refactor(endpoints): reconcile every endpoint event through one path

Now that a deletion recomputes the remaining endpoints, AddOrModify and
Delete differed only in how the event mutated the provider. Keeping two
near-identical bodies is what let their behaviour drift apart in the first
place, so fold them into a single Reconcile that dispatches the mutation
in applyEvent.

Deletions consequently pick up the parts of the add path they were still
missing: the IPv6 egress guard, the leader election start, and the
endpointless opt-in branch, which now keeps an opted-in Cluster service up
after its last EndpointSlice disappears instead of tearing it down.

Assisted-by: GitHub-Copilot:unspecified
Signed-off-by: Marcel Fest <marcel.fest@telekom.de>

.github/workflows/anchore-syft.yml

@@ -22,10 +22,10 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v4
+        uses: actions/checkout@v7
         with:
           ref: ${{ github.ref_name }}
       - name: Anchore SBOM Action
-        uses: anchore/sbom-action@v0.17.5
+        uses: anchore/sbom-action@v0.24.0
         with:
           format: cyclonedx-json

.github/workflows/ci-pull-request.yaml

@@ -0,0 +1,88 @@
+name: For each PR
+on:
+  pull_request:
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  e2e-tests:
+    runs-on: ubuntu-latest
+    name: E2E tests
+    timeout-minutes: 120
+    env:
+      GINKGO_PROCS: ${{ matrix.ginkgo-procs }}
+    strategy:
+      fail-fast: false
+      max-parallel: 3
+      matrix:
+        include:
+          - mode: arp
+            ginkgo-procs: 4
+          - mode: rt
+            ginkgo-procs: 4
+          - mode: bgp
+            ginkgo-procs: 4
+    steps:
+      - name: Get current date
+        id: date
+        run: echo "date=$(date +'%Y-%m-%d-%H-%M')" >> "$GITHUB_OUTPUT"
+      - name: Ensure fs wont cause issues
+        run: sudo sysctl fs.inotify.max_user_instances=8192 &&  sudo sysctl fs.inotify.max_user_watches=524288
+      - name: Checkout code
+        uses: actions/checkout@v7
+      - name: Install Go
+        uses: actions/setup-go@v7
+        with:
+          go-version-file: go.mod
+      - name: Build image locally
+        run: make dockerx86Local
+      - name: Run Manifest generation tests
+        run: make manifest-test
+        if: matrix.mode == 'arp'
+      - name: Run ARP mode tests
+        run: DOCKER_API_VERSION=1.48 E2E_KEEP_LOGS=true GINKGO_ARGS="--output-dir=/tmp --json-report=kube-vip-test-report-arp.json" make e2e-tests-arp
+        if: matrix.mode == 'arp'
+      - name: Run RT mode tests
+        run: DOCKER_API_VERSION=1.48 E2E_KEEP_LOGS=true GINKGO_ARGS="--output-dir=/tmp --json-report=kube-vip-test-report-rt.json" make e2e-tests-rt
+        if: matrix.mode == 'rt'
+      - name: Get GoBGP binaries
+        run: make get-gobgp
+        if: matrix.mode == 'bgp'
+      - name: Run BGP mode tests
+        run: sudo -E PATH=$PATH DOCKER_API_VERSION=1.48 E2E_KEEP_LOGS=true GINKGO_ARGS="--output-dir=/tmp --json-report=kube-vip-test-report-bgp.json" make e2e-tests-bgp
+        if: matrix.mode == 'bgp'
+      - name: Change log directory permissions
+        run: sudo chmod -R 755 /tmp/kube-vip-test*
+        if: matrix.mode == 'bgp' && always()
+      - name: Save logs
+        uses: actions/upload-artifact@v7
+        with:
+          name: e2e-test-logs-${{ matrix.mode }}-${{ steps.date.outputs.date }}
+          path: /tmp/kube-vip-test*
+        if: always()
+  service-e2e-tests:
+    runs-on: ubuntu-latest
+    name: E2E service tests
+    timeout-minutes: 60
+    steps:
+      - name: Get current date
+        id: date
+        run: echo "date=$(date +'%Y-%m-%d-%H-%M')" >> "$GITHUB_OUTPUT"
+      - name: Checkout code
+        uses: actions/checkout@v7
+      - name: Install Go
+        uses: actions/setup-go@v7
+        with:
+          go-version-file: go.mod
+      - name: Build image with iptables
+        run: make dockerx86ActionIPTables
+      - name: Run tests
+        run: DOCKER_API_VERSION=1.48 DOCKERTAG=action E2E_KEEP_LOGS=true make service-tests
+      - name: Save logs
+        uses: actions/upload-artifact@v7
+        with:
+          name: services-test-logs-${{ steps.date.outputs.date }}
+          path: /tmp/kube-vip-services*
+        if: always()

.github/workflows/ci.yaml

@@ -1,34 +1,52 @@
-name: For each commit and PR
+name: For each commit
 on:
   push:
+    branches: [main]
+    tags: ['v*']
   pull_request:
-env:
-  GO_VERSION: "1.21"
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: ${{ github.event_name == 'pull_request' }}
+
 jobs:
   validation:
     runs-on: ubuntu-latest
     name: Checks and linters
+    timeout-minutes: 30
     steps:
       - name: Init
-        run: sudo apt-get update && sudo apt-get install -y build-essential golint &&  sudo sysctl fs.inotify.max_user_instances=8192 &&  sudo sysctl fs.inotify.max_user_watches=524288
+        run: sudo apt-get update && sudo apt-get install -y build-essential &&  sudo sysctl fs.inotify.max_user_instances=8192 &&  sudo sysctl fs.inotify.max_user_watches=524288
       - name: Install golangci-lint
-        run: curl -sSfL https://raw.githubusercontent.com/golangci/golangci-lint/master/install.sh | sh -s -- -b $(go env GOPATH)/bin v1.55.2
+        uses: golangci/golangci-lint-action@v9
+        with:
+          version: v2.12
+          install-only: true
       - name: Checkout code
-        uses: actions/checkout@v4
+        uses: actions/checkout@v7
       - name: Install Go
-        uses: actions/setup-go@v5
+        uses: actions/setup-go@v7
         with:
           go-version-file: go.mod
+      - name: Verify gofmt
+        run: |
+          unformatted=$(gofmt -l .)
+          if [ -n "$unformatted" ]; then
+            echo "The following files are not gofmt-formatted:"
+            echo "$unformatted"
+            exit 1
+          fi
       - name: All checks
         run: make check
   unit-tests:
     runs-on: ubuntu-latest
     name: Unit tests
+    timeout-minutes: 30
     steps:
       - name: Checkout code
-        uses: actions/checkout@v4
+        uses: actions/checkout@v7
       - name: Install Go
-        uses: actions/setup-go@v5
+        uses: actions/setup-go@v7
         with:
           go-version-file: go.mod
       - name: Run tests
@@ -36,64 +54,31 @@ jobs:
   integration-tests:
     name: Integration tests
     runs-on: ubuntu-latest
-    steps:
-    - name: Checkout code
-      uses: actions/checkout@v4
-    - name: Install Go
-      uses: actions/setup-go@v5
-      with:
-        go-version-file: go.mod
-    - name: Run tests
-      run: make integration-tests
-  e2e-tests:
-    runs-on: ubuntu-latest
-    name: E2E ARP tests
-    steps:
-      - name: Ensure fs wont cause issues
-        run: sudo sysctl fs.inotify.max_user_instances=8192 &&  sudo sysctl fs.inotify.max_user_watches=524288
-      - name: Checkout code
-        uses: actions/checkout@v4
-      - name: Install Go
-        uses: actions/setup-go@v5
-        with:
-          go-version-file: go.mod
-      - name: Build image locally
-        run: make dockerx86Local
-      - name: Run Manifest generation tests
-        run: make manifest-test
-      - name: Run Control plane tests
-        run: make e2e-tests
-      - name: Run Control plane tests v1.29.0 onwards
-        run: make e2e-tests129
-  service-e2e-tests:
-    runs-on: ubuntu-latest
-    name: E2E service tests
+    timeout-minutes: 20
     steps:
       - name: Checkout code
-        uses: actions/checkout@v4
+        uses: actions/checkout@v7
       - name: Install Go
-        uses: actions/setup-go@v5
+        uses: actions/setup-go@v7
         with:
           go-version-file: go.mod
-      - name: Build image with iptables
-        run: make dockerx86ActionIPTables
       - name: Run tests
-        run: DOCKERTAG=action make service-tests
+        run: make integration-tests
   image-vul-check:
     runs-on: ubuntu-latest
     name: Image vulnerability scan
+    timeout-minutes: 30
     steps:
       - name: Checkout code
-        uses: actions/checkout@v4
-      - name: Build image with iptables
-        run: make dockerx86ActionIPTables
+        uses: actions/checkout@v7
+      - name: Build image
+        run: make dockerx86Action
       - name: Run Trivy vulnerability scanner
         uses: aquasecurity/trivy-action@master
         with:
-          image-ref: 'plndr/kube-vip:action'
-          format: 'table'
-          exit-code: '1'
+          image-ref: "plndr/kube-vip:action"
+          format: "table"
+          exit-code: "1"
           ignore-unfixed: true
-          vuln-type: 'os,library'
-          severity: 'CRITICAL,HIGH'
-      
+          vuln-type: "os,library"
+          severity: "CRITICAL,HIGH"

.github/workflows/codeql-analysis.yml

@@ -38,16 +38,16 @@ jobs:
 
     steps:
     - name: Checkout repository
-      uses: actions/checkout@v4
+      uses: actions/checkout@v7
       
     - name: Install Go
-      uses: actions/setup-go@v5
+      uses: actions/setup-go@v7
       with:
         go-version-file: go.mod
         
     # Initializes the CodeQL tools for scanning.
     - name: Initialize CodeQL
-      uses: github/codeql-action/init@v3
+      uses: github/codeql-action/init@v4
       with:
         languages: ${{ matrix.language }}
         # If you wish to specify custom queries, you can do so here or in a config file.
@@ -58,7 +58,7 @@ jobs:
     # Autobuild attempts to build any compiled languages  (C/C++, C#, or Java).
     # If this step fails, then you should remove it and run the build manually (see below)
     - name: Autobuild
-      uses: github/codeql-action/autobuild@v3
+      uses: github/codeql-action/autobuild@v4
 
     # ℹ️ Command-line programs to run using the OS shell.
     # 📚 https://git.io/JvXDl
@@ -72,4 +72,4 @@ jobs:
     #   make release
 
     - name: Perform CodeQL Analysis
-      uses: github/codeql-action/analyze@v3
+      uses: github/codeql-action/analyze@v4

.github/workflows/main.yaml

@@ -11,25 +11,25 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v4
+        uses: actions/checkout@v7
       - name: Set up QEMU
-        uses: docker/setup-qemu-action@v3
+        uses: docker/setup-qemu-action@v4
       - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
+        uses: docker/setup-buildx-action@v4
       - name: Login to DockerHub
-        uses: docker/login-action@v3
+        uses: docker/login-action@v4
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
       - name: Login to Github Packages
-        uses: docker/login-action@v3
+        uses: docker/login-action@v4
         with:
           registry: ghcr.io
           username: ${{ github.actor }}
           password: ${{ secrets.GITHUB_TOKEN }}   
       - name: Build standard version
         id: docker_build
-        uses: docker/build-push-action@v6
+        uses: docker/build-push-action@v7
         with:
           context: .
           platforms: linux/amd64,linux/arm/v7,linux/arm64,linux/ppc64le,linux/s390x
@@ -39,7 +39,7 @@ jobs:
             ghcr.io/kube-vip/kube-vip:${{ github.ref_name }}
       - name: Build iptables version
         id: docker_build_iptables
-        uses: docker/build-push-action@v6
+        uses: docker/build-push-action@v7
         with:
           context: .
           file: Dockerfile_iptables

.github/workflows/release.yaml

@@ -11,29 +11,36 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v4
+        uses: actions/checkout@v7
       - name: Set up QEMU
-        uses: docker/setup-qemu-action@v3
+        uses: docker/setup-qemu-action@v4
       - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
+        uses: docker/setup-buildx-action@v4
       - name: Login to DockerHub
-        uses: docker/login-action@v3
+        uses: docker/login-action@v4
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
       - name: Login to Github Packages
-        uses: docker/login-action@v3
+        uses: docker/login-action@v4
         with:
           registry: ghcr.io
           username: ${{ github.actor }}
           password: ${{ secrets.GITHUB_TOKEN }}
+      - name: Generate Metadata
+        uses: docker/metadata-action@v6.2.0
+        id: metadata
+        with:
+          labels: |
+            org.opencontainers.image.documentation=https://kube-vip.io/docs/
       - name: Build and push main branch
         id: docker_build
-        uses: docker/build-push-action@v6
+        uses: docker/build-push-action@v7
         with:
           context: .
           platforms: linux/amd64,linux/arm/v7,linux/arm64,linux/ppc64le,linux/s390x
           push: ${{ github.event_name != 'pull_request' }}
+          labels: ${{ steps.metadata.outputs.labels }}
           tags: >-
             plndr/kube-vip:${{ github.ref_name }},
             plndr/kube-vip:latest,
@@ -41,12 +48,13 @@ jobs:
             ghcr.io/kube-vip/kube-vip:latest
       - name: Build iptables version and push main branch
         id: docker_build_iptables
-        uses: docker/build-push-action@v6
+        uses: docker/build-push-action@v7
         with:
           context: .
           file: Dockerfile_iptables
           platforms: linux/amd64,linux/arm/v7,linux/arm64,linux/ppc64le,linux/s390x
           push: ${{ github.event_name != 'pull_request' }}
+          labels: ${{ steps.metadata.outputs.labels }}
           tags: >-
             plndr/kube-vip-iptables:${{ github.ref_name }},
             plndr/kube-vip-iptables:latest,

.gitignore

@@ -5,3 +5,4 @@ bin
 testing/e2e/etcd/certs
 pkg/etcd/etcd.pid
 pkg/etcd/etcd-data
+testing/e2e/e2e.test

.golangci.yml

@@ -1,13 +1,38 @@
-run:
-  timeout: 10m
-  
+version: "2"
 linters:
   enable:
-  - bodyclose
-  - gofmt
-  - goimports
-  - revive
-  - gosec
-  - misspell
-  - unconvert
-  - unparam
+    - bodyclose
+    - gosec
+    - misspell
+    - unconvert
+    - unparam
+  settings:
+    misspell:
+      ignore-rules:
+        - creater
+    staticcheck:
+      checks:
+        - all
+        # Disable QF1008 to retain embedded fields for better readability.
+        - "-QF1008"
+  exclusions:
+    generated: lax
+    presets:
+      - comments
+      - common-false-positives
+      - legacy
+      - std-error-handling
+    paths:
+      - third_party$
+      - builtin$
+      - examples$
+formatters:
+  enable:
+    - gofmt
+    - goimports
+  exclusions:
+    generated: lax
+    paths:
+      - third_party$
+      - builtin$
+      - examples$

CHANGELOG.md

@@ -0,0 +1,61 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [Unreleased]
+
+### Fixed
+- Add a configurable kube-vip instance name and use it to isolate internal nftables egress tables, persist table ownership on Services, and migrate per-Service chains without affecting other deployments. Fixes #1634.
+- Retry on 403 Forbidden and 401 Unauthorized in `ServicesWatcher` at startup with exponential backoff. Fixes #1464.
+- Reintroduce BGP config via node annotations. Fixes #1488.
+- Fail fast in runtime `manager` and `service` paths when legacy `vip_address` is used without `vip_subnet` in control-plane ARP, BGP, or Routing Table mode.
+- Cancel the mode context on init or configuration failure before waiting on goroutines during shutdown.
+
+
+### Added
+- Configurable control-plane health check for BGP mode without leader election
+  - Polls a configurable HTTP(S) endpoint (e.g. `https://localhost:6443/livez`) to verify the exposed service is healthy (usually the local kube-apiserver)
+  - Withdraws the BGP route after a configurable number of consecutive failures, removing the unhealthy node from the ECMP set
+  - Re-announces the route automatically once the endpoint recovers
+  - Gracefully withdraws the route on shutdown (SIGTERM)
+  - Supports custom CA certificates for TLS verification
+  - Configuration via environment variables or CLI flags:
+    - `control_plane_health_check_address` / `--controlPlaneHealthCheckAddress`: URL to poll
+    - `control_plane_health_check_period_seconds` / `--controlPlaneHealthCheckPeriodSeconds`: interval between checks (default: 5)
+    - `control_plane_health_check_timeout_seconds` / `--controlPlaneHealthCheckTimeoutSeconds`: per-request timeout (default: 3)
+    - `control_plane_health_check_failure_threshold` / `--controlPlaneHealthCheckFailureThreshold`: consecutive failures before withdrawal (default: 3)
+    - `control_plane_health_check_ca_path` / `--controlPlaneHealthCheckCAPath`: CA cert for HTTPS verification
+- SIGUSR1 signal handler for runtime configuration dumps (#1301)
+  - Send SIGUSR1 to kube-vip process to dump current configuration to stdout
+  - Configuration dump includes:
+    - Basic configuration (VIP, interface, port, namespace settings)
+    - BGP configuration (enabled status, AS number, router ID, peers)
+    - ARP/NDP configuration (enabled status, broadcast rate)
+    - Services configuration (enabled status, load balancer settings)
+    - Network interfaces status
+    - Leader election configuration (type, lease details)
+    - Runtime statistics (load balancer, Prometheus, health check settings)
+  - Output format: Human-readable plaintext via fmt.Printf()
+  - Thread-safe implementation using mutex protection
+  - Non-disruptive: Process continues running after configuration dump
+  - Added comprehensive unit tests for all dump methods
+  - Added E2E tests for signal handling
+- Opt-in support for endpointless `LoadBalancer` services with `externalTrafficPolicy: Cluster`
+  - Annotation: `kube-vip.io/allow-reconcile-without-endpoints: "true"`
+  - Starts service handling path for opted-in endpointless Cluster services while preserving default endpoint-gated behavior for non-opt-in services and `Local` policy
+  - Added endpoint behavior tests and README usage documentation
+  - Added support in ipoib interfaces in ARP mode. Fixes #694
+
+### Changed
+- BGP mode now honours `enable_leader_election` for services: a single global services leader advertises the service VIPs instead of every node advertising them. Deployments that enabled `enable_leader_election` for the control plane and relied on ECMP/multipath for services must unset it (or switch to `enable_service_election`) to keep the previous datapath. kube-vip logs a warning on startup when this path is taken.
+- Updated signal handlers in manager_arp.go, manager_bgp.go, manager_wireguard.go, and manager_table.go to use switch statement pattern for handling multiple signals (SIGUSR1, SIGINT, SIGTERM)
+- wireguard.go now manages a complete wireguard interface on the current network namespace
+- manager_wireguard.go uses the new wireguard.go implementation
+
+## [v1.0.1] - Previous Release
+
+### Previous changes
+- See git history for changes prior to CHANGELOG.md introduction

CONTRIBUTING.md

@@ -113,8 +113,8 @@ and *merged* sorts of commits.
 To make it easier for reviewers to review your PR, consider the following:
 
 1. Follow the golang [coding conventions](https://github.com/golang/go/wiki/CodeReviewComments).
-2. Format your code with `make golangci-fix`; if the [linters](ci/README.md) flag an issue that
-   cannot be fixed automatically, an error message will be displayed so you can address the issue.
+2. Format your code with `make simplify` to automatically fix formatting issues.
+2. Lint your code with `make check`; if the linters flag an issue that cannot be fixed automatically, an error message will be displayed so you can address the issue.
 3. Follow [git commit](https://chris.beams.io/posts/git-commit/) guidelines.
 4. Follow [logging](https://github.com/kubernetes/community/blob/master/contributors/devel/sig-instrumentation/logging.md) guidelines.
 

Dockerfile

@@ -1,6 +1,6 @@
 # syntax=docker/dockerfile:experimental
 
-FROM golang:1.23.2-alpine3.20 as dev
+FROM golang:1.27.0-alpine3.23 as dev
 RUN apk add --no-cache git ca-certificates make
 RUN adduser -D appuser
 COPY . /src/

Dockerfile_iptables

@@ -1,6 +1,6 @@
 # syntax=docker/dockerfile:experimental
 
-FROM golang:1.23.2-alpine3.20 as dev
+FROM golang:1.27.0-alpine3.23 as dev
 RUN apk add --no-cache git make
 RUN adduser -D appuser
 COPY . /src/
@@ -11,7 +11,7 @@ RUN --mount=type=cache,sharing=locked,id=gomod,target=/go/pkg/mod/cache \
     --mount=type=cache,sharing=locked,id=goroot,target=/root/.cache/go-build \
     CGO_ENABLED=0 GOOS=linux make build
 
-FROM alpine:3.20.3
+FROM alpine:3.24.1
 # Update pkgs and add iptables
 RUN apk upgrade && \
     apk add --no-cache iptables iptables-legacy

Makefile

@@ -5,7 +5,7 @@ TARGET := kube-vip
 .DEFAULT_GOAL := $(TARGET)
 
 # These will be provided to the target
-VERSION := v0.8.5
+VERSION := v1.2.3
 
 BUILD := `git rev-parse HEAD`
 
@@ -15,9 +15,15 @@ TARGETOS=linux
 # Use linker flags to provide version/build settings to the target
 LDFLAGS=-ldflags "-s -w -X=main.Version=$(VERSION) -X=main.Build=$(BUILD) -extldflags -static"
 DOCKERTAG ?= $(VERSION)
-REPOSITORY ?= plndr
+REPOSITORY ?= docker.io/plndr
+GO_VERSION := 1.25.6
+K8S_VERSION ?= v1.35.0
+GINKGO_ARGS ?=
+GINKGO_PROCS ?=
+GINKGO_PARALLEL := $(if $(GINKGO_PROCS),--procs=$(GINKGO_PROCS),-p)
+BUILDX_CACHE_FLAGS ?=
 
-.PHONY: all build clean install uninstall simplify check run e2e-tests
+.PHONY: all build clean install uninstall simplify check run e2e-tests unit-tests integration-tests unit-tests-docker integration-tests-docker
 
 all: check install
 
@@ -75,17 +81,17 @@ docker:
 # This will build a local docker image (x86 only), use make dockerLocal for all architectures
 dockerx86Local:
 	@-rm ./kube-vip
-	@docker buildx build  --platform linux/amd64 --load -t $(REPOSITORY)/$(TARGET):$(DOCKERTAG) .
+	@docker buildx build --platform linux/amd64 --load -t $(REPOSITORY)/$(TARGET):$(DOCKERTAG) $(BUILDX_CACHE_FLAGS) .
 	@echo New Multi Architecture Docker image created
 
 dockerx86Action:
 	@-rm ./kube-vip
-	@docker buildx build  --platform linux/amd64 --load -t $(REPOSITORY)/$(TARGET):action .
+	@docker buildx build --platform linux/amd64 --load -t $(REPOSITORY)/$(TARGET):action $(BUILDX_CACHE_FLAGS) .
 	@echo New Multi Architecture Docker image created
 
 dockerx86ActionIPTables:
 	@-rm ./kube-vip
-	@docker buildx build  --platform linux/amd64 -f ./Dockerfile_iptables --load -t $(REPOSITORY)/$(TARGET):action .
+	@docker buildx build --platform linux/amd64 -f ./Dockerfile_iptables --load -t $(REPOSITORY)/$(TARGET):action $(BUILDX_CACHE_FLAGS) .
 	@echo New Multi Architecture Docker image created
 
 dockerLocal:
@@ -109,37 +115,46 @@ run: install
 manifests:
 	@make build
 	@mkdir -p ./docs/manifests/$(VERSION)/
-	@./kube-vip manifest pod --interface eth0 --vip 192.168.0.1 --arp --leaderElection --controlplane --services > ./docs/manifests/$(VERSION)/kube-vip-arp.yaml
-	@./kube-vip manifest pod --interface eth0 --vip 192.168.0.1 --arp --leaderElection --controlplane --services --enableLoadBalancer > ./docs/manifests/$(VERSION)/kube-vip-arp-lb.yaml
-	@./kube-vip manifest pod --interface eth0 --vip 192.168.0.1 --bgp --controlplane --services > ./docs/manifests/$(VERSION)/kube-vip-bgp.yaml
-	@./kube-vip manifest daemonset --interface eth0 --vip 192.168.0.1 --arp --leaderElection --controlplane --services --inCluster > ./docs/manifests/$(VERSION)/kube-vip-arp-ds.yaml
-	@./kube-vip manifest daemonset --interface eth0 --vip 192.168.0.1 --arp --leaderElection --controlplane --services --inCluster --enableLoadBalancer > ./docs/manifests/$(VERSION)/kube-vip-arp-ds-lb.yaml
-	@./kube-vip manifest daemonset --interface eth0 --vip 192.168.0.1 --bgp --leaderElection --controlplane --services --inCluster > ./docs/manifests/$(VERSION)/kube-vip-bgp-ds.yaml
-	@./kube-vip manifest daemonset --interface eth0 --vip 192.168.0.1 --bgp --leaderElection --controlplane --services --inCluster --provider-config /etc/cloud-sa/cloud-sa.json > ./docs/manifests/$(VERSION)/kube-vip-bgp-em-ds.yaml
+	@./kube-vip manifest pod --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --arp --leaderElection --controlplane --services > ./docs/manifests/$(VERSION)/kube-vip-arp.yaml
+	@./kube-vip manifest pod --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --arp --leaderElection --controlplane --services --enableLoadBalancer > ./docs/manifests/$(VERSION)/kube-vip-arp-lb.yaml
+	@./kube-vip manifest pod --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --bgp --controlplane --services > ./docs/manifests/$(VERSION)/kube-vip-bgp.yaml
+	@./kube-vip manifest daemonset --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --arp --leaderElection --controlplane --services --inCluster > ./docs/manifests/$(VERSION)/kube-vip-arp-ds.yaml
+	@./kube-vip manifest daemonset --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --arp --leaderElection --controlplane --services --inCluster --enableLoadBalancer > ./docs/manifests/$(VERSION)/kube-vip-arp-ds-lb.yaml
+	@./kube-vip manifest daemonset --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --bgp --leaderElection --controlplane --services --inCluster > ./docs/manifests/$(VERSION)/kube-vip-bgp-ds.yaml
+	@./kube-vip manifest daemonset --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --bgp --leaderElection --controlplane --services --inCluster > ./docs/manifests/$(VERSION)/kube-vip-bgp-em-ds.yaml
 	@-rm ./kube-vip
 
 manifest-test: 
-	docker run $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest pod --interface eth0 --vip 192.168.0.1 --arp --leaderElection --controlplane --services
-	docker run $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest pod --interface eth0 --vip 192.168.0.1 --arp --leaderElection --controlplane --services --enableLoadBalancer
-	docker run $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest pod --interface eth0 --vip 192.168.0.1 --bgp --controlplane --services 
-	docker run $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest daemonset --interface eth0 --vip 192.168.0.1 --arp --leaderElection --controlplane --services --inCluster
-	docker run $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest daemonset --interface eth0 --vip 192.168.0.1 --arp --leaderElection --controlplane --services --inCluster --enableLoadBalancer
-	docker run $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest daemonset --interface eth0 --vip 192.168.0.1 --bgp --leaderElection --controlplane --services --inCluster
+	docker run $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest pod --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --arp --leaderElection --controlplane --services
+	docker run $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest pod --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --arp --leaderElection --controlplane --services --enableLoadBalancer
+	docker run $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest pod --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --bgp --controlplane --services
+	docker run $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest daemonset --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --arp --leaderElection --controlplane --services --inCluster
+	docker run $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest daemonset --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --arp --leaderElection --controlplane --services --inCluster --enableLoadBalancer
+	docker run $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest daemonset --interface eth0 --vip 192.168.0.1 --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --bgp --leaderElection --controlplane --services --inCluster
 
 unit-tests:
-	go test ./...
+	go test -race ./...
+
+unit-tests-docker:
+	docker run --rm -w /kube-vip -v $$(pwd):/kube-vip -v kube-vip-gomod-cache:/go/pkg/mod -v kube-vip-gobuild-cache:/root/.cache/go-build golang:$(GO_VERSION) make unit-tests
 
 integration-tests:
 	go test -tags=integration,e2e -v ./pkg/etcd
 
-e2e-tests:
-	E2E_IMAGE_PATH=$(REPOSITORY)/$(TARGET):$(DOCKERTAG) go run github.com/onsi/ginkgo/v2/ginkgo --tags=e2e -v -p ./testing/e2e ./testing/e2e/etcd
+e2e-tests-arp: get-whoami
+	GOMAXPROCS=4 TEST_MODE=arp K8S_IMAGE_PATH=kindest/node:$(K8S_VERSION) E2E_IMAGE_PATH=$(REPOSITORY)/$(TARGET):$(DOCKERTAG) go run github.com/onsi/ginkgo/v2/ginkgo --tags=e2e -v $(GINKGO_PARALLEL) $(GINKGO_ARGS) ./testing/e2e
 
-e2e-tests129:
-	V129=true K8S_IMAGE_PATH=kindest/node:v1.29.0 E2E_IMAGE_PATH=$(REPOSITORY)/$(TARGET):$(DOCKERTAG) go run github.com/onsi/ginkgo/v2/ginkgo --tags=e2e -v -p ./testing/e2e
+e2e-tests-rt: get-whoami
+	GOMAXPROCS=4 TEST_MODE=rt K8S_IMAGE_PATH=kindest/node:$(K8S_VERSION) E2E_IMAGE_PATH=$(REPOSITORY)/$(TARGET):$(DOCKERTAG) go run github.com/onsi/ginkgo/v2/ginkgo --tags=e2e -v $(GINKGO_PARALLEL) $(GINKGO_ARGS) ./testing/e2e
+
+e2e-tests-bgp: get-whoami get-gobgp
+	GOMAXPROCS=4 TEST_MODE=bgp K8S_IMAGE_PATH=kindest/node:$(K8S_VERSION) E2E_IMAGE_PATH=$(REPOSITORY)/$(TARGET):$(DOCKERTAG) go run github.com/onsi/ginkgo/v2/ginkgo --tags=e2e -v $(GINKGO_PARALLEL) $(GINKGO_ARGS) ./testing/e2e
+
+e2e-tests: e2e-tests-arp e2e-tests-rt e2e-tests-bgp
 
 service-tests:
-	E2E_IMAGE_PATH=$(REPOSITORY)/$(TARGET):$(DOCKERTAG) go run ./testing/e2e/services -Services
+	$(MAKE) -C testing/e2e/e2e dockerLocal
+	E2E_IMAGE_PATH=$(REPOSITORY)/$(TARGET):$(DOCKERTAG) go run ./testing/services -Services -simple -deployments -leaderActive -leaderFailover -localDeploy -electionFaults -egress -egressIPv6 -dualStack -egressInternal
 
 trivy: dockerx86ActionIPTables
 	docker run -v /var/run/docker.sock:/var/run/docker.sock aquasec/trivy:0.47.0 \
@@ -151,3 +166,23 @@ trivy: dockerx86ActionIPTables
 		--severity  'CRITICAL,HIGH'  \
 		$(REPOSITORY)/$(TARGET):action
 
+kind-quick:
+	echo "Standing up your cluster"
+	kind create cluster --config ./testing/kind/kind.yaml --name kube-vip
+	kubectl apply -f https://kube-vip.io/manifests/rbac.yaml
+	kubectl create configmap --namespace kube-system kubevip --from-literal range-global=172.18.100.10-172.18.100.30
+	kubectl apply -f https://raw.githubusercontent.com/kube-vip/kube-vip-cloud-provider/main/manifest/kube-vip-cloud-controller.yaml
+	kind load docker-image --name kube-vip $(REPOSITORY)/$(TARGET):$(DOCKERTAG)
+	docker run --network host --rm $(REPOSITORY)/$(TARGET):$(DOCKERTAG) manifest daemonset --services --inCluster --image "$(REPOSITORY)/$(TARGET):$(DOCKERTAG)" --arp --servicesElection --interface  eth0 | kubectl apply -f -
+
+kind-reload:
+	kind load docker-image $(REPOSITORY)/$(TARGET):$(DOCKERTAG) --name services
+	kubectl rollout restart -n kube-system daemonset/kube-vip-ds
+
+get-gobgp:
+	mkdir -p bin
+	wget -nc --directory-prefix=bin https://github.com/osrg/gobgp/releases/download/v4.6.0/gobgp_4.6.0_linux_amd64.tar.gz
+	tar -xvzf bin/gobgp_4.6.0_linux_amd64.tar.gz -C bin
+
+get-whoami:
+	docker pull ghcr.io/traefik/whoami:v1.11

README.md

@@ -4,7 +4,7 @@ High Availability and Load-Balancing
 
 ![](https://github.com/kube-vip/kube-vip/raw/main/kube-vip.png)
 
-[![Build and publish main image regularly](https://github.com/kube-vip/kube-vip/actions/workflows/main.yaml/badge.svg)](https://github.com/kube-vip/kube-vip/actions/workflows/main.yaml)
+[![Build and publish main image regularly](https://github.com/kube-vip/kube-vip/actions/workflows/main.yaml/badge.svg)](https://github.com/kube-vip/kube-vip/actions/workflows/main.yaml) [![LFX Health Score](https://img.shields.io/static/v1?label=Health%20Score&message=Healthy&color=A7F3D0&logo=linuxfoundation&logoColor=white&style=flat)](https://insights.linuxfoundation.org/project/kube-vip) [![LFX Active Contributors](https://img.shields.io/static/v1?label=Active%20contributors%20(1Y)&message=212&color=0094FF&logo=linuxfoundation&logoColor=white&style=flat)](https://insights.linuxfoundation.org/project/kube-vip)
 
 ## Overview
 Kubernetes Virtual IP and Load-Balancer for both control plane and Kubernetes services
@@ -18,6 +18,8 @@ The idea behind `kube-vip` is a small self-contained Highly-Available option for
 
 **NOTE:** All documentation of both usage and architecture are now available at [https://kube-vip.io](https://kube-vip.io).
 
+For upgrading an existing install in place (static Pod or DaemonSet), see the [upgrade guide](https://kube-vip.io/docs/upgrade/).
+
 ## Features
 
 Kube-Vip was originally created to provide a HA solution for the Kubernetes control plane, over time it has evolved to incorporate that same functionality into Kubernetes service type [load-balancers](https://kubernetes.io/docs/concepts/services-networking/service/#loadbalancer).
@@ -32,6 +34,7 @@ Kube-Vip was originally created to provide a HA solution for the Kubernetes cont
 - Service LoadBalancer address pools per namespace or global
 - Service LoadBalancer address via (existing network DHCP)
 - Service LoadBalancer address exposure to gateway via UPNP
+- Egress! Kube-vip will utilise a service loadbalancer as both the ingress and **egress** for a pod. 
 - ... manifest generation, vendor API integrations and many more...
 
 ## Why?
@@ -57,14 +60,81 @@ All of these would require a separate level of configuration and in some infrast
 
 ## Troubleshooting and Feedback
 
+### SELinux and IPVS kernel modules
+
+When using IPVS load balancing on nodes with SELinux enforcing, kube-vip may be
+blocked from requesting kernel modules from inside the container. Symptoms can
+include the kube-vip pod entering `Error` or `CrashLoopBackOff`, logs that show
+`ensure IPVS kernel modules are loaded`, or audit denials for `module_request`
+from `container_t`.
+
+Load the required IPVS modules on every node that can run kube-vip before
+deploying it:
+
+```shell
+sudo modprobe ip_vs
+sudo modprobe ip_vs_rr
+```
+
+To persist this across reboots, add the modules to a file such as
+`/etc/modules-load.d/kube-vip-ipvs.conf`:
+
+```text
+ip_vs
+ip_vs_rr
+```
+
+Preloading only the required modules is preferred to enabling the SELinux
+`domain_kernel_load_modules` boolean for containers.
+
+### Gateway API `LoadBalancer` services with no endpoints
+
+Some Gateway API controllers create `LoadBalancer` services that intentionally have no Endpoints/EndpointSlices backends.
+
+If you want kube-vip to reconcile such a service, opt in with:
+
+```yaml
+metadata:
+  annotations:
+    kube-vip.io/allow-reconcile-without-endpoints: "true"
+spec:
+  type: LoadBalancer
+  externalTrafficPolicy: Cluster
+```
+
+Scope:
+- Works only with `externalTrafficPolicy: Cluster`
+- No effect for `Local`
+- Default endpoint-gated behavior remains unchanged for services without this annotation
+
 Please raise issues on the GitHub repository and as mentioned check the documentation at [https://kube-vip.io](https://kube-vip.io/).
 
+## Community Tools
+
+- **[KubeStellar Console — Guided kube-vip Install](https://console.kubestellar.io/missions/install-kube-vip)** — A step-by-step guided installation experience for kube-vip with pre-flight checks, validation, troubleshooting, and rollback support.
+
 ## Contributing
 
 Thanks for taking the time to join our community and start contributing! We welcome pull requests. Feel free to dig through the [issues](https://github.com/kube-vip/kube-vip/issues) and jump in.
 
 :warning: This project has issue compiling on MacOS, please compile it on linux distribution
 
+Additionally it is now relatively easy and quick to develop with [skaffold](https://skaffold.dev/), and the `skaffold.yaml` exists within the root folder of the gir repository.
+
+### Set up a kind development environment
+
+1. `kind create cluster --config ./testing/kind.yaml`
+2. `kubectl apply -f https://kube-vip.io/manifests/rbac.yaml`
+3. Create a load balancer range `configMap` from the kind cluster
+4. Apply the CCM manifest
+5. Start skaffold with `skaffold dev`
+6. Start developing and see changes applied in real-time.
+
 ## Star History
 
 [![Star History Chart](https://api.star-history.com/svg?repos=kube-vip/kube-vip&type=Date)](https://star-history.com/#kube-vip/kube-vip&Date)
+[![FOSSA Status](https://app.fossa.com/api/projects/git%2Bgithub.com%2Fkube-vip%2Fkube-vip.svg?type=shield)](https://app.fossa.com/projects/git%2Bgithub.com%2Fkube-vip%2Fkube-vip?ref=badge_shield)
+
+
+## License
+[![FOSSA Status](https://app.fossa.com/api/projects/git%2Bgithub.com%2Fkube-vip%2Fkube-vip.svg?type=large)](https://app.fossa.com/projects/git%2Bgithub.com%2Fkube-vip%2Fkube-vip?ref=badge_large)

cmd/kube-vip-kubeadm.go

@@ -4,8 +4,9 @@ import (
 	"fmt"
 	"os"
 
+	log "log/slog"
+
 	"github.com/kube-vip/kube-vip/pkg/kubevip"
-	log "github.com/sirupsen/logrus"
 	"github.com/spf13/cobra"
 )
 
@@ -22,46 +23,73 @@ func init() {
 var kubeKubeadm = &cobra.Command{
 	Use:   "kubeadm",
 	Short: "Kubeadm functions",
-	Run: func(cmd *cobra.Command, args []string) {
+	Long: `This command group provides utilities for generating static Pod manifests specifically tailored for the kubeadm bootstrapping process.
+	It contains two subcommands:
+  		- init:  Generates a manifest to be used during 'kubeadm init' on the first control-plane node.
+  		- join:  Generates a manifest to be used during 'kubeadm join' for additional control-plane nodes.
+
+The generated YAML manifest should be saved to the kubeadm static Pod directory (typically /etc/kubernetes/manifests/) so that kubeadm launches the kube-vip static Pod automatically.`,
+	Run: func(cmd *cobra.Command, _ []string) {
 		_ = cmd.Help()
-		// TODO - A load of text detailing what's actually happening
 	},
 }
 
 var kubeKubeadmInit = &cobra.Command{
 	Use:   "init",
 	Short: "kube-vip init",
-	Long:  "The \"init\" subcommand will generate the Kubernetes manifest that will be started by kubeadm through the kubeadm init process",
-	Run: func(cmd *cobra.Command, args []string) {
-		// Set the logging level for all subsequent functions
-		log.SetLevel(log.Level(logLevel))
+	Long: `The 'init' subcommand generates a Kubernetes Pod manifest that kubeadm will start as a static Pod during the cluster initialisation phase.
+
+This manifest runs kube-vip on the first control-plane node to advertise the Virtual IP (VIP) for the API server. The VIP is typically configured using ARP (Layer 2) or BGP (dynamic routing).
+
+Required flags for this command:
+  --interface   : The network interface to bind the VIP to (e.g., eth0).
+  --vip or --address : The Virtual IP address or DNS name to use.
+
+Example:
+  kube-vip kubeadm init --interface eth0 --vip 192.168.1.100 --controlplane
+
+The output YAML should be written to the kubeadm manifests directory, e.g.:
+  kube-vip kubeadm init ... > /etc/kubernetes/manifests/kube-vip.yaml`,
+	Run: func(cmd *cobra.Command, _ []string) {
+
 		initConfig.LoadBalancers = append(initConfig.LoadBalancers, initLoadBalancer)
-		// TODO - A load of text detailing what's actually happening
 		err := kubevip.ParseEnvironment(&initConfig)
 		if err != nil {
-			log.Fatalf("Error parsing environment from config: %v", err)
+			log.Error("parsing environment", "err", err)
+			return
+		}
+		if err := initConfig.Validate(); err != nil {
+			log.Error("validating configuration", "err", err)
+			return
 		}
 
 		// TODO - check for certain things VIP/interfaces
 		if initConfig.Interface == "" {
 			_ = cmd.Help()
-			log.Fatalln("No interface is specified for kube-vip to bind to")
+			log.Error("No interface is specified for kube-vip to bind to")
+			return
 		}
 
 		if initConfig.VIP == "" && initConfig.Address == "" {
 			_ = cmd.Help()
-			log.Fatalln("No address is specified for kube-vip to expose services on")
+			log.Error("No address is specified for kube-vip to expose services on")
+			return
 		}
 
 		// Ensure there is an address to generate the CIDR from
-		if initConfig.VIPCIDR == "" && initConfig.Address != "" {
-			initConfig.VIPCIDR, err = GenerateCidrRange(initConfig.Address)
+		if initConfig.VIPSubnet == "" && initConfig.Address != "" {
+			initConfig.VIPSubnet, err = GenerateCidrRange(initConfig.Address, initConfig.DNSMode)
 			if err != nil {
-				log.Fatalln(err)
+				log.Error("generating VIPSubnet", "err", err)
+				return
 			}
 		}
 
-		cfg := kubevip.GeneratePodManifestFromConfig(&initConfig, Release.Version, inCluster)
+		cfg, err := kubevip.GeneratePodManifestFromConfig(&initConfig, image, Release.Version, inCluster)
+		if err != nil {
+			log.Error("unable to create manifest", "err", err)
+			return
+		}
 		fmt.Println(cfg) // output manifest to stdout
 	},
 }
@@ -69,41 +97,63 @@ var kubeKubeadmInit = &cobra.Command{
 var kubeKubeadmJoin = &cobra.Command{
 	Use:   "join",
 	Short: "kube-vip join",
-	Run: func(cmd *cobra.Command, args []string) {
-		// Set the logging level for all subsequent functions
-		log.SetLevel(log.Level(logLevel))
+	Long: `The 'join' subcommand generates a Kubernetes Pod manifest for additional control-plane nodes joining an existing cluster via 'kubeadm join'.
+
+It functions identically to the 'init' subcommand, but is intended for secondary control-plane nodes. It validates that the kubeconfig file (specified by --config, defaulting to /etc/kubernetes/admin.conf) exists on the node to ensure the node can authenticate with the cluster.
+
+Required flags for this command:
+  --interface   : The network interface to bind the VIP to.
+  --vip or --address : The Virtual IP address or DNS name (must match the VIP used during 'init').
+
+Example:
+  kube-vip kubeadm join --interface eth0 --vip 192.168.1.100
+
+The output YAML should be saved to the kubeadm manifests directory on the joining node.`,
+	Run: func(cmd *cobra.Command, _ []string) {
 
 		initConfig.LoadBalancers = append(initConfig.LoadBalancers, initLoadBalancer)
-		// TODO - A load of text detailing what's actually happening
 		err := kubevip.ParseEnvironment(&initConfig)
 		if err != nil {
-			log.Fatalf("Error parsing environment from config: %v", err)
+			log.Error("parsing environment", "err", err)
+			return
+		}
+		if err := initConfig.Validate(); err != nil {
+			log.Error("validating configuration", "err", err)
+			return
 		}
 
 		// TODO - check for certain things VIP/interfaces
 		if initConfig.Interface == "" {
 			_ = cmd.Help()
-			log.Fatalln("No interface is specified for kube-vip to bind to")
+			log.Error("No interface is specified for kube-vip to bind to")
+			return
 		}
 
 		if initConfig.VIP == "" && initConfig.Address == "" {
 			_ = cmd.Help()
-			log.Fatalln("No address is specified for kube-vip to expose services on")
+			log.Error("No address is specified for kube-vip to expose services on")
+			return
 		}
 
 		if _, err := os.Stat(kubeConfigPath); os.IsNotExist(err) {
-			log.Fatalf("Unable to find file [%s]", kubeConfigPath)
+			log.Error("kubeConfig not found", "Path", kubeConfigPath)
+			return
 		}
 
 		// Ensure there is an address to generate the CIDR from
-		if initConfig.VIPCIDR == "" && initConfig.Address != "" {
-			initConfig.VIPCIDR, err = GenerateCidrRange(initConfig.Address)
+		if initConfig.VIPSubnet == "" && initConfig.Address != "" {
+			initConfig.VIPSubnet, err = GenerateCidrRange(initConfig.Address, initConfig.DNSMode)
 			if err != nil {
-				log.Fatalln(err)
+				log.Error("generating VIPSubnet", "err", err)
+				return
 			}
 		}
 
-		cfg := kubevip.GeneratePodManifestFromConfig(&initConfig, Release.Version, inCluster)
+		cfg, err := kubevip.GeneratePodManifestFromConfig(&initConfig, image, Release.Version, inCluster)
+		if err != nil {
+			log.Error("unable to create manifest", "err", err)
+			return
+		}
 		fmt.Println(cfg) // output manifest to stdout
 	},
 }

cmd/kube-vip-manifests.go

@@ -3,23 +3,26 @@ package cmd
 import (
 	"fmt"
 
+	log "log/slog"
+
 	"github.com/kube-vip/kube-vip/pkg/kubevip"
-	log "github.com/sirupsen/logrus"
 	"github.com/spf13/cobra"
-	"gopkg.in/yaml.v2"
 )
 
 // manifests will eventually deprecate the kubeadm set of subcommands
 // manifests will be used to generate:
 // - Pod spec manifest, mainly used for a static pod (kubeadm)
 // - Daemonset manifest, mainly used to run kube-vip as a deamonset within Kubernetes (k3s/rke)
+// - RBAC manifest, used to generate the RBAC permissions for kube-vip
 
-// var inCluster bool
-var taint bool
+var taint, role, rolebinding bool
 
 func init() {
 	kubeManifest.PersistentFlags().BoolVar(&inCluster, "inCluster", false, "Use the incluster token to authenticate to Kubernetes")
+	kubeManifest.PersistentFlags().StringVar(&image, "image", "ghcr.io/kube-vip/kube-vip", "Define a hardcoded image with or without tag for the manifest")
 	kubeManifestDaemon.PersistentFlags().BoolVar(&taint, "taint", false, "Taint the manifest for only running on control planes")
+	kubeManifestRbac.PersistentFlags().BoolVar(&role, "role", false, "Generate only a Role inside the serviceNamespace access")
+	kubeManifestRbac.PersistentFlags().BoolVar(&rolebinding, "rolebinding", false, "Generate only a RoleBinding for namespaced access")
 
 	kubeManifest.AddCommand(kubeManifestPod)
 	kubeManifest.AddCommand(kubeManifestDaemon)
@@ -29,41 +32,71 @@ func init() {
 var kubeManifest = &cobra.Command{
 	Use:   "manifest",
 	Short: "Manifest functions",
-	Run: func(cmd *cobra.Command, args []string) {
+	Long: `This command group provides flexible manifest generation for deploying kube-vip in various Kubernetes environments.
+
+Unlike the 'kubeadm' subcommands, which are tightly coupled to kubeadm's static Pod requirements, these generators produce standard Kubernetes manifests (Pod, DaemonSet, RBAC) that can be used with any Kubernetes distribution (e.g., k3s, RKE, or vanilla Kubernetes).
+
+Subcommands:
+  pod        : Generates a standalone Pod manifest (similar to a static pod).
+  daemonset  : Generates a DaemonSet manifest to run kube-vip on selected nodes.
+  rbac       : Generates the necessary ServiceAccount, Role/ClusterRole, and Binding manifests.
+
+All output is written to stdout as YAML, typically piped to 'kubectl apply -f -' or saved to a file.`,
+	Run: func(cmd *cobra.Command, _ []string) {
 		_ = cmd.Help()
-		// TODO - A load of text detailing what's actually happening
 	},
 }
 
 var kubeManifestPod = &cobra.Command{
 	Use:   "pod",
 	Short: "Generate a Pod Manifest",
-	Run: func(cmd *cobra.Command, args []string) {
+	Long: `Generate a standalone Pod manifest for kube-vip.
+
+This is ideal for environments that do not use DaemonSets or where you want to run kube-vip as a static Pod (similar to the 'kubeadm' subcommand, but without kubeadm-specific assumptions). It includes all the necessary container specifications, volumes, and environment variables derived from the provided flags.
+
+Key flags:
+  --interface : Network interface for the VIP.
+  --vip or --address : The Virtual IP address or DNS name.
+  --image     : Override the container image (default: ghcr.io/kube-vip/kube-vip).
+
+The manifest is generated based on the current configuration flags set on the root command.
+
+Example:
+  kube-vip manifest pod --interface eth0 --vip 10.0.0.100 --controlplane | kubectl apply -f -`,
+	Run: func(cmd *cobra.Command, _ []string) {
 		var err error
 
-		// Set the logging level for all subsequent functions
-		log.SetLevel(log.Level(logLevel))
 		initConfig.LoadBalancers = append(initConfig.LoadBalancers, initLoadBalancer)
-		// TODO - A load of text detailing what's actually happening
 		if err := kubevip.ParseEnvironment(&initConfig); err != nil {
-			log.Fatalf("Error parsing environment from config: %v", err)
+			log.Error("parsing environment", "err", err)
+			return
+		}
+		if err := initConfig.Validate(); err != nil {
+			log.Error("validating configuration", "err", err)
+			return
 		}
 
 		// The control plane has a requirement for a VIP being specified
 		if initConfig.EnableControlPlane && (initConfig.VIP == "" && initConfig.Address == "" && !initConfig.DDNS) {
 			_ = cmd.Help()
-			log.Fatalln("No address is specified for kube-vip to expose services on")
+			log.Error("no address is specified for kube-vip to expose services on")
+			return
 		}
 
 		// Ensure there is an address to generate the CIDR from
-		if initConfig.VIPCIDR == "" && initConfig.Address != "" {
-			initConfig.VIPCIDR, err = GenerateCidrRange(initConfig.Address)
+		if initConfig.VIPSubnet == "" && initConfig.Address != "" {
+			initConfig.VIPSubnet, err = GenerateCidrRange(initConfig.Address, initConfig.DNSMode)
 			if err != nil {
-				log.Fatalln(err)
+				log.Error("config parse", "err", err)
+				return
 			}
 		}
 
-		cfg := kubevip.GeneratePodManifestFromConfig(&initConfig, Release.Version, inCluster)
+		cfg, err := kubevip.GeneratePodManifestFromConfig(&initConfig, image, Release.Version, inCluster)
+		if err != nil {
+			log.Error("unable to create manifest", "err", err)
+			return
+		}
 		fmt.Println(cfg) // output manifest to stdout
 	},
 }
@@ -71,34 +104,50 @@ var kubeManifestPod = &cobra.Command{
 var kubeManifestDaemon = &cobra.Command{
 	Use:   "daemonset",
 	Short: "Generate a Daemonset Manifest",
-	Run: func(cmd *cobra.Command, args []string) {
+	Long: `Generate a DaemonSet manifest to run kube-vip across multiple nodes.
+
+This is the recommended deployment method for production clusters running kube-vip as a service. It ensures that kube-vip runs on all control-plane nodes (or selected nodes via tolerations) and can handle both control-plane HA and service load-balancing.
+
+Flags specific to this subcommand:
+  --taint  : Adds a toleration to the DaemonSet so that pods are scheduled only on nodes with the control-plane taint (node-role.kubernetes.io/control-plane:NoSchedule). This is essential for control-plane-only deployments.
+
+All other standard kube-vip flags (--interface, --vip, --enableARP, --enableBGP, etc.) are respected and embedded into the DaemonSet pod template.
+
+Example:
+  kube-vip manifest daemonset --interface eth0 --vip 192.168.1.100 --controlplane --taint | kubectl apply -f -`,
+	Run: func(cmd *cobra.Command, _ []string) {
 		var err error
 
-		// Set the logging level for all subsequent functions
-		log.SetLevel(log.Level(logLevel))
 		initConfig.LoadBalancers = append(initConfig.LoadBalancers, initLoadBalancer)
-		// TODO - A load of text detailing what's actually happening
 		if err := kubevip.ParseEnvironment(&initConfig); err != nil {
-			log.Fatalf("error parsing environment config: %v", err)
+			log.Error("parsing environment", "err", err)
+			return
+		}
+		if err := initConfig.Validate(); err != nil {
+			log.Error("validating configuration", "err", err)
+			return
 		}
-
-		// TODO - check for certain things VIP/interfaces
-
 		// The control plane has a requirement for a VIP being specified
 		if initConfig.EnableControlPlane && (initConfig.VIP == "" && initConfig.Address == "" && !initConfig.DDNS) {
 			_ = cmd.Help()
-			log.Fatalln("No address is specified for kube-vip to expose services on")
+			log.Error("no address is specified for kube-vip to expose services on")
+			return
 		}
 
 		// Ensure there is an address to generate the CIDR from
-		if initConfig.VIPCIDR == "" && initConfig.Address != "" {
-			initConfig.VIPCIDR, err = GenerateCidrRange(initConfig.Address)
+		if initConfig.VIPSubnet == "" && initConfig.Address != "" {
+			initConfig.VIPSubnet, err = GenerateCidrRange(initConfig.Address, initConfig.DNSMode)
 			if err != nil {
-				log.Fatalln(err)
+				log.Error("config parse", "err", err)
+				return
 			}
 		}
 
-		cfg := kubevip.GenerateDaemonsetManifestFromConfig(&initConfig, Release.Version, inCluster, taint)
+		cfg, err := kubevip.GenerateDaemonsetManifestFromConfig(&initConfig, image, Release.Version, inCluster, taint)
+		if err != nil {
+			log.Error("unable to create manifest", "err", err)
+			return
+		}
 		fmt.Println(cfg) // output manifest to stdout
 	},
 }
@@ -106,33 +155,58 @@ var kubeManifestDaemon = &cobra.Command{
 var kubeManifestRbac = &cobra.Command{
 	Use:   "rbac",
 	Short: "Generate an RBAC Manifest",
-	Run: func(cmd *cobra.Command, args []string) {
-		var err error
+	Long: `Generate the RBAC (Role-Based Access Control) manifests required for kube-vip to interact with the Kubernetes API.
 
-		// Set the logging level for all subsequent functions
-		log.SetLevel(log.Level(logLevel))
+kube-vip needs permissions to watch services, endpoints, configmaps, and manage leader election leases. This command outputs the minimum required ServiceAccount, Role (or ClusterRole), and the corresponding binding.
+
+Flags:
+  --role        : If true, generates a namespaced Role instead of a ClusterRole. The namespace is taken from the root --namespace flag (default: kube-system).
+  --rolebinding : If true, generates a RoleBinding (if --role is also true). If --role is false, a ClusterRoleBinding is generated automatically.
+
+The output is a multi-document YAML (separated by '---'). It is safe to apply directly:
+  kube-vip manifest rbac --role --rolebinding | kubectl apply -f -
+
+Without --role, it generates a ClusterRole and ClusterRoleBinding, which is the default behaviour and suitable for most cluster-wide deployments.`,
+	Run: func(cmd *cobra.Command, _ []string) {
 		initConfig.LoadBalancers = append(initConfig.LoadBalancers, initLoadBalancer)
-		// TODO - A load of text detailing what's actually happening
 		if err := kubevip.ParseEnvironment(&initConfig); err != nil {
-			log.Fatalf("Error parsing environment from config: %v", err)
+			log.Error("parsing environment", "err", err)
+			return
+		}
+		if err := initConfig.Validate(); err != nil {
+			log.Error("validating configuration", "err", err)
+			return
 		}
 
 		// The control plane has a requirement for a VIP being specified
 		if initConfig.EnableControlPlane && (initConfig.VIP == "" && initConfig.Address == "" && !initConfig.DDNS) {
 			_ = cmd.Help()
-			log.Fatalln("No address is specified for kube-vip to expose services on")
+			log.Error("no address is specified for kube-vip to expose services on")
+			return
 		}
 
 		// Ensure there is an address to generate the CIDR from
-		if initConfig.VIPCIDR == "" && initConfig.Address != "" {
-			initConfig.VIPCIDR, err = GenerateCidrRange(initConfig.Address)
+		if initConfig.VIPSubnet == "" && initConfig.Address != "" {
+			var err error
+			initConfig.VIPSubnet, err = GenerateCidrRange(initConfig.Address, initConfig.DNSMode)
 			if err != nil {
-				log.Fatalln(err)
+				log.Error("generating VIPSubnet", "err", err)
+				return
 			}
 		}
+		saCfg := kubevip.GenerateSA(&initConfig)
+		roleCfg := kubevip.GenerateRole(&initConfig, role)
+		if role {
+			rolebinding = true
+		}
+		roleBindingCfg := kubevip.GenerateRoleBinding(rolebinding, saCfg, roleCfg)
 
-		cfg := kubevip.GenerateSA()
-		b, _ := yaml.Marshal(cfg)
-		fmt.Println(string(b)) // output manifest to stdout
+		// Output the YAML manifests to stdout
+		fmt.Println("---") // Separator for YAML documents
+		fmt.Println(kubevip.TransformApplyObjectToManifest(saCfg))
+		fmt.Println("---") // Separator for YAML documents
+		fmt.Println(kubevip.TransformApplyObjectToManifest(roleCfg))
+		fmt.Println("---") // Separator for YAML documents
+		fmt.Println(kubevip.TransformApplyObjectToManifest(roleBindingCfg))
 	},
 }

cmd/kube-vip.go

@@ -6,33 +6,36 @@ import (
 	"net"
 	"net/http"
 	"os"
+	"slices"
+	"strconv"
 	"strings"
+	"sync"
 	"time"
 
-	"github.com/prometheus/client_golang/prometheus"
+	log "log/slog"
+
 	"github.com/prometheus/client_golang/prometheus/promhttp"
-	log "github.com/sirupsen/logrus"
 	"github.com/spf13/cobra"
 	"github.com/vishvananda/netlink"
+	"golang.org/x/sys/unix"
 
-	"github.com/kube-vip/kube-vip/pkg/equinixmetal"
+	"github.com/kube-vip/kube-vip/pkg/debouncer"
 	"github.com/kube-vip/kube-vip/pkg/kubevip"
 	"github.com/kube-vip/kube-vip/pkg/manager"
+	"github.com/kube-vip/kube-vip/pkg/metrics"
+	"github.com/kube-vip/kube-vip/pkg/utils"
 	"github.com/kube-vip/kube-vip/pkg/vip"
 )
 
+// Is an option to set the image
+var image string
+
 // Is kube-vip running within cluster
 var inCluster bool
 
 // ConfigMap name within a Kubernetes cluster
 var configMap string
 
-// Configure the level of logging
-var logLevel uint32
-
-// Provider Config
-var providerConfig string
-
 // Points to a kubernetes configuration file
 var kubeConfigPath string
 
@@ -49,29 +52,33 @@ var (
 )
 
 var kubeVipCmd = &cobra.Command{
-	Use:   "kube-vip",
-	Short: "This is a server for providing a Virtual IP and load-balancer for the Kubernetes control-plane",
+	Use:           "kube-vip",
+	Short:         "This is a server for providing a Virtual IP and load-balancer for the Kubernetes control-plane",
+	SilenceErrors: true,
 }
 
 func init() {
 	// Basic flags
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.Interface, "interface", "", "Name of the interface to bind to")
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.ServicesInterface, "serviceInterface", "", "Name of the interface to bind to (for services)")
+	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.AllowInterfaceNotUp, "allowInterfaceNotUp", false, "Allow kube-vip to start even if the interface is not up")
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.VIP, "vip", "", "The Virtual IP address")
-	kubeVipCmd.PersistentFlags().StringVar(&initConfig.VIPSubnet, "vipSubnet", "", "The Virtual IP address subnet e.g. /32 /24 /8 etc..")
+	kubeVipCmd.PersistentFlags().StringVar(&initConfig.VIPSubnet, "vipSubnet", "", "The Virtual IP address subnet e.g. /32 /24 /8 etc.. (Default to 32 for IPv4 and 128 for IPv6)")
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.NodeName, "nodeName", "", "Name to be used for lease holder. Must be unique for each node/instance")
 
-	kubeVipCmd.PersistentFlags().StringVar(&initConfig.VIPCIDR, "cidr", "", "The CIDR range for the virtual IP address. Default to 32 for IPv4 and 128 for IPv6") // todo: deprecate
-
+	// VIP flags
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.Address, "address", "", "an address (IP or DNS name) to use as a VIP")
-	kubeVipCmd.PersistentFlags().IntVar(&initConfig.Port, "port", 6443, "Port for the VIP")
+	kubeVipCmd.PersistentFlags().Uint16Var(&initConfig.Port, "port", 6443, "Port for the VIP")
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableARP, "arp", false, "Enable Arp for VIP changes")
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableWireguard, "wireguard", false, "Enable Wireguard for services VIPs")
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableRoutingTable, "table", false, "Enable Routing Table for services VIPs")
+	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.PreserveVIPOnLeadershipLoss, "preserveVipOnLeadershipLoss", false, "Preserve ARP VIP addresses on interface when leadership is lost (default: false for backward compatibility)")
+	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.LoseLeadership, "loseLeadership", false, "Lose leadership when VIP interface goes down")
+	kubeVipCmd.PersistentFlags().IntVar(&initConfig.LoseLeadershipTimeoutSeconds, "loseLeadershiptTimeoutSeconds", 30, "Timeout before re-electing a leader when the VIP interface is down")
 
 	// LoadBalancer flags
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableLoadBalancer, "enableLoadBalancer", false, "enable loadbalancing on the VIP with IPVS")
-	kubeVipCmd.PersistentFlags().IntVar(&initConfig.LoadBalancerPort, "lbPort", 6443, "loadbalancer port for the VIP")
+	kubeVipCmd.PersistentFlags().Uint16Var(&initConfig.LoadBalancerPort, "lbPort", 6443, "loadbalancer port for the VIP")
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.LoadBalancerForwardingMethod, "lbForwardingMethod", "local", "loadbalancer forwarding method")
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.DDNS, "ddns", false, "use Dynamic DNS + DHCP to allocate VIP for address")
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.MirrorDestInterface, "mirrorDestInterface", "", "network interface where all traffic that traverses the service interface will be mirrored to. Source interface will use default interface is servicesInterface is not set.")
@@ -80,19 +87,13 @@ func init() {
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableLeaderElection, "leaderElection", false, "Use the Kubernetes leader election mechanism for clustering")
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.LeaderElectionType, "leaderElectionType", "kubernetes", "Defines the backend to run the leader election: kubernetes or etcd. Defaults to kubernetes.")
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.LeaseName, "leaseName", "plndr-cp-lock", "Name of the lease that is used for leader election")
-	kubeVipCmd.PersistentFlags().IntVar(&initConfig.LeaseDuration, "leaseDuration", 5, "Length of time (in seconds) a Kubernetes leader lease can be held for")
-	kubeVipCmd.PersistentFlags().IntVar(&initConfig.RenewDeadline, "leaseRenewDuration", 3, "Length of time (in seconds) a Kubernetes leader can attempt to renew its lease")
-	kubeVipCmd.PersistentFlags().IntVar(&initConfig.RetryPeriod, "leaseRetry", 1, "Length of time (in seconds) the LeaderElector clients should wait between tries of actions")
-
-	// Equinix Metal flags
-	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableMetal, "metal", false, "This will use the Equinix Metal API (requires the token ENV) to update the EIP <-> VIP")
-	kubeVipCmd.PersistentFlags().StringVar(&initConfig.MetalAPIKey, "metalKey", "", "The API token for authenticating with the Equinix Metal API")
-	kubeVipCmd.PersistentFlags().StringVar(&initConfig.MetalProject, "metalProject", "", "The name of project already created within Equinix Metal")
-	kubeVipCmd.PersistentFlags().StringVar(&initConfig.MetalProjectID, "metalProjectID", "", "The ID of project already created within Equinix Metal")
-	kubeVipCmd.PersistentFlags().StringVar(&initConfig.ProviderConfig, "provider-config", "", "The path to a provider configuration")
+	kubeVipCmd.PersistentFlags().IntVar(&initConfig.LeaseDuration, "leaseDuration", 15, "Length of time (in seconds) a Kubernetes leader lease can be held for")
+	kubeVipCmd.PersistentFlags().IntVar(&initConfig.RenewDeadline, "leaseRenewDuration", 10, "Length of time (in seconds) a Kubernetes leader can attempt to renew its lease")
+	kubeVipCmd.PersistentFlags().IntVar(&initConfig.RetryPeriod, "leaseRetry", 2, "Length of time (in seconds) the LeaderElector clients should wait between tries of actions")
 
 	// BGP flags
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableBGP, "bgp", false, "This will enable BGP support within kube-vip")
+	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.BGPAttachIPToInterface, "bgpAttachIPToInterface", false, "Assign BGP service VIPs to the configured interface")
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.BGPConfig.RouterID, "bgpRouterID", "", "The routerID for the bgp server")
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.BGPConfig.SourceIF, "sourceIF", "", "The source interface for bgp peering (not to be used with sourceIP)")
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.BGPConfig.SourceIP, "sourceIP", "", "The source address for bgp peering (not to be used with sourceIF)")
@@ -106,18 +107,28 @@ func init() {
 	kubeVipCmd.PersistentFlags().StringSliceVar(&initConfig.BGPPeers, "bgppeers", []string{}, "Comma separated BGP Peer, format: address:as:password:multihop")
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.Annotations, "annotations", "", "Set Node annotations prefix for parsing")
 
+	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.BGPConfig.Zebra.Enabled, "zebra", false, "This will enable Zebra support within kube-vip")
+	kubeVipCmd.PersistentFlags().StringVar(&initConfig.BGPConfig.Zebra.URL, "zebraUrl", "unix:/var/run/frr/zserv.api", "Path to the unix domain socket for connecting to Zebra daemon")
+	kubeVipCmd.PersistentFlags().Uint32Var(&initConfig.BGPConfig.Zebra.Version, "zebraVersion", 6, "Zebra API Version")
+	kubeVipCmd.PersistentFlags().StringVar(&initConfig.BGPConfig.Zebra.SoftwareName, "zebraSoftwareName", "frr8.3", "Software Name for Zebra")
+	kubeVipCmd.PersistentFlags().StringVar(&initConfig.ControlPlaneHealthCheck.Address, "controlPlaneHealthCheckAddress", "", "URL to poll for the control-plane health check when using BGP without leader election")
+	kubeVipCmd.PersistentFlags().IntVar(&initConfig.ControlPlaneHealthCheck.PeriodSeconds, "controlPlaneHealthCheckPeriodSeconds", 5, "Seconds between control-plane health checks")
+	kubeVipCmd.PersistentFlags().IntVar(&initConfig.ControlPlaneHealthCheck.TimeoutSeconds, "controlPlaneHealthCheckTimeoutSeconds", 3, "Timeout for each control-plane health check request")
+	kubeVipCmd.PersistentFlags().IntVar(&initConfig.ControlPlaneHealthCheck.FailureThreshold, "controlPlaneHealthCheckFailureThreshold", 3, "Consecutive control-plane health check failures before withdrawing the BGP route")
+	kubeVipCmd.PersistentFlags().StringVar(&initConfig.ControlPlaneHealthCheck.CAPath, "controlPlaneHealthCheckCAPath", "", "Path to CA certificate for TLS verification when the control-plane health check URL is HTTPS")
+
 	// Namespace for kube-vip
 	kubeVipCmd.PersistentFlags().StringVarP(&initConfig.Namespace, "namespace", "n", "kube-system", "The namespace for the configmap defined within the cluster")
 
 	// Manage logging
-	kubeVipCmd.PersistentFlags().Uint32Var(&logLevel, "log", 4, "Set the level of logging")
+	kubeVipCmd.PersistentFlags().Int32Var(&initConfig.Logging, "log", 0, "Set the level of logging")
 
 	// Service flags
 	kubeVipService.Flags().StringVarP(&configMap, "configMap", "c", "plndr", "The configuration map defined within the cluster")
 
 	// Routing Table flags
 	kubeVipCmd.PersistentFlags().IntVar(&initConfig.RoutingTableID, "tableID", 198, "The routing table used for all table entries")
-	kubeVipCmd.PersistentFlags().IntVar(&initConfig.RoutingTableType, "tableType", 0, "The type of route that will be added to the routing table")
+	kubeVipCmd.PersistentFlags().IntVar(&initConfig.RoutingTableType, "tableType", unix.RTN_UNICAST, "The type of route that will be added to the routing table")
 	kubeVipCmd.PersistentFlags().IntVar(&initConfig.RoutingProtocol, "routingProtocol", 248, "The routing protocol value used to create routes")
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.CleanRoutingTable, "cleanRoutingTable", false, "Clean routing table of redundant routes on start")
 
@@ -128,16 +139,22 @@ func init() {
 
 	// Extended behaviour flags
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableServicesElection, "servicesElection", false, "Enable leader election per kubernetes service")
-	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.LoadBalancerClassOnly, "lbClassOnly", false, "Enable load balancing only for services with LoadBalancerClass \"kube-vip.io/kube-vip-class\"")
-	kubeVipCmd.PersistentFlags().StringVar(&initConfig.LoadBalancerClassName, "lbClassName", "kube-vip.io/kube-vip-class", "Name of load balancer class for kube-VIP, defaults to \"kube-vip.io/kube-vip-class\"")
+	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.LoadBalancerClassOnly, "lbClassOnly", false, fmt.Sprintf("Enable load balancing only for services with LoadBalancerClass %q", kubevip.LBClassName))
+	kubeVipCmd.PersistentFlags().StringVar(&initConfig.LoadBalancerClassName, "lbClassName", kubevip.LBClassName, fmt.Sprintf("Name of load balancer class for kube-VIP, defaults to %q", kubevip.LBClassName))
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.LoadBalancerClassLegacyHandling, "lbClassNameLegacyHandling", true, "Use legacy LoadBalancer class name handling (e.g. accepting services both with empty and non-empty class)")
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableServiceSecurity, "onlyAllowTrafficServicePorts", false, "Only allow traffic to service ports, others will be dropped, defaults to false")
-	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableNodeLabeling, "enableNodeLabeling", false, "Enable leader node labeling with \"kube-vip.io/has-ip=<VIP address>\", defaults to false")
+	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableNodeLabeling, "enableNodeLabeling", false, fmt.Sprintf("Enable leader node labeling with %q, defaults to false", kubevip.HasIP))
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.ServicesLeaseName, "servicesLeaseName", "plndr-svcs-lock", "Name of the lease that is used for leader election for services (in arp mode)")
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.DNSMode, "dnsMode", "first", "Name of the mode that DNS lookup will be performed (first, ipv4, ipv6, dual)")
+	kubeVipCmd.PersistentFlags().StringVar(&initConfig.DHCPMode, "dhcpMode", "", "Mode DHCP resolving will use to obtain IP addresses (ipv4, ipv6, dual)")
+	kubeVipCmd.PersistentFlags().UintVar(&initConfig.DHCPBackoffAttempts, "dhcpBackoffAttempts", kubevip.DefaultDHCPBackoffAttempts,
+		fmt.Sprintf("number of times DHCP client will try to obtain an IP address (defaults to: %d, 0 for unlimited retries)", kubevip.DefaultDHCPBackoffAttempts))
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.DisableServiceUpdates, "disableServiceUpdates", false, "If true, kube-vip will process services as usual, but will not update service's Status.LoadBalancer.Ingress slice")
-	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableEndpointSlices, "enableEndpointSlices", false, "If enabled, kube-vip will only advertise services, but will use EndpointSlices instead of endpoints to get IPs of Pods")
+	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EnableEndpoints, "enableEndpoints", false, "If enabled, kube-vip will only advertise services, but will use the (deprecated since v1.33) endpoints for IP addresses")
 	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.LoInterfaceGlobalScope, "loInterfaceGlobalScope", false, "If true, kube-vip will set global scope when using the lo interface, otherwise a host scope will be used by default")
+	kubeVipCmd.PersistentFlags().IntVar(&initConfig.HealthCheckPort, "healthCheckPort", 0, "If set to non-zero (> 1024), then this is the port that the healthcheck will listen on")
+	kubeVipCmd.PersistentFlags().StringVar(&initConfig.DebounceTime, "debounceTime", debouncer.DefaultTime,
+		"Configures the time that the event debouncer will wait for the events arrival (default 0s - debouncer disabled, enable with min. 200ms)")
 
 	// Prometheus HTTP Server
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.PrometheusHTTPServer, "prometheusHTTPServer", ":2112", "Host and port used to expose Prometheus metrics via an HTTP server")
@@ -149,29 +166,44 @@ func init() {
 	kubeVipCmd.PersistentFlags().StringSliceVar(&initConfig.Etcd.Endpoints, "etcdEndpoints", nil, "Etcd member endpoints")
 
 	// Kubernetes client specific flags
-
 	kubeVipCmd.PersistentFlags().StringVar(&initConfig.K8sConfigFile, "k8sConfigPath", "/etc/kubernetes/admin.conf", "Path to the configuration file used with the Kubernetes client")
 
+	// Configuration file flag
+	kubeVipCmd.PersistentFlags().StringVar(&initConfig.ConfigFile, "config-file", "", "Path to a JSON/YAML configuration file to load settings from")
+	kubeVipCmd.PersistentFlags().StringVar(&initConfig.InstanceName, "instanceName", "", "Unique name for this kube-vip instance (currently used to isolate nftables egress tables)")
+
+	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.EgressWithNftables, "egressWithNftables", true, "Use nftables-based egress implementation")
+
+	kubeVipCmd.PersistentFlags().BoolVar(&initConfig.PerServiceElectionOnDemand, "perServiceElectionOnDemand", false, "Allow kube-vip to use per-service election for annotated services")
+
 	kubeVipCmd.AddCommand(kubeKubeadm)
 	kubeVipCmd.AddCommand(kubeManifest)
 	kubeVipCmd.AddCommand(kubeVipManager)
 	kubeVipCmd.AddCommand(kubeVipSample)
 	kubeVipCmd.AddCommand(kubeVipService)
 	kubeVipCmd.AddCommand(kubeVipVersion)
+
+	// Set the logging level for all subsequent functions
+	log.SetLogLoggerLevel(log.Level(initConfig.Logging))
 }
 
 // Execute - starts the command parsing process
-func Execute() {
-	if err := kubeVipCmd.Execute(); err != nil {
-		fmt.Println(err)
-		os.Exit(1)
+func Execute() int {
+	cmd, err := kubeVipCmd.ExecuteC()
+	if err != nil {
+		log.Error("command failed", "err", err)
+		if cmd == kubeVipCmd {
+			_ = cmd.Usage()
+		}
+		return 1
 	}
+	return 0
 }
 
 var kubeVipVersion = &cobra.Command{
 	Use:   "version",
 	Short: "Version and Release information about the Kubernetes Virtual IP Server",
-	Run: func(cmd *cobra.Command, args []string) {
+	Run: func(cmd *cobra.Command, args []string) { //nolint TODO
 		fmt.Printf("Kube-VIP Release Information\n")
 		fmt.Printf("Version:  %s\n", Release.Version)
 		fmt.Printf("Build:    %s\n", Release.Build)
@@ -181,7 +213,7 @@ var kubeVipVersion = &cobra.Command{
 var kubeVipSample = &cobra.Command{
 	Use:   "sample",
 	Short: "Generate a Sample configuration",
-	Run: func(cmd *cobra.Command, args []string) {
+	Run: func(cmd *cobra.Command, args []string) { //nolint TODO
 		_ = cmd.Help()
 	},
 }
@@ -189,18 +221,36 @@ var kubeVipSample = &cobra.Command{
 var kubeVipService = &cobra.Command{
 	Use:   "service",
 	Short: "Start the Virtual IP / Load balancer as a service within a Kubernetes cluster",
-	Run: func(cmd *cobra.Command, args []string) {
-		// Set the logging level for all subsequent functions
-		log.SetLevel(log.Level(logLevel))
+	RunE: func(cmd *cobra.Command, args []string) error { //nolint TODO
+		cmd.SilenceUsage = true
 
-		// parse environment variables, these will overwrite anything loaded or flags
-		err := kubevip.ParseEnvironment(&initConfig)
-		if err != nil {
-			log.Fatalln(err)
+		// Load configuration from file if specified (lowest priority)
+		if initConfig.ConfigFile != "" {
+			err := kubevip.MergeConfigFromFile(&initConfig, initConfig.ConfigFile)
+			if err != nil {
+				return fmt.Errorf("loading config file: %w", err)
+			}
 		}
 
+		// parse environment variables, these will overwrite anything loaded from config file
+		err := kubevip.ParseEnvironment(&initConfig)
+		if err != nil {
+			return fmt.Errorf("parsing environment: %w", err)
+		}
+		if err := initConfig.Validate(); err != nil {
+			return fmt.Errorf("validating configuration: %w", err)
+		}
+
+		// Change RTN_UNSPEC to default type
+		if initConfig.RoutingProtocol == unix.RTN_UNSPEC {
+			initConfig.RoutingProtocol = unix.RTN_UNICAST
+		}
+
+		// Set the logging level for all subsequent functions
+		log.SetLogLoggerLevel(log.Level(initConfig.Logging))
+
 		if err := initConfig.CheckInterface(); err != nil {
-			log.Fatalln(err)
+			return fmt.Errorf("checking interface: %w", err)
 		}
 
 		// User Environment variables as an option to make manifest clearer
@@ -209,92 +259,153 @@ var kubeVipService = &cobra.Command{
 			configMap = envConfigMap
 		}
 
-		// Ensure there is an address to generate the CIDR from
-		if initConfig.VIPCIDR == "" && initConfig.Address != "" {
-			initConfig.VIPCIDR, err = GenerateCidrRange(initConfig.Address)
-			if err != nil {
-				log.Fatalln(err)
+		// Legacy vip_address requires vip_subnet for control-plane ARP, BGP, and Routing Table modes.
+		if initConfig.EnableControlPlane &&
+			(initConfig.EnableARP || initConfig.EnableBGP || initConfig.EnableRoutingTable) {
+			if err := initConfig.CheckSubnetExists(); err != nil {
+				return fmt.Errorf("checking subnet exists if vip_address defined: %w", err)
 			}
 		}
 
+		// Ensure there is an address to generate the CIDR from
+		if initConfig.VIPSubnet == "" && initConfig.Address != "" {
+			initConfig.VIPSubnet, err = GenerateCidrRange(initConfig.Address, initConfig.DNSMode)
+			if err != nil {
+				return fmt.Errorf("generating CIDR: %w", err)
+			}
+		}
+
+		ctx, cancel := context.WithCancel(cmd.Context())
+		defer cancel()
+
 		// Define the new service manager
-		mgr, err := manager.New(configMap, &initConfig)
+		mgr, err := manager.New(ctx, configMap, &initConfig)
 		if err != nil {
-			log.Fatalf("%v", err)
+			return fmt.Errorf("new manager: %w", err)
 		}
 
 		// Start the service manager, this will watch the config Map and construct kube-vip services for it
-		err = mgr.Start()
+		err = mgr.Start(ctx)
 		if err != nil {
-			log.Fatalf("%v", err)
+			return fmt.Errorf("manager start: %w", err)
 		}
+		return nil
 	},
 }
 
 var kubeVipManager = &cobra.Command{
 	Use:   "manager",
 	Short: "Start the kube-vip manager",
-	Run: func(cmd *cobra.Command, args []string) {
-		// parse environment variables, these will overwrite anything loaded or flags
-		err := kubevip.ParseEnvironment(&initConfig)
-		if err != nil {
-			log.Fatalln(err)
-		}
+	RunE: func(cmd *cobra.Command, args []string) error { //nolint TODO
+		cmd.SilenceUsage = true
 
-		// Ensure there is an address to generate the CIDR from
-		if initConfig.VIPCIDR == "" && initConfig.Address != "" {
-			initConfig.VIPCIDR, err = GenerateCidrRange(initConfig.Address)
+		// Load configuration from file if specified (lowest priority)
+		if initConfig.ConfigFile != "" {
+			err := kubevip.MergeConfigFromFile(&initConfig, initConfig.ConfigFile)
 			if err != nil {
-				log.Fatalln(err)
+				return fmt.Errorf("loading config file: %w", err)
 			}
 		}
 
+		// parse environment variables, these will overwrite anything loaded from config file
+		err := kubevip.ParseEnvironment(&initConfig)
+		if err != nil {
+			return fmt.Errorf("parsing environment: %w", err)
+		}
+		if err := initConfig.Validate(); err != nil {
+			return fmt.Errorf("validating configuration: %w", err)
+		}
+
+		// Change RTN_UNSPEC to default type
+		if initConfig.RoutingProtocol == unix.RTN_UNSPEC {
+			initConfig.RoutingProtocol = unix.RTN_UNICAST
+		}
+
 		// Set the logging level for all subsequent functions
-		log.SetLevel(log.Level(initConfig.Logging))
+		log.SetLogLoggerLevel(log.Level(initConfig.Logging))
+
+		// Legacy vip_address requires vip_subnet for control-plane ARP, BGP, and Routing Table modes.
+		if initConfig.EnableControlPlane &&
+			(initConfig.EnableARP || initConfig.EnableBGP || initConfig.EnableRoutingTable) {
+			if err := initConfig.CheckSubnetExists(); err != nil {
+				return fmt.Errorf("checking subnet exists if vip_address defined: %w", err)
+			}
+		}
+
+		// Ensure there is an address to generate the CIDR from
+		if initConfig.VIPSubnet == "" && initConfig.Address != "" {
+			initConfig.VIPSubnet, err = GenerateCidrRange(initConfig.Address, initConfig.DNSMode)
+			if err != nil {
+				return fmt.Errorf("generating CIDR: %w", err)
+			}
+		}
 
 		// Welome messages
-		log.Infof("Starting kube-vip.io [%s]", Release.Version)
-		log.Debugf("Build kube-vip.io [%s]", Release.Build)
+		log.Info("kube-vip.io", "version", Release.Version, "build", Release.Build)
+
+		wg := sync.WaitGroup{}
+		defer wg.Wait()
+
+		// create main manager context
+		ctx, cancel := context.WithCancel(cmd.Context())
+		defer cancel()
 
 		// start prometheus server
 		if initConfig.PrometheusHTTPServer != "" {
-			go servePrometheusHTTPServer(cmd.Context(), PrometheusHTTPServerConfig{
-				Addr: initConfig.PrometheusHTTPServer,
+			wg.Go(func() {
+				servePrometheusHTTPServer(ctx, PrometheusHTTPServerConfig{
+					Addr: initConfig.PrometheusHTTPServer,
+				})
 			})
 		}
 
 		// Determine the kube-vip mode
-		var mode string
+		var (
+			mode         string
+			modesEnabled int
+		)
 		if initConfig.EnableARP {
 			mode = "ARP"
+			modesEnabled++
 		}
 
 		if initConfig.EnableBGP {
 			mode = "BGP"
+			modesEnabled++
 		}
 
 		if initConfig.EnableWireguard {
 			mode = "Wireguard"
+			modesEnabled++
 		}
 
 		if initConfig.EnableRoutingTable {
 			mode = "Routing Table"
+			modesEnabled++
+		}
+
+		if mode == "" {
+			return fmt.Errorf("no valid kube-vip mode detected, ensure a supported mode is configured")
+		}
+
+		if modesEnabled > 1 {
+			return fmt.Errorf("multiple kube-vip modes detected, ensure only one mode is configured")
 		}
 
 		// Provide configuration to output/logging
-		log.Infof("namespace [%s], Mode: [%s], Features(s): Control Plane:[%t], Services:[%t]", initConfig.Namespace, mode, initConfig.EnableControlPlane, initConfig.EnableServices)
+		log.Info("starting", "namespace", initConfig.Namespace, "Mode", mode, "Control Plane", initConfig.EnableControlPlane, "Services", initConfig.EnableServices)
 
 		// End if nothing is enabled
 		if !initConfig.EnableServices && !initConfig.EnableControlPlane {
-			log.Fatalln("no features are enabled")
+			return fmt.Errorf("no features are enabled")
 		}
 
-		if !initConfig.EnableARP && strings.Contains(initConfig.VIPCIDR, kubevip.Auto) {
-			log.Fatalln("auto subnet discovery cannot be used outside ARP mode")
+		if !initConfig.EnableARP && strings.Contains(initConfig.VIPSubnet, kubevip.Auto) {
+			return fmt.Errorf("auto subnet discovery cannot be used outside ARP mode")
 		}
 
-		if strings.Contains(initConfig.VIPCIDR, kubevip.Auto) && initConfig.Address != "" {
-			log.Fatalln("auto subnet discovery cannot be used if VIP address was provided")
+		if strings.Contains(initConfig.VIPSubnet, kubevip.Auto) && initConfig.Address != "" {
+			return fmt.Errorf("auto subnet discovery cannot be used if VIP address was provided")
 		}
 
 		// If we're using wireguard then all traffic goes through the wg0 interface
@@ -304,48 +415,50 @@ var kubeVipManager = &cobra.Command{
 				initConfig.Interface = "wg0"
 			}
 
-			log.Infof("configuring Wireguard networking")
+			log.Info("configuring Wireguard networking")
 			l, err := netlink.LinkByName(initConfig.Interface)
 			if err != nil {
 				if strings.Contains(err.Error(), "Link not found") {
-					log.Warnf("interface \"%s\" doesn't exist, attempting to create wireguard interface", initConfig.Interface)
+					log.Warn("attempting to create wireguard interface", "interface not found", initConfig.Interface)
 					err = netlink.LinkAdd(&netlink.Wireguard{LinkAttrs: netlink.LinkAttrs{Name: initConfig.Interface}})
 					if err != nil {
-						log.Fatalln(err)
+						return fmt.Errorf("adding link: %w", err)
 					}
 					l, err = netlink.LinkByName(initConfig.Interface)
 					if err != nil {
-						log.Fatalln(err)
+						return fmt.Errorf("finding link: %w", err)
 					}
 				}
 			}
 			err = netlink.LinkSetUp(l)
 			if err != nil {
-				log.Fatalln(err)
+				return fmt.Errorf("setting link UP: %w", err)
 			}
 
 		} else { // if we're not using Wireguard then we'll need to use an actual interface
 			// Check if the interface needs auto-detecting
 			if initConfig.Interface == "" {
-				log.Infof("No interface is specified for VIP in config, auto-detecting default Interface")
+				log.Info("No interface is specified for VIP in config, auto-detecting default Interface")
 				defaultIF, err := vip.GetDefaultGatewayInterface()
 				if err != nil {
 					_ = cmd.Help()
-					log.Fatalf("unable to detect default interface -> [%v]", err)
+					return fmt.Errorf("detecting interface: %w", err)
 				}
 				initConfig.Interface = defaultIF.Name
-				log.Infof("kube-vip will bind to interface [%s]", initConfig.Interface)
+				log.Info("kube-vip bind", "interface", initConfig.Interface)
 
-				go func() {
-					if err := vip.MonitorDefaultInterface(context.TODO(), defaultIF); err != nil {
-						log.Fatalf("crash: %s", err.Error())
+				wg.Go(func() {
+					if err := vip.MonitorDefaultInterface(ctx, defaultIF); err != nil {
+
+						log.Error("interface monitor", "err", err)
+						return
 					}
-				}()
+				})
 			}
 		}
-		// Perform a check on th state of the interface
+		// Perform a check on the state of the interface
 		if err := initConfig.CheckInterface(); err != nil {
-			log.Fatalln(err)
+			return fmt.Errorf("checking interface: %w", err)
 		}
 
 		// User Environment variables as an option to make manifest clearer
@@ -354,31 +467,21 @@ var kubeVipManager = &cobra.Command{
 			configMap = envConfigMap
 		}
 
-		// If Equinix Metal is enabled and there is a provider configuration passed
-		if initConfig.EnableMetal {
-			if providerConfig != "" {
-				providerAPI, providerProject, err := equinixmetal.GetPacketConfig(providerConfig)
-				if err != nil {
-					log.Fatalf("%v", err)
-				}
-				initConfig.MetalAPIKey = providerAPI
-				initConfig.MetalProject = providerProject
-			}
-		}
-
 		// Define the new service manager
-		mgr, err := manager.New(configMap, &initConfig)
+		mgr, err := manager.New(ctx, configMap, &initConfig)
 		if err != nil {
-			log.Fatalf("configuring new Manager error -> %v", err)
+			return fmt.Errorf("new manager: %w", err)
 		}
 
-		prometheus.MustRegister(mgr.PrometheusCollector()...)
+		metrics.RegisterPrometheusMetrics()
+		metrics.BuildInfo.WithLabelValues(Release.Version, Release.Build, initConfig.NodeName)
 
 		// Start the service manager, this will watch the config Map and construct kube-vip services for it
-		err = mgr.Start()
+		err = mgr.Start(ctx)
 		if err != nil {
-			log.Fatalf("starting new Manager error -> %v", err)
+			return fmt.Errorf("start manager: %w", err)
 		}
+		return nil
 	},
 }
 
@@ -392,7 +495,7 @@ func servePrometheusHTTPServer(ctx context.Context, config PrometheusHTTPServerC
 	var err error
 	mux := http.NewServeMux()
 	mux.Handle("/metrics", promhttp.Handler())
-	mux.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
+	mux.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) { //nolint TODO
 		_, _ = w.Write([]byte(`<html>
 			<head><title>kube-vip</title></head>
 			<body>
@@ -408,49 +511,70 @@ func servePrometheusHTTPServer(ctx context.Context, config PrometheusHTTPServerC
 		ReadHeaderTimeout: 2 * time.Second,
 	}
 
-	go func() {
-		if err = srv.ListenAndServe(); err != nil && err != http.ErrServerClosed {
-			log.Fatalf("listen:%+s\n", err)
-		}
-	}()
+	wg := sync.WaitGroup{}
 
-	log.Printf("prometheus HTTP server started")
+	wg.Go(func() {
+		if err = srv.ListenAndServe(); err != nil && err != http.ErrServerClosed {
+			log.Error("prometheus HTTP server", "err", err)
+			return
+		}
+	})
+
+	log.Info("prometheus HTTP server started")
 
 	<-ctx.Done()
 
-	log.Printf("prometheus HTTP server stopped")
-
+	// create prometheus shutdown context (independent of other contexts)
 	ctxShutDown, cancel := context.WithTimeout(context.Background(), 5*time.Second)
 	defer func() {
 		cancel()
 	}()
 
 	if err = srv.Shutdown(ctxShutDown); err != nil {
-		log.Fatalf("server Shutdown Failed:%+s", err)
+		log.Error("shutting down prometheus HTTP server", "err", err)
+		return
 	}
 
 	if err == http.ErrServerClosed {
 		err = nil
 	}
+
+	log.Info("prometheus HTTP server stopped")
+
+	wg.Wait()
 }
 
-func GenerateCidrRange(address string) (string, error) {
+func GenerateCidrRange(address string, dnsMode string) (string, error) {
 	var cidrs []string
 
 	addresses := strings.Split(address, ",")
 	for _, a := range addresses {
 		ip := net.ParseIP(a)
-
 		if ip == nil {
-			return "", fmt.Errorf("invalid IP address: %s from [%s]", a, address)
-		}
-
-		if ip.To4() != nil {
-			cidrs = append(cidrs, "32")
+			// we probably are a DNS name
+			ips, err := utils.LookupHost(a, dnsMode, true)
+			if len(ips) == 0 || err != nil {
+				return "", fmt.Errorf("invalid IP address: %s from [%s], %v", a, address, err)
+			}
+			for _, addr := range ips {
+				ip = net.ParseIP(addr)
+				if ip.To4() != nil {
+					cidrs = append(cidrs, strconv.Itoa(vip.DefaultMaskIPv4))
+				} else {
+					cidrs = append(cidrs, strconv.Itoa(vip.DefaultMaskIPv6))
+				}
+			}
 		} else {
-			cidrs = append(cidrs, "128")
+			if ip.To4() != nil {
+				cidrs = append(cidrs, strconv.Itoa(vip.DefaultMaskIPv4))
+			} else {
+				cidrs = append(cidrs, strconv.Itoa(vip.DefaultMaskIPv6))
+			}
 		}
 	}
-
+	// compact as DNS could have a lot of addresses
+	slices.Sort(cidrs)
+	cidrs = slices.Compact(cidrs)
+	slices.Reverse(cidrs)
 	return strings.Join(cidrs, ","), nil
 }

demo/client/main.go

@@ -19,7 +19,7 @@ func main() {
 	var errorOccurred bool
 	for {
 		p := make([]byte, 2048)
-		conn, err := net.Dial("udp", fmt.Sprintf("%s:%d", *address, *port))
+		conn, err := net.Dial("udp", net.JoinHostPort(*address, fmt.Sprint(port)))
 		if err != nil {
 			if !errorOccurred {
 				errorTime = time.Now()

demo/server/main.go

@@ -60,7 +60,7 @@ func main() {
 				fmt.Println("error: ", err)
 			}
 
-			ServerConn.WriteTo(buf[0:n])
+			ServerConn.WriteTo(buf[0:n], addr)
 		}
 	}
 }

example/deploy/0.1.2.yaml

@@ -1,74 +0,0 @@
-apiVersion: apps/v1
-kind: Deployment
-metadata:
-  creationTimestamp: null
-  labels:
-    app: kube-vip-cluster
-  name: kube-vip-cluster
-spec:
-  replicas: 3
-  selector:
-    matchLabels:
-      app: kube-vip-cluster
-  strategy: {}
-  template:
-    metadata:
-      creationTimestamp: null
-      labels:
-        app: kube-vip-cluster
-    spec:
-      affinity:
-        podAntiAffinity:
-          requiredDuringSchedulingIgnoredDuringExecution:
-          - labelSelector:
-              matchExpressions:
-                - key: "app"
-                  operator: In
-                  values:
-                  - kube-vip-cluster
-            topologyKey: "kubernetes.io/hostname"
-      containers:
-      - image: ghcr.io/kube-vip/kube-vip:0.3.7
-        imagePullPolicy: Always
-        name: kube-vip
-        command:
-        - /kube-vip
-        - service
-        - --configMap
-        - plndr-configmap
-        - --arp
-        - --interface
-        - ens192
-        - --log
-        - "5"
-        resources: {}
-        securityContext:
-          capabilities:
-            add:
-            - NET_ADMIN
-      hostNetwork: true
-status: {}
----
-kind: Role
-apiVersion: rbac.authorization.k8s.io/v1
-metadata:
-  name: lease-access
-rules:
-  - apiGroups: ["coordination.k8s.io"]
-    resources: ["leases"]
-    verbs: ["get", "create", "update", "list", "put"]
-  - apiGroups: [""]
-    resources: ["configMap"]
-    verbs: ["get"]
----
-kind: RoleBinding
-apiVersion: rbac.authorization.k8s.io/v1
-metadata:
-  name: lease-access
-subjects:
-  - kind: User
-    name: system:serviceaccount:default:default
-roleRef:
-  kind: Role
-  name: lease-access
-  apiGroup: rbac.authorization.k8s.io

example/deploy/0.1.3.yaml

@@ -1,83 +0,0 @@
-apiVersion: v1
-kind: ServiceAccount
-metadata:
-  name: vip
----
-kind: Role
-apiVersion: rbac.authorization.k8s.io/v1
-metadata:
-  name: vip-role
-rules:
-  - apiGroups: ["coordination.k8s.io"]
-    resources: ["leases"]
-    verbs: ["get", "create", "update", "list", "put"]
-  - apiGroups: [""]
-    resources: ["configmaps", "endpoints"]
-    verbs: ["watch", "get"]
----
-kind: RoleBinding
-apiVersion: rbac.authorization.k8s.io/v1
-metadata:
-  name: vip-role-bind
-subjects:
-  - kind: ServiceAccount
-    name: vip 
-    apiGroup: ""
-roleRef:
-  kind: Role
-  name: vip-role
-  apiGroup: "" 
----
-apiVersion: apps/v1
-kind: Deployment
-metadata:
-  creationTimestamp: null
-  labels:
-    app: kube-vip-cluster
-  name: kube-vip-cluster
-spec:
-  replicas: 3
-  selector:
-    matchLabels:
-      app: kube-vip-cluster
-  strategy: {}
-  template:
-    metadata:
-      creationTimestamp: null
-      labels:
-        app: kube-vip-cluster
-    spec:
-      affinity:
-        podAntiAffinity:
-          requiredDuringSchedulingIgnoredDuringExecution:
-          - labelSelector:
-              matchExpressions:
-                - key: "app"
-                  operator: In
-                  values:
-                  - kube-vip-cluster
-            topologyKey: "kubernetes.io/hostname"
-      containers:
-      - image: ghcr.io/kube-vip/kube-vip:0.3.7
-        imagePullPolicy: Always
-        name: kube-vip
-        command:
-        - /kube-vip
-        - service
-        env:
-          - name: vip_interface
-            value: "ens192"
-          - name: vip_configmap
-            value: "plndr" 
-          - name: vip_arp
-            value: "true"
-          - name: vip_loglevel
-            value: "5"
-        resources: {}
-        securityContext:
-          capabilities:
-            add:
-            - NET_ADMIN
-      hostNetwork: true
-      serviceAccountName: vip
-status: {}

example/deploy/0.1.4.yaml

@@ -1,83 +0,0 @@
-apiVersion: v1
-kind: ServiceAccount
-metadata:
-  name: vip
----
-kind: Role
-apiVersion: rbac.authorization.k8s.io/v1
-metadata:
-  name: vip-role
-rules:
-  - apiGroups: ["coordination.k8s.io"]
-    resources: ["leases"]
-    verbs: ["get", "create", "update", "list", "put"]
-  - apiGroups: [""]
-    resources: ["configmaps", "endpoints"]
-    verbs: ["watch", "get"]
----
-kind: RoleBinding
-apiVersion: rbac.authorization.k8s.io/v1
-metadata:
-  name: vip-role-bind
-subjects:
-  - kind: ServiceAccount
-    name: vip 
-    apiGroup: ""
-roleRef:
-  kind: Role
-  name: vip-role
-  apiGroup: "" 
----
-apiVersion: apps/v1
-kind: Deployment
-metadata:
-  creationTimestamp: null
-  labels:
-    app: kube-vip-cluster
-  name: kube-vip-cluster
-spec:
-  replicas: 3
-  selector:
-    matchLabels:
-      app: kube-vip-cluster
-  strategy: {}
-  template:
-    metadata:
-      creationTimestamp: null
-      labels:
-        app: kube-vip-cluster
-    spec:
-      affinity:
-        podAntiAffinity:
-          requiredDuringSchedulingIgnoredDuringExecution:
-          - labelSelector:
-              matchExpressions:
-                - key: "app"
-                  operator: In
-                  values:
-                  - kube-vip-cluster
-            topologyKey: "kubernetes.io/hostname"
-      containers:
-      - image: plndr/kube-vip:0.1.4
-        imagePullPolicy: Always
-        name: kube-vip
-        command:
-        - /kube-vip
-        - service
-        env:
-          - name: vip_interface
-            value: "ens192"
-          - name: vip_configmap
-            value: "plndr" 
-          - name: vip_arp
-            value: "true"
-          - name: vip_loglevel
-            value: "5"
-        resources: {}
-        securityContext:
-          capabilities:
-            add:
-            - NET_ADMIN
-      hostNetwork: true
-      serviceAccountName: vip
-status: {}

example/deploy/0.3.5.yaml

@@ -1,55 +0,0 @@
-apiVersion: apps/v1
-kind: DaemonSet
-metadata:
-  creationTimestamp: null
-  name: kube-vip-ds
-  namespace: kube-system
-spec:
-  selector:
-    matchLabels:
-      name: kube-vip-ds
-  template:
-    metadata:
-      creationTimestamp: null
-      labels:
-        name: kube-vip-ds
-    spec:
-      containers:
-      - args:
-        - manager
-        env:
-        - name: vip_arp
-          value: "true"
-        - name: vip_interface
-          value: eth0
-        - name: port
-          value: "6443"
-        - name: vip_cidr
-          value: "32"
-        - name: svc_enable
-          value: "true"
-        - name: vip_startleader
-          value: "false"
-        - name: vip_addpeerstolb
-          value: "true"
-        - name: vip_localpeer
-          value: ip-172-20-40-207:172.20.40.207:10000
-        - name: vip_address
-        image: plndr/kube-vip:v0.3.5
-        imagePullPolicy: Always
-        name: kube-vip
-        resources: {}
-        securityContext:
-          capabilities:
-            add:
-            - NET_ADMIN
-            - NET_RAW
-            - SYS_TIME
-      hostNetwork: true
-      serviceAccountName: kube-vip
-  updateStrategy: {}
-status:
-  currentNumberScheduled: 0
-  desiredNumberScheduled: 0
-  numberMisscheduled: 0
-  numberReady: 0

go.mod

@@ -1,133 +1,154 @@
 module github.com/kube-vip/kube-vip
 
-go 1.21
-
-toolchain go1.21.3
+go 1.26.4
 
 require (
-	github.com/cloudflare/ipvs v0.10.2
-	github.com/davecgh/go-spew v1.1.1
-	github.com/florianl/go-conntrack v0.4.0
-	github.com/golang/protobuf v1.5.4
-	github.com/google/go-cmp v0.6.0
+	github.com/cloudflare/ipvs v0.12.0
+	github.com/containernetworking/plugins v1.9.1
+	github.com/docker/docker v28.5.2+incompatible
+	github.com/florianl/go-conntrack v0.7.0
+	github.com/google/go-cmp v0.7.0
+	github.com/google/go-containerregistry v0.21.9
+	github.com/google/nftables v0.3.0
+	github.com/gookit/slog v0.7.1
 	github.com/huin/goupnp v1.3.0
-	github.com/insomniacslk/dhcp v0.0.0-20230731140434-0f9eb93a696c
+	github.com/insomniacslk/dhcp v0.0.0-20260719225207-c76316d4aa82
 	github.com/jpillora/backoff v1.0.0
-	github.com/mdlayher/ndp v1.0.1
-	github.com/onsi/ginkgo/v2 v2.19.0
-	github.com/onsi/gomega v1.33.1
-	github.com/osrg/gobgp/v3 v3.27.0
-	github.com/packethost/packngo v0.31.0
+	github.com/mdlayher/ndp v1.1.0
+	github.com/onsi/ginkgo/v2 v2.32.1
+	github.com/onsi/gomega v1.42.1
+	github.com/osrg/gobgp/v4 v4.8.0
 	github.com/pkg/errors v0.9.1
-	github.com/prometheus/client_golang v1.19.0
-	github.com/sirupsen/logrus v1.9.3
-	github.com/spf13/cobra v1.8.1
-	github.com/stretchr/testify v1.9.0
-	github.com/vishvananda/netlink v1.2.1-beta.2
-	go.etcd.io/etcd/api/v3 v3.5.13
-	go.etcd.io/etcd/client/pkg/v3 v3.5.13
-	go.etcd.io/etcd/client/v3 v3.5.13
-	go.uber.org/zap v1.27.0
-	golang.org/x/exp v0.0.0-20240409090435-93d18d7e34b8
-	golang.org/x/sync v0.7.0
-	golang.org/x/sys v0.20.0
-	golang.zx2c4.com/wireguard/wgctrl v0.0.0-20230429144221-925a1e7659e6
-	gopkg.in/yaml.v2 v2.4.0
-	k8s.io/api v0.29.1
-	k8s.io/apimachinery v0.29.3
-	k8s.io/client-go v0.29.1
-	k8s.io/klog/v2 v2.130.1
-	sigs.k8s.io/kind v0.22.0
-	sigs.k8s.io/yaml v1.4.0
+	github.com/prometheus/client_golang v1.24.1
+	github.com/spf13/cobra v1.10.2
+	github.com/stretchr/testify v1.11.1
+	github.com/vishvananda/netlink v1.3.1
+	go.etcd.io/etcd/api/v3 v3.7.1
+	go.etcd.io/etcd/client/pkg/v3 v3.7.1
+	go.etcd.io/etcd/client/v3 v3.7.1
+	go.uber.org/zap v1.28.0
+	golang.org/x/exp v0.0.0-20250103183323-7d7fa50e5329
+	golang.org/x/sync v0.22.0
+	golang.org/x/sys v0.47.0
+	golang.zx2c4.com/wireguard/wgctrl v0.0.0-20241231184526-a9ab2273dd10
+	google.golang.org/grpc v1.83.1
+	gopkg.in/yaml.v3 v3.0.1
+	k8s.io/api v0.36.4
+	k8s.io/apimachinery v0.36.4
+	k8s.io/client-go v0.36.4
+	k8s.io/klog/v2 v2.140.0
+	sigs.k8s.io/kind v0.32.0
+	sigs.k8s.io/yaml v1.6.0
 )
 
 require (
-	github.com/BurntSushi/toml v1.3.2 // indirect
-	github.com/alessio/shellescape v1.4.1 // indirect
+	al.essio.dev/pkg/shellescape v1.5.1 // indirect
+	github.com/BurntSushi/toml v1.5.0 // indirect
+	github.com/Masterminds/semver/v3 v3.4.0 // indirect
+	github.com/Microsoft/go-winio v0.6.2 // indirect
 	github.com/beorn7/perks v1.0.1 // indirect
-	github.com/cespare/xxhash/v2 v2.2.0 // indirect
-	github.com/coreos/go-semver v0.3.0 // indirect
-	github.com/coreos/go-systemd/v22 v22.5.0 // indirect
-	github.com/dgryski/go-farm v0.0.0-20200201041132-a6ae2369ad13 // indirect
+	github.com/cespare/xxhash/v2 v2.3.0 // indirect
+	github.com/containerd/errdefs v1.0.0 // indirect
+	github.com/containerd/errdefs/pkg v0.3.0 // indirect
+	github.com/containerd/log v0.1.0 // indirect
+	github.com/coreos/go-semver v0.3.1 // indirect
+	github.com/coreos/go-systemd/v22 v22.7.0 // indirect
+	github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc // indirect
+	github.com/dgryski/go-farm v0.0.0-20240924180020-3414d57e47da // indirect
+	github.com/distribution/reference v0.6.0 // indirect
+	github.com/docker/go-connections v0.7.0 // indirect
+	github.com/docker/go-units v0.5.0 // indirect
 	github.com/eapache/channels v1.1.0 // indirect
 	github.com/eapache/queue v1.1.0 // indirect
-	github.com/emicklei/go-restful/v3 v3.11.0 // indirect
-	github.com/evanphx/json-patch/v5 v5.6.0 // indirect
-	github.com/fsnotify/fsnotify v1.6.0 // indirect
-	github.com/go-logr/logr v1.4.1 // indirect
-	github.com/go-openapi/jsonpointer v0.19.6 // indirect
-	github.com/go-openapi/jsonreference v0.20.2 // indirect
-	github.com/go-openapi/swag v0.22.3 // indirect
+	github.com/emicklei/go-restful/v3 v3.13.0 // indirect
+	github.com/evanphx/json-patch/v5 v5.9.0 // indirect
+	github.com/felixge/httpsnoop v1.0.4 // indirect
+	github.com/fsnotify/fsnotify v1.9.0 // indirect
+	github.com/fxamacker/cbor/v2 v2.9.0 // indirect
+	github.com/gaissmai/bart v0.26.1 // indirect
+	github.com/go-logr/logr v1.4.3 // indirect
+	github.com/go-logr/stdr v1.2.2 // indirect
+	github.com/go-openapi/jsonpointer v0.21.0 // indirect
+	github.com/go-openapi/jsonreference v0.21.0 // indirect
+	github.com/go-openapi/swag v0.23.0 // indirect
 	github.com/go-task/slim-sprig/v3 v3.0.0 // indirect
-	github.com/gogo/protobuf v1.3.2 // indirect
-	github.com/google/gnostic-models v0.6.8 // indirect
-	github.com/google/gofuzz v1.2.0 // indirect
-	github.com/google/pprof v0.0.0-20240424215950-a892ee059fd6 // indirect
-	github.com/google/safetext v0.0.0-20220905092116-b49f7bc46da2 // indirect
-	github.com/google/uuid v1.3.1 // indirect
-	github.com/hashicorp/hcl v1.0.0 // indirect
-	github.com/imdario/mergo v0.3.12 // indirect
+	github.com/go-viper/mapstructure/v2 v2.4.0 // indirect
+	github.com/golang/protobuf v1.5.4 // indirect
+	github.com/google/gnostic-models v0.7.0 // indirect
+	github.com/google/pprof v0.0.0-20260402051712-545e8a4df936 // indirect
+	github.com/google/uuid v1.6.0 // indirect
+	github.com/gookit/color v1.6.1 // indirect
+	github.com/gookit/goutil v0.7.6 // indirect
+	github.com/gookit/gsr v0.1.1 // indirect
+	github.com/grpc-ecosystem/grpc-gateway/v2 v2.29.0 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
 	github.com/josharian/intern v1.0.0 // indirect
 	github.com/josharian/native v1.1.0 // indirect
 	github.com/json-iterator/go v1.1.12 // indirect
 	github.com/k-sone/critbitgo v1.4.0 // indirect
-	github.com/magiconair/properties v1.8.7 // indirect
-	github.com/mailru/easyjson v0.7.7 // indirect
-	github.com/mattn/go-isatty v0.0.14 // indirect
+	github.com/kylelemons/godebug v1.1.0 // indirect
+	github.com/mailru/easyjson v0.9.0 // indirect
+	github.com/mattn/go-isatty v0.0.20 // indirect
 	github.com/mdlayher/genetlink v1.3.2 // indirect
-	github.com/mdlayher/netlink v1.7.2 // indirect
+	github.com/mdlayher/netlink v1.8.0 // indirect
 	github.com/mdlayher/packet v1.1.2 // indirect
-	github.com/mdlayher/socket v0.4.1 // indirect
-	github.com/mitchellh/mapstructure v1.5.0 // indirect
+	github.com/mdlayher/socket v0.5.1 // indirect
+	github.com/moby/docker-image-spec v1.3.1 // indirect
+	github.com/moby/sys/atomicwriter v0.1.0 // indirect
+	github.com/moby/term v0.5.2 // indirect
 	github.com/modern-go/concurrent v0.0.0-20180306012644-bacd9c7ef1dd // indirect
-	github.com/modern-go/reflect2 v1.0.2 // indirect
+	github.com/modern-go/reflect2 v1.0.3-0.20250322232337-35a7c28c31ee // indirect
+	github.com/morikuni/aec v1.1.0 // indirect
 	github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 // indirect
-	github.com/pelletier/go-toml v1.9.4 // indirect
-	github.com/pelletier/go-toml/v2 v2.0.8 // indirect
-	github.com/pierrec/lz4/v4 v4.1.18 // indirect
-	github.com/pmezard/go-difflib v1.0.0 // indirect
-	github.com/prometheus/client_model v0.5.0 // indirect
-	github.com/prometheus/common v0.48.0 // indirect
-	github.com/prometheus/procfs v0.12.0 // indirect
-	github.com/remyoudompheng/bigfft v0.0.0-20200410134404-eec4a21b6bb0 // indirect
-	github.com/spf13/afero v1.9.5 // indirect
-	github.com/spf13/cast v1.5.1 // indirect
-	github.com/spf13/jwalterweatherman v1.1.0 // indirect
-	github.com/spf13/pflag v1.0.5 // indirect
-	github.com/spf13/viper v1.16.0 // indirect
-	github.com/subosito/gotenv v1.4.2 // indirect
-	github.com/tj/go-spin v1.1.0 // indirect
-	github.com/u-root/uio v0.0.0-20230305220412-3e8cd9d6bf63 // indirect
-	github.com/vishvananda/netns v0.0.4 // indirect
-	github.com/xlab/c-for-go v0.0.0-20230906092656-a1822f0a09c1 // indirect
-	github.com/xlab/pkgconfig v0.0.0-20170226114623-cea12a0fd245 // indirect
-	go.uber.org/multierr v1.10.0 // indirect
-	golang.org/x/crypto v0.23.0 // indirect
-	golang.org/x/mod v0.17.0 // indirect
-	golang.org/x/net v0.25.0 // indirect
-	golang.org/x/oauth2 v0.16.0 // indirect
-	golang.org/x/term v0.20.0 // indirect
-	golang.org/x/text v0.15.0 // indirect
-	golang.org/x/time v0.3.0 // indirect
-	golang.org/x/tools v0.21.0 // indirect
-	golang.zx2c4.com/wireguard v0.0.0-20230325221338-052af4a8072b // indirect
-	google.golang.org/appengine v1.6.7 // indirect
-	google.golang.org/genproto v0.0.0-20230822172742-b8732ec3820d // indirect
-	google.golang.org/genproto/googleapis/api v0.0.0-20230822172742-b8732ec3820d // indirect
-	google.golang.org/genproto/googleapis/rpc v0.0.0-20230822172742-b8732ec3820d // indirect
-	google.golang.org/grpc v1.59.0 // indirect
-	google.golang.org/protobuf v1.33.0 // indirect
+	github.com/opencontainers/go-digest v1.0.0 // indirect
+	github.com/opencontainers/image-spec v1.1.1 // indirect
+	github.com/orcaman/concurrent-map/v2 v2.0.1 // indirect
+	github.com/pelletier/go-toml v1.9.5 // indirect
+	github.com/pelletier/go-toml/v2 v2.2.3 // indirect
+	github.com/pierrec/lz4/v4 v4.1.22 // indirect
+	github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 // indirect
+	github.com/prometheus/client_model v0.6.2 // indirect
+	github.com/prometheus/common v0.70.1 // indirect
+	github.com/prometheus/procfs v0.21.1 // indirect
+	github.com/sagikazarmark/locafero v0.7.0 // indirect
+	github.com/segmentio/fasthash v1.0.3 // indirect
+	github.com/sourcegraph/conc v0.3.0 // indirect
+	github.com/spf13/afero v1.12.0 // indirect
+	github.com/spf13/cast v1.7.1 // indirect
+	github.com/spf13/pflag v1.0.10 // indirect
+	github.com/spf13/viper v1.20.1 // indirect
+	github.com/subosito/gotenv v1.6.0 // indirect
+	github.com/u-root/uio v0.0.0-20240224005618-d2acac8f3701 // indirect
+	github.com/valyala/bytebufferpool v1.0.0 // indirect
+	github.com/vishvananda/netns v0.0.5 // indirect
+	github.com/x448/float16 v0.8.4 // indirect
+	github.com/xo/terminfo v0.0.0-20220910002029-abceb7e1c41e // indirect
+	go.opentelemetry.io/auto/sdk v1.2.1 // indirect
+	go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.61.0 // indirect
+	go.opentelemetry.io/otel v1.44.0 // indirect
+	go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracehttp v1.43.0 // indirect
+	go.opentelemetry.io/otel/metric v1.44.0 // indirect
+	go.opentelemetry.io/otel/trace v1.44.0 // indirect
+	go.uber.org/multierr v1.11.0 // indirect
+	go.yaml.in/yaml/v2 v2.4.4 // indirect
+	go.yaml.in/yaml/v3 v3.0.4 // indirect
+	golang.org/x/crypto v0.54.0 // indirect
+	golang.org/x/mod v0.38.0 // indirect
+	golang.org/x/net v0.57.0 // indirect
+	golang.org/x/oauth2 v0.36.0 // indirect
+	golang.org/x/term v0.45.0 // indirect
+	golang.org/x/text v0.40.0 // indirect
+	golang.org/x/time v0.14.0 // indirect
+	golang.org/x/tools v0.48.0 // indirect
+	golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173 // indirect
+	google.golang.org/genproto/googleapis/api v0.0.0-20260526163538-3dc84a4a5aaa // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20260526163538-3dc84a4a5aaa // indirect
+	google.golang.org/protobuf v1.36.12-0.20260120151049-f2248ac996af // indirect
+	gopkg.in/evanphx/json-patch.v4 v4.13.0 // indirect
 	gopkg.in/inf.v0 v0.9.1 // indirect
-	gopkg.in/ini.v1 v1.67.0 // indirect
-	gopkg.in/yaml.v3 v3.0.1 // indirect
-	k8s.io/kube-openapi v0.0.0-20231010175941-2dd684a91f00 // indirect
-	k8s.io/utils v0.0.0-20230726121419-3b25d923346b // indirect
-	modernc.org/cc/v4 v4.1.0 // indirect
-	modernc.org/mathutil v1.5.0 // indirect
-	modernc.org/opt v0.1.3 // indirect
-	modernc.org/strutil v1.1.3 // indirect
-	modernc.org/token v1.0.1 // indirect
-	sigs.k8s.io/json v0.0.0-20221116044647-bc3834ca7abd // indirect
-	sigs.k8s.io/structured-merge-diff/v4 v4.4.1 // indirect
+	k8s.io/kube-openapi v0.0.0-20260317180543-43fb72c5454a // indirect
+	k8s.io/utils v0.0.0-20260210185600-b8788abfbbc2 // indirect
+	sigs.k8s.io/json v0.0.0-20250730193827-2d320260d730 // indirect
+	sigs.k8s.io/randfill v1.0.0 // indirect
+	sigs.k8s.io/structured-merge-diff/v6 v6.3.3 // indirect
 )

main.go

@@ -1,6 +1,8 @@
 package main
 
 import (
+	"os"
+
 	"github.com/kube-vip/kube-vip/cmd"
 )
 
@@ -14,5 +16,5 @@ func main() {
 
 	cmd.Release.Version = Version
 	cmd.Release.Build = Build
-	cmd.Execute()
+	os.Exit(cmd.Execute())
 }

pkg/arp/arp.go

@@ -0,0 +1,292 @@
+package arp
+
+import (
+	"context"
+	"fmt"
+	log "log/slog"
+	"sync"
+	"time"
+
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	"github.com/kube-vip/kube-vip/pkg/vip"
+	"github.com/vishvananda/netlink"
+)
+
+type Manager struct {
+	instances sync.Map
+	config    *kubevip.Config
+}
+
+type Instance struct {
+	network vip.Network
+	ndp     *vip.NdpResponder
+	mu      sync.Mutex
+	counter int
+}
+
+func NewManager(config *kubevip.Config) *Manager {
+	if config.ArpBroadcastRate < 500 {
+		log.Warn("[ARP manager] arp broadcast rate is too low", "rate (ms)", config.ArpBroadcastRate, "setting to (ms)", "3000")
+		config.ArpBroadcastRate = 3000
+	}
+	return &Manager{
+		config: config,
+	}
+}
+
+func NewInstance(network vip.Network, ndp *vip.NdpResponder) *Instance {
+	return &Instance{
+		ndp:     ndp,
+		network: network,
+		counter: 1,
+	}
+}
+
+func (i *Instance) Name() string {
+	return i.network.ARPName()
+}
+
+func (m *Manager) Insert(instance *Instance) {
+	i, err := m.get(instance.Name())
+	if err != nil {
+		log.Error("[ARP manager] unable to insert instance", "err", err)
+		return
+	}
+	if i == nil {
+		log.Info("[ARP manager] inserting ARP/NDP instance", "name", instance.Name())
+		m.instances.Store(instance.Name(), instance)
+	} else {
+		i.mu.Lock()
+		defer i.mu.Unlock()
+		i.counter++
+	}
+}
+
+func (m *Manager) Remove(instance *Instance) {
+	m.RemoveWithIPDelete(instance, true)
+}
+
+// RemoveOnLeadershipLoss removes an ARP instance when leadership is lost
+func (m *Manager) RemoveOnLeadershipLoss(instance *Instance) {
+	// Use the inverse of PreserveVIPOnLeadershipLoss to decide whether to delete the IP
+	// If preserve is true, don't delete IP (deleteIP = false)
+	// If preserve is false, delete IP (deleteIP = true), This is the legacy behavior
+	deleteIP := !m.config.PreserveVIPOnLeadershipLoss
+	m.RemoveWithIPDelete(instance, deleteIP)
+}
+
+func (m *Manager) RemoveWithIPDelete(instance *Instance, deleteIP bool) {
+	i, err := m.get(instance.Name())
+	if err != nil {
+		log.Error("[ARP manager] unable to remove the instance", "err", err)
+		return
+	}
+	if i != nil {
+		i.mu.Lock()
+		defer i.mu.Unlock()
+		i.counter--
+		if i.counter == 0 {
+			log.Info("[ARP manager] removing ARP/NDP instance", "name", instance.Name())
+			if deleteIP {
+				if _, err := instance.network.DeleteIP(); err != nil {
+					log.Error("failed to delete IP", "address", instance.network.IP(), "err", err)
+				}
+			}
+			m.instances.Delete(instance.Name())
+		}
+	} else {
+		log.Warn("[ARP manager] unable to remove the instance - instance not found", "name", instance.Name())
+	}
+}
+
+func (m *Manager) Count(name string) int {
+	i, err := m.get(name)
+	if err != nil {
+		log.Error("[ARP manager] unable to count instance", "err", err)
+		return -1
+	}
+	if i != nil {
+		i.mu.Lock()
+		defer i.mu.Unlock()
+		return i.counter
+	}
+	return 0
+}
+
+func (m *Manager) StartAdvertisement(ctx context.Context, killFunc func()) {
+	if m.config.LoseLeadership {
+		var wg sync.WaitGroup
+		defer wg.Wait()
+
+		log.Info("[ARP manager] starting watching network device", "interface", m.config.Interface)
+
+		duration := time.Duration(m.config.LoseLeadershipTimeoutSeconds) * time.Second
+		timeout := time.NewTimer(duration)
+		timeout.Stop()
+
+		wg.Go(func() {
+			select {
+			case <-timeout.C:
+				killFunc()
+			case <-ctx.Done():
+				return
+			}
+		})
+
+		wg.Go(func() {
+			if err := watch(ctx, m.config.Interface, func(s netlink.LinkOperState) {
+				if isUp(s) {
+					timeout.Stop()
+					return
+				}
+				timeout.Reset(duration)
+			}); err != nil {
+				log.Error("[ARP manager] stopped watching interface", "err", err)
+				killFunc()
+			}
+		})
+	}
+
+	log.Info("[ARP manager] starting ARP/NDP advertisement")
+
+	ticker := time.NewTicker(time.Duration(m.config.ArpBroadcastRate) * time.Millisecond)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-ctx.Done(): // if cancel() execute
+			return
+		case <-ticker.C: // send gratuitous ARP/NDP on each tick
+			m.instances.Range(func(_ any, instance any) bool {
+				if i, ok := instance.(*Instance); ok {
+					i.mu.Lock()
+					defer i.mu.Unlock()
+					if i.counter > 0 {
+						ensureIPAndSendGratuitous(i)
+					} else {
+						// this instance should not be advertised - delete the IP just in case...
+						if _, err := i.network.DeleteIP(); err != nil {
+							log.Error("[ARP manager] failed to delete IP", "address", i.network.IP(), "err", err)
+						}
+					}
+				}
+				return true
+			})
+		}
+	}
+}
+
+func (m *Manager) get(name string) (*Instance, error) {
+	i, exists := m.instances.Load(name)
+	if !exists {
+		return nil, nil
+	}
+	inst, ok := i.(*Instance)
+	if !ok {
+		return nil, fmt.Errorf("value for name %q is not of Instance pointer type", name)
+	}
+	return inst, nil
+}
+
+// ensureIPAndSendGratuitous - adds IP to the interface if missing, and send
+// either a gratuitous ARP or gratuitous NDP. Re-adds the interface if it is IPv6
+// and in a dadfailed state.
+func ensureIPAndSendGratuitous(instance *Instance) {
+	iface := instance.network.Interface()
+	ipString := instance.network.IP()
+
+	// Check if IP is dadfailed
+	if instance.network.IsDADFAIL() {
+		log.Warn("IP address is in dadfailed state, removing config", "ip", ipString, "interface", iface)
+		deleted, err := instance.network.DeleteIP()
+		if err != nil {
+			log.Warn(err.Error())
+		}
+		if deleted {
+			log.Info("deleted and recreating address with NODAD flag to skip DAD", "IP", ipString, "interface", iface)
+			// Re-add immediately without DAD check since we're recovering from DADFAILED
+			// The AddIP function will set IFA_F_NODAD flag for IPv6 addresses when skipDAD=true
+			if _, err := instance.network.AddIP(false, true); err != nil {
+				log.Error("failed to recreate address after DADFAILED", "IP", ipString, "interface", iface, "err", err)
+			} else {
+				log.Info("successfully recreated address after DADFAILED recovery", "IP", ipString, "interface", iface)
+			}
+		}
+		// Return early after DADFAILED recovery to avoid double IP addition
+		return
+	}
+
+	// Normal case: add IP with precheck and normal DAD process
+	if added, err := instance.network.AddIP(true, false); err != nil {
+		log.Warn(err.Error())
+	} else if added {
+		log.Warn("Re-applied the VIP configuration", "ip", ipString, "interface", iface)
+	}
+
+	if utils.IsIPv6(ipString) {
+		// Gratuitous NDP, will broadcast new MAC <-> IPv6 address
+		if instance.ndp == nil {
+			log.Error("NDP responder was not created")
+		} else {
+			err := instance.ndp.SendGratuitous(ipString)
+			if err != nil {
+				log.Warn(err.Error())
+			}
+		}
+	} else {
+		// Gratuitous ARP, will broadcast to new MAC <-> IPv4 address
+		err := vip.ARPSendGratuitous(ipString, iface)
+		if err != nil {
+			log.Warn(err.Error())
+		}
+	}
+}
+
+// watch subscribing to the network interface events and calls handler
+func watch(ctx context.Context, interfaceName string, operStateHandler func(netlink.LinkOperState)) error {
+	ifname, err := netlink.LinkByName(interfaceName)
+	if err != nil {
+		return fmt.Errorf("failed to watch interface %q: %w", interfaceName, err)
+	}
+
+	// verify if this interface is physical device
+	if _, ok := ifname.(*netlink.Device); !ok {
+		return fmt.Errorf("interface %s is not physical, ignoring", interfaceName)
+	}
+
+	events := make(chan netlink.LinkUpdate)
+	done := make(chan struct{})
+
+	if err := netlink.LinkSubscribe(events, done); err != nil {
+		return fmt.Errorf("failed to subscribe to the interface events: %w", err)
+	}
+	defer close(done)
+
+	//  handle initial state
+	operStateHandler(ifname.Attrs().OperState)
+
+	for {
+		select {
+		case <-ctx.Done():
+			return ctx.Err()
+		case event, ok := <-events:
+			if !ok {
+				return fmt.Errorf("interface events channel closed")
+			}
+
+			attrs := event.Attrs()
+			// LinkSubscribe captures events for all network devices found
+			// so we only care about vip interface
+			if ifname.Attrs().Name != attrs.Name {
+				continue
+			}
+			log.Debug("handling device change", "state", attrs.OperState)
+			operStateHandler(attrs.OperState)
+		}
+	}
+}
+
+func isUp(operState netlink.LinkOperState) bool {
+	return operState == netlink.OperUp
+}

pkg/backend/backend.go

@@ -1,24 +1,42 @@
 package backend
 
 import (
+	"context"
 	"fmt"
+	"sync"
 	"time"
 
+	log "log/slog"
+
 	"github.com/kube-vip/kube-vip/pkg/k8s"
 	"github.com/kube-vip/kube-vip/pkg/utils"
-	"github.com/kube-vip/kube-vip/pkg/vip"
-	log "github.com/sirupsen/logrus"
 	"k8s.io/client-go/kubernetes"
 	"k8s.io/client-go/rest"
 )
 
 type Entry struct {
-	Addr string
-	Port int
+	Addr    string
+	Port    uint16
+	IsLocal bool
 }
 
 type Map map[Entry]bool
 
+// kubeConfigPath is an explicitly configured kubeconfig used by Check when
+// set; static pod deployments configure it since neither admin.conf nor
+// in-cluster config are available there.
+var (
+	kubeConfigPath string
+	pathMtx        sync.Mutex
+)
+
+// SetKubeConfigPath configures the kubeconfig used by backend health checks.
+func SetKubeConfigPath(path string) {
+	pathMtx.Lock()
+	defer pathMtx.Unlock()
+	kubeConfigPath = path
+}
+
 func (e *Entry) Check() bool {
 	var client *kubernetes.Clientset
 	var err error
@@ -29,43 +47,48 @@ func (e *Entry) Check() bool {
 	// homeConfigPath := filepath.Join(os.Getenv("HOME"), ".kube", "config")
 
 	var k8sAddr string
-	if vip.IsIPv4(e.Addr) {
-		k8sAddr = fmt.Sprintf("%s:%v", e.Addr, e.Port)
-	} else {
+	if utils.IsIPv6(e.Addr) {
 		k8sAddr = fmt.Sprintf("[%s]:%v", e.Addr, e.Port)
+	} else {
+		k8sAddr = fmt.Sprintf("%s:%v", e.Addr, e.Port)
 	}
 
 	switch {
+	case kubeConfigPath != "" && utils.FileExists(kubeConfigPath):
+		config, err = k8s.NewRestConfig(kubeConfigPath, false, k8sAddr)
+		if err != nil {
+			log.Error("create k8s REST config", "path", kubeConfigPath, "err", err)
+			return false
+		}
 	case utils.FileExists(adminConfigPath):
 		config, err = k8s.NewRestConfig(adminConfigPath, false, k8sAddr)
-		// client, err = k8s.NewClientset(adminConfigPath, false, k8sAddr)
 		if err != nil {
-			log.Errorf("could not create k8s REST config for external file: %q: %v", adminConfigPath, err)
+			log.Error("create k8s REST config", "path", adminConfigPath, "err", err)
 			return false
 		}
 	default:
 		config, err = k8s.NewRestConfig("", true, k8sAddr)
 		if err != nil {
-			log.Errorf("could not create k8s REST config %v", err)
+			log.Error("create k8s REST config", "err", err)
 			return false
 		}
 	}
 
 	client, err = k8s.NewClientset(config)
 	if err != nil {
-		log.Errorf("failed to create k8s client: %v", err)
+		log.Error("create k8s client", "err", err)
 		return false
 	}
 
 	_, err = client.DiscoveryClient.ServerVersion()
 	if err != nil {
-		log.Errorf("failed check k8s server version: %s", err)
+		log.Error("discover k8s version", "err", err)
 		return false
 	}
 	return true
 }
 
-func Watch(tickAction func(), interval int, stop chan struct{}) {
+func Watch(ctx context.Context, interval int, tickAction func()) {
 	if interval <= 0 {
 		interval = 5
 	}
@@ -75,13 +98,10 @@ func Watch(tickAction func(), interval int, stop chan struct{}) {
 
 	for {
 		select {
-		case <-stop:
-			ticker.Stop()
+		case <-ctx.Done():
 			return
 		case <-ticker.C:
-			ticker.Stop()
 			tickAction()
-			ticker.Reset(time.Second * time.Duration(interval))
 		}
 	}
 }

pkg/bgp/hosts.go

@@ -3,46 +3,80 @@ package bgp
 import (
 	"context"
 	"fmt"
+	log "log/slog"
 	"net"
 
-	api "github.com/osrg/gobgp/v3/api"
+	"github.com/osrg/gobgp/v4/pkg/apiutil"
 )
 
 // AddHost will update peers of a host
-func (b *Server) AddHost(addr string) (err error) {
-	ip, _, err := net.ParseCIDR(addr)
-	if err != nil {
-		return err
+func (b *Server) AddHost(ctx context.Context, addr string, object string) error {
+	b.mtx.Lock()
+	defer b.mtx.Unlock()
+
+	objects, exists := b.tracker[addr]
+
+	if !exists {
+		b.tracker[addr] = make(map[string]bool)
+		objects = b.tracker[addr]
+
+		ip, _, err := net.ParseCIDR(addr)
+		if err != nil {
+			return err
+		}
+
+		p := b.getPath(ip)
+		if p == nil {
+			return fmt.Errorf("failed to get path for %v", ip)
+		}
+
+		if _, err := b.s.AddPath(apiutil.AddPathRequest{
+			Paths: []*apiutil.Path{p},
+		}); err != nil {
+			return err
+		}
+		log.Debug("[BGP] added host", "addr", addr, "cnt", len(objects)+1, "object", object)
 	}
 
-	p := b.getPath(ip)
-	if p == nil {
-		return fmt.Errorf("failed to get path for %v", ip)
-	}
+	objects[object] = true
 
-	_, err = b.s.AddPath(context.Background(), &api.AddPathRequest{
-		Path: p,
-	})
-
-	if err != nil {
-		return err
-	}
-
-	return
+	return nil
 }
 
 // DelHost will inform peers to remove a host
-func (b *Server) DelHost(addr string) (err error) {
+func (b *Server) DelHost(ctx context.Context, addr string, object string) error {
+	b.mtx.Lock()
+	defer b.mtx.Unlock()
+
+	objects, exists := b.tracker[addr]
+	if !exists {
+		log.Debug("[BGP] deleting host - nothing to delete", "addr", addr, "object", object)
+		return nil
+	}
+
 	ip, _, err := net.ParseCIDR(addr)
 	if err != nil {
 		return err
 	}
-	p := b.getPath(ip)
-	if p == nil {
-		return
+
+	delete(objects, object)
+
+	if len(objects) == 0 {
+		p := b.getPath(ip)
+		if p == nil {
+			return nil
+		}
+
+		if err := b.s.DeletePath(apiutil.DeletePathRequest{
+			Paths: []*apiutil.Path{p},
+		}); err != nil {
+			return err
+		}
+		delete(b.tracker, addr)
+		log.Debug("[BGP] deleted host", "addr", addr, "cnt", len(objects), "object", object)
+	} else {
+		log.Debug("[BGP] deleting from tracker only", "addr", addr, "object", object)
 	}
 
-	return b.s.DeletePath(context.Background(), &api.DeletePathRequest{
-		Path: p,
-	})
+	return nil
 }

pkg/bgp/peers.go

@@ -3,22 +3,34 @@ package bgp
 import (
 	"context"
 	"fmt"
+	log "log/slog"
 	"net"
+	"net/netip"
 	"strconv"
-	"strings"
+	"time"
 
-	"github.com/golang/protobuf/ptypes" //nolint
-	"github.com/golang/protobuf/ptypes/any"
-	api "github.com/osrg/gobgp/v3/api"
+	"github.com/jpillora/backoff"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/vip"
+	api "github.com/osrg/gobgp/v4/api"
+
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	"github.com/osrg/gobgp/v4/pkg/apiutil"
+	"github.com/osrg/gobgp/v4/pkg/config/oc"
+	bgp "github.com/osrg/gobgp/v4/pkg/packet/bgp"
+	"github.com/osrg/gobgp/v4/pkg/server"
 )
 
+const defaultBGPPort uint32 = 179
+
 // AddPeer will add peers to the BGP configuration
-func (b *Server) AddPeer(peer Peer) (err error) {
+func (b *Server) AddPeer(ctx context.Context, peer kubevip.BGPPeer) (err error) {
 	p := &api.Peer{
 		Conf: &api.PeerConf{
-			NeighborAddress: peer.Address,
-			PeerAsn:         peer.AS,
-			AuthPassword:    peer.Password,
+			NeighborAddress:   peer.Address,
+			PeerAsn:           peer.AS,
+			NeighborInterface: peer.Interface,
+			AuthPassword:      peer.Password,
 		},
 
 		Timers: &api.Timers{
@@ -38,138 +50,268 @@ func (b *Server) AddPeer(peer Peer) (err error) {
 		Transport: &api.Transport{
 			MtuDiscovery:  true,
 			RemoteAddress: peer.Address,
-			RemotePort:    uint32(179),
+			RemotePort:    defaultBGPPort,
 		},
 	}
 
-	if b.c.SourceIP != "" {
-		p.Transport.LocalAddress = b.c.SourceIP
+	if peer.BFDEnabled {
+		p.Bfd = &api.BfdPeerConfig{
+			Enabled:                  true,
+			DesiredMinimumTxInterval: peer.BFDTransmitInterval,
+			RequiredMinimumReceive:   peer.BFDReceiveInterval,
+			DetectionMultiplier:      peer.BFDDetectMultiplier,
+			Port:                     3784, // TODO: Should this be configurable??
+		}
 	}
 
-	if b.c.SourceIF != "" {
-		p.Transport.BindInterface = b.c.SourceIF
+	if peer.Interface != "" {
+		neighborAddress, err := getIPv6LinkLocalNeighborAddress(ctx, peer.Interface)
+		if err != nil {
+			return fmt.Errorf("failed to get link-local address of interface %s: %w", peer.Interface, err)
+		}
+
+		p.State = &api.PeerState{
+			NeighborAddress: neighborAddress,
+		}
 	}
 
-	return b.s.AddPeer(context.Background(), &api.AddPeerRequest{
-		Peer: p,
-	})
-}
+	mpBGP := b.c.MpbgpNexthop
 
-func (b *Server) getPath(ip net.IP) (path *api.Path) {
-	isV6 := ip.To4() == nil
+	if peer.MpbgpNexthop != "" {
+		mpBGP = peer.MpbgpNexthop
+	}
 
-	//nolint
-	originAttr, _ := ptypes.MarshalAny(&api.OriginAttribute{
-		Origin: 0,
-	})
+	if mpBGP != "" {
+		ipv4Address, ipv6Address, err := peer.FindMpbgpAddresses(p, b.c)
+		if err != nil {
+			log.Error("failed to get MP-BGP addresses, will not us MP-BGP for this host", "error", err)
+		} else {
+			p.AfiSafis = []*api.AfiSafi{
+				{
+					Config: &api.AfiSafiConfig{
+						Family: &api.Family{
+							Afi:  api.Family_AFI_IP,
+							Safi: api.Family_SAFI_UNICAST,
+						},
+						Enabled: true,
+					},
+				},
+				{
+					Config: &api.AfiSafiConfig{
+						Family: &api.Family{
+							Afi:  api.Family_AFI_IP6,
+							Safi: api.Family_SAFI_UNICAST,
+						},
+						Enabled: true,
+					},
+				},
+			}
 
-	if !isV6 {
-		//nolint
-		nlri, _ := ptypes.MarshalAny(&api.IPAddressPrefix{
-			Prefix:    ip.String(),
-			PrefixLen: 32,
-		})
+			peer.SetMpbgpOptions(b.c)
 
-		//nolint
-		nhAttr, _ := ptypes.MarshalAny(&api.NextHopAttribute{
-			NextHop: "0.0.0.0", // gobgp will fill this
-		})
+			mask := strconv.Itoa(vip.DefaultMaskIPv6)
+			address := ipv4Address
+			family := api.Family_AFI_IP
+			if utils.IsIPv4(p.Conf.NeighborAddress) {
+				mask = strconv.Itoa(vip.DefaultMaskIPv4)
+				address = ipv6Address
+				family = api.Family_AFI_IP6
+			}
 
-		path = &api.Path{
-			Family: &api.Family{
-				Afi:  api.Family_AFI_IP,
-				Safi: api.Family_SAFI_UNICAST,
-			},
-			Nlri:   nlri,
-			Pattrs: []*any.Any{originAttr, nhAttr},
+			err = b.s.AddDefinedSet(ctx, &api.AddDefinedSetRequest{
+				DefinedSet: &api.DefinedSet{
+					DefinedType: api.DefinedType_DEFINED_TYPE_NEIGHBOR,
+					Name:        fmt.Sprintf("peer-%s", p.Conf.NeighborAddress),
+					List:        []string{fmt.Sprintf("%s/%s", p.Conf.NeighborAddress, mask)},
+				},
+			})
+			if err != nil {
+				return fmt.Errorf("failed to add defined set: %v", err)
+			}
+
+			if address != "" {
+				if err := insertPolicy(ctx, b.s, address, p, family); err != nil {
+					return fmt.Errorf("failed to add policy: %w", err)
+				}
+			}
 		}
 	} else {
-		//nolint
-		nlri, _ := ptypes.MarshalAny(&api.IPAddressPrefix{
-			Prefix:    ip.String(),
-			PrefixLen: 128,
-		})
-
-		v6Family := &api.Family{
-			Afi:  api.Family_AFI_IP6,
-			Safi: api.Family_SAFI_UNICAST,
+		if b.c.SourceIP != "" {
+			p.Transport.LocalAddress = b.c.SourceIP
 		}
 
-		//nolint
-		mpAttr, _ := ptypes.MarshalAny(&api.MpReachNLRIAttribute{
-			Family:   v6Family,
-			NextHops: []string{"::"}, // gobgp will fill this
-			Nlris:    []*any.Any{nlri},
-		})
-
-		path = &api.Path{
-			Family: v6Family,
-			Nlri:   nlri,
-			Pattrs: []*any.Any{originAttr, mpAttr},
+		if b.c.SourceIF != "" {
+			p.Transport.BindInterface = b.c.SourceIF
 		}
 	}
-	return
+
+	if err := b.s.AddPeer(ctx, &api.AddPeerRequest{Peer: p}); err != nil {
+		return fmt.Errorf("failed to add peer: %v", err)
+	}
+	log.Info("[BGP]", "peer", p.Conf.NeighborAddress, "AS", p.Conf.PeerAsn, "BFD", p.Bfd)
+	return nil
 }
 
-// ParseBGPPeerConfig - take a string and parses it into an array of peers
-func ParseBGPPeerConfig(config string) (bgpPeers []Peer, err error) {
-	peers := strings.Split(config, ",")
-	if len(peers) == 0 {
-		return nil, fmt.Errorf("No BGP Peer configurations found")
-	}
+func (b *Server) getPath(ip net.IP) *apiutil.Path {
+	isV6 := ip.To4() == nil
 
-	for x := range peers {
-		peerStr := peers[x]
-		if peerStr == "" {
-			continue
-		}
-		isV6Peer := peerStr[0] == '['
-
-		address := ""
-		if isV6Peer {
-			addressEndPos := strings.IndexByte(peerStr, ']')
-			if addressEndPos == -1 {
-				return nil, fmt.Errorf("no matching ] found for IPv6 BGP Peer")
-			}
-			address = peerStr[1:addressEndPos]
-			peerStr = peerStr[addressEndPos+1:]
-		}
-
-		peer := strings.Split(peerStr, ":")
-		if len(peer) < 2 {
-			return nil, fmt.Errorf("mandatory peering params <host>:<AS> incomplete")
-		}
-
-		if !isV6Peer {
-			address = peer[0]
-		}
-
-		ASNumber, err := strconv.ParseUint(peer[1], 10, 32)
+	if !isV6 {
+		prefix, err := bgp.NewIPAddrPrefix(netip.MustParsePrefix(
+			fmt.Sprintf("%s/%d", ip.String(), vip.DefaultMaskIPv4),
+		))
 		if err != nil {
-			return nil, fmt.Errorf("BGP Peer AS format error [%s]", peer[1])
+			return nil
 		}
 
-		password := ""
-		if len(peer) >= 3 {
-			password = peer[2]
+		nh, err := bgp.NewPathAttributeNextHop(netip.MustParseAddr("0.0.0.0"))
+		if err != nil {
+			return nil
 		}
 
-		multiHop := false
-		if len(peer) >= 4 {
-			multiHop, err = strconv.ParseBool(peer[3])
+		return &apiutil.Path{
+			Family: bgp.RF_IPv4_UC,
+			Nlri:   prefix,
+			Attrs: []bgp.PathAttributeInterface{
+				bgp.NewPathAttributeOrigin(0),
+				nh,
+			},
+		}
+	}
+
+	prefix, err := bgp.NewIPAddrPrefix(netip.MustParsePrefix(
+		fmt.Sprintf("%s/%d", ip.String(), vip.DefaultMaskIPv6),
+	))
+	if err != nil {
+		return nil
+	}
+
+	mpReach, err := bgp.NewPathAttributeMpReachNLRI(
+		bgp.RF_IPv6_UC,
+		[]bgp.PathNLRI{{NLRI: prefix}},
+		netip.MustParseAddr("::"),
+	)
+	if err != nil {
+		return nil
+	}
+
+	return &apiutil.Path{
+		Family: bgp.RF_IPv6_UC,
+		Nlri:   prefix,
+		Attrs: []bgp.PathAttributeInterface{
+			bgp.NewPathAttributeOrigin(0),
+			mpReach,
+		},
+	}
+}
+func insertPolicy(ctx context.Context, s *server.BgpServer, address string, p *api.Peer, family api.Family_Afi) error {
+	familyType := "v4"
+	if family == api.Family_AFI_IP6 {
+		familyType = "v6"
+	}
+
+	setName := fmt.Sprintf("peer-%s", p.Conf.NeighborAddress)
+	policyName := fmt.Sprintf("%s-%s", setName, familyType)
+
+	policy := &api.Policy{
+		Name: policyName,
+		Statements: []*api.Statement{
+			{
+				Conditions: &api.Conditions{
+					AfiSafiIn: []*api.Family{
+						{
+							Afi:  family,
+							Safi: api.Family_SAFI_UNICAST,
+						},
+					},
+					NeighborSet: &api.MatchSet{
+						Type: api.MatchSet_TYPE_ANY,
+						Name: setName,
+					},
+				},
+				Actions: &api.Actions{
+					RouteAction: api.RouteAction_ROUTE_ACTION_ACCEPT,
+					Nexthop: &api.NexthopAction{
+						Address: address,
+					},
+				},
+			},
+			{
+				Conditions: &api.Conditions{
+					NeighborSet: &api.MatchSet{
+						Type: api.MatchSet_TYPE_ANY,
+						Name: setName,
+					},
+				},
+				Actions: &api.Actions{
+					RouteAction: api.RouteAction_ROUTE_ACTION_ACCEPT,
+				},
+			},
+		},
+	}
+
+	err := s.AddPolicy(ctx, &api.AddPolicyRequest{
+		Policy: policy,
+	})
+	if err != nil {
+		return fmt.Errorf("failed to add policy: %w", err)
+	}
+
+	err = s.AddPolicyAssignment(ctx, &api.AddPolicyAssignmentRequest{
+		Assignment: &api.PolicyAssignment{
+			Name:      "global",
+			Direction: api.PolicyDirection_POLICY_DIRECTION_EXPORT,
+			Policies: []*api.Policy{
+				{
+					Name: policy.Name,
+				},
+			},
+		},
+	})
+	if err != nil {
+		return fmt.Errorf("failed to add policy assignment: %v", err)
+	}
+
+	return nil
+}
+
+func getIPv6LinkLocalNeighborAddress(ctx context.Context, peerInterface string) (string, error) {
+	neighCtx, neighCancel := context.WithTimeout(ctx, time.Minute)
+	defer neighCancel()
+
+	bo := backoff.Backoff{
+		Factor: 2,
+		Jitter: true,
+		Min:    1 * time.Second,
+		Max:    5 * time.Second,
+	}
+
+	maxAttempts := 20.0
+
+	var err error
+	for {
+		select {
+		case <-neighCtx.Done():
 			if err != nil {
-				return nil, fmt.Errorf("BGP MultiHop format error (true/false) [%s]", peer[1])
+				return "", fmt.Errorf("failed to get link-local address of interface %s: %w", peerInterface, err)
+			}
+			return "", fmt.Errorf("failed to get link-local address of interface %s: %w", peerInterface, neighCtx.Err())
+		default:
+			dur := bo.Duration()
+			var neighborAddress string
+			neighborAddress, err = oc.GetIPv6LinkLocalNeighborAddress(peerInterface)
+			if err != nil && bo.Attempt() >= maxAttempts {
+				return "", fmt.Errorf("failed to get link-local address of interface %s: %w", peerInterface, err)
+			}
+			if neighborAddress != "" {
+				return neighborAddress, nil
+			}
+			t := time.NewTimer(dur)
+			select {
+			case <-neighCtx.Done():
+				t.Stop()
+			case <-t.C:
 			}
 		}
 
-		peerConfig := Peer{
-			Address:  address,
-			AS:       uint32(ASNumber),
-			Password: password,
-			MultiHop: multiHop,
-		}
-
-		bgpPeers = append(bgpPeers, peerConfig)
 	}
-	return
 }

pkg/bgp/server.go

@@ -3,56 +3,89 @@ package bgp
 import (
 	"context"
 	"fmt"
+	"sync"
 	"time"
 
-	api "github.com/osrg/gobgp/v3/api"
-	gobgp "github.com/osrg/gobgp/v3/pkg/server"
-	log "github.com/sirupsen/logrus"
+	log "log/slog"
+
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	api "github.com/osrg/gobgp/v4/api"
+	"github.com/osrg/gobgp/v4/pkg/apiutil"
+	bgp "github.com/osrg/gobgp/v4/pkg/packet/bgp"
+	gobgp "github.com/osrg/gobgp/v4/pkg/server"
 )
 
+// Server manages a server object
+type Server struct {
+	s       *gobgp.BgpServer
+	c       *kubevip.BGPConfig
+	mtx     sync.Mutex
+	tracker map[string]map[string]bool
+}
+
 // NewBGPServer takes a configuration and returns a running BGP server instance
-func NewBGPServer(c *Config, peerStateChangeCallback func(*api.WatchEventResponse_PeerEvent)) (b *Server, err error) {
+func NewBGPServer(c kubevip.BGPConfig, logLevel log.Level) (b *Server, err error) {
 	if c.AS == 0 {
-		return nil, fmt.Errorf("You need to provide AS")
+		return nil, fmt.Errorf("you need to provide AS")
 	}
 
 	if c.SourceIP != "" && c.SourceIF != "" {
-		return nil, fmt.Errorf("SourceIP and SourceIF are mutually exclusive")
+		return nil, fmt.Errorf("sourceIP and SourceIF are mutually exclusive")
 	}
 
 	if len(c.Peers) == 0 {
-		return nil, fmt.Errorf("You need to provide at least one peer")
+		return nil, fmt.Errorf("you need to provide at least one peer")
 	}
+	bgpLogger := log.Default()
+	lvl := &log.LevelVar{}
+	lvl.Set(logLevel)
 
 	b = &Server{
-		s: gobgp.NewBgpServer(),
-		c: c,
+		s:       gobgp.NewBgpServer(gobgp.LoggerOption(bgpLogger, lvl)),
+		c:       &c,
+		tracker: make(map[string]map[string]bool),
 	}
+	return
+}
+
+// Start starts the BGP server
+func (b *Server) Start(ctx context.Context, peerStateChangeCallback func(*apiutil.WatchEventMessage_PeerEvent)) (err error) {
 	go b.s.Serve()
 
-	if err = b.s.StartBgp(context.Background(), &api.StartBgpRequest{
+	if err = b.s.StartBgp(ctx, &api.StartBgpRequest{
 		Global: &api.Global{
-			Asn:        c.AS,
-			RouterId:   c.RouterID,
+			Asn:        b.c.AS,
+			RouterId:   b.c.RouterID,
 			ListenPort: -1,
 		},
 	}); err != nil {
 		return
 	}
 
-	if err = b.s.WatchEvent(context.Background(), &api.WatchEventRequest{Peer: &api.WatchEventRequest_Peer{}}, func(r *api.WatchEventResponse) {
-		if p := r.GetPeer(); p != nil && p.Type == api.WatchEventResponse_PeerEvent_STATE {
-			log.Infof("[BGP] %s", p.String())
+	if err = b.s.WatchEvent(ctx, gobgp.WatchEventMessageCallbacks{
+		OnPeerUpdate: func(p *apiutil.WatchEventMessage_PeerEvent, _ time.Time) {
+			log.Info("[BGP]", "peer", fmt.Sprintf("%+v", p))
 			if peerStateChangeCallback != nil {
 				peerStateChangeCallback(p)
 			}
-		}
-	}); err != nil {
+		},
+	}, gobgp.WatchPeer()); err != nil {
 		return
 	}
 
-	for _, p := range c.Peers {
-		if err = b.AddPeer(p); err != nil {
+	for _, p := range b.c.Peers {
+		if err = b.AddPeer(ctx, p); err != nil {
+			return
+		}
+	}
+
+	if b.c.Zebra.Enabled {
+		if err = b.s.EnableZebra(ctx, &api.EnableZebraRequest{
+			Url:          b.c.Zebra.URL,
+			Version:      b.c.Zebra.Version,
+			SoftwareName: b.c.Zebra.SoftwareName,
+		}); err != nil {
+			log.Error(err.Error())
 			return
 		}
 	}
@@ -62,7 +95,51 @@ func NewBGPServer(c *Config, peerStateChangeCallback func(*api.WatchEventRespons
 
 // Close will stop a running BGP Server
 func (b *Server) Close() error {
-	ctx, cf := context.WithTimeout(context.Background(), 5*time.Second)
-	defer cf()
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
 	return b.s.StopBgp(ctx, &api.StopBgpRequest{})
 }
+
+// ListAdvertisedRoutes retrieves all active routes inside GoBGP's local RIB.
+// It queries the GLOBAL table type to find routes that kube-vip has requested GoBGP to advertise.
+func (b *Server) ListAdvertisedRoutes(ctx context.Context, isIPv6 bool) ([]*api.Destination, error) {
+	afi := bgp.AFI_IP
+
+	if isIPv6 {
+		afi = bgp.AFI_IP6
+	}
+
+	family := bgp.NewFamily(uint16(afi), bgp.SAFI_UNICAST)
+
+	var destinations []*api.Destination
+
+	req := apiutil.ListPathRequest{
+		TableType: api.TableType_TABLE_TYPE_GLOBAL,
+		Family:    family,
+	}
+
+	// GoBGP's embedded server API uses a callback function to stream results
+	// locally without requiring a gRPC client stream setup.
+	err := b.s.ListPath(req, func(prefix bgp.NLRI, paths []*apiutil.Path) {
+		var newPaths []*api.Path
+		for _, p := range paths {
+			np, err := apiutil.NewPath(p.Family, p.Nlri, p.Withdrawal, p.Attrs, time.Unix(p.Age, 0))
+			if err != nil {
+				log.Error("failed to create BGP path details", "err", err)
+				continue
+			}
+			newPaths = append(newPaths, np)
+		}
+		d := &api.Destination{
+			Prefix: prefix.String(),
+			Paths:  newPaths,
+		}
+		destinations = append(destinations, d)
+	})
+
+	if err != nil {
+		return nil, fmt.Errorf("failed to extract local RIB: %w", err)
+	}
+
+	return destinations, nil
+}

pkg/bgp/types.go

@@ -1,30 +0,0 @@
-package bgp
-
-import gobgp "github.com/osrg/gobgp/v3/pkg/server"
-
-// Peer defines a BGP Peer
-type Peer struct {
-	Address  string
-	AS       uint32
-	Password string
-	MultiHop bool
-}
-
-// Config defines the BGP server configuration
-type Config struct {
-	AS       uint32
-	RouterID string
-	SourceIP string
-	SourceIF string
-
-	HoldTime          uint64
-	KeepaliveInterval uint64
-
-	Peers []Peer
-}
-
-// Server manages a server object
-type Server struct {
-	s *gobgp.BgpServer
-	c *Config
-}

pkg/cluster/cluster.go

@@ -1,45 +1,73 @@
 package cluster
 
 import (
-	"sync"
+	"crypto/tls"
+	"crypto/x509"
+	"fmt"
+	"net/http"
+	"os"
+	"time"
 
-	log "github.com/sirupsen/logrus"
+	log "log/slog"
 
+	"github.com/kube-vip/kube-vip/pkg/arp"
 	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/networkinterface"
+	"github.com/kube-vip/kube-vip/pkg/node"
+	"github.com/kube-vip/kube-vip/pkg/route"
+	"github.com/kube-vip/kube-vip/pkg/utils"
 	"github.com/kube-vip/kube-vip/pkg/vip"
 )
 
 // Cluster - The Cluster object manages the state of the cluster for a particular node
 type Cluster struct {
-	stop      chan bool
-	completed chan bool
-	once      sync.Once
-	Network   []vip.Network
+	stop                  chan bool
+	Network               []vip.Network
+	arpMgr                *arp.Manager
+	routeMgr              *route.Manager
+	nodeLabelMgr          node.Labeler
+	labelAdded            bool
+	healthCheckHTTPClient *http.Client
 }
 
 // InitCluster - Will attempt to initialise all of the required settings for the cluster
-func InitCluster(c *kubevip.Config, disableVIP bool) (*Cluster, error) {
+func InitCluster(c *kubevip.Config, disableVIP bool, intfMgr *networkinterface.Manager, arpMgr *arp.Manager,
+	routeMgr *route.Manager, nodeLabelMgr node.Labeler) (*Cluster, error) {
 	var networks []vip.Network
+	var healthCheckHTTPClient *http.Client
 	var err error
 
 	if !disableVIP {
 		// Start the Virtual IP Networking configuration
-		networks, err = startNetworking(c)
+		networks, err = startNetworking(c, intfMgr)
 		if err != nil {
 			return nil, err
 		}
 	}
-	// Initialise the Cluster structure
-	newCluster := &Cluster{
-		Network: networks,
+
+	if c.ControlPlaneHealthCheck.Address != "" {
+		healthCheckHTTPClient, err = newHealthCheckHTTPClient(c)
+		if err != nil {
+			return nil, fmt.Errorf("initializing BGP health check client: %w", err)
+		}
 	}
 
-	log.Debugf("init enable service security: %t", c.EnableServiceSecurity)
+	// Initialise the Cluster structure
+	newCluster := &Cluster{
+		Network:               networks,
+		arpMgr:                arpMgr,
+		stop:                  make(chan bool),
+		routeMgr:              routeMgr,
+		nodeLabelMgr:          nodeLabelMgr,
+		healthCheckHTTPClient: healthCheckHTTPClient,
+	}
+
+	log.Debug("service security", "enabled", c.EnableServiceSecurity)
 
 	return newCluster, nil
 }
 
-func startNetworking(c *kubevip.Config) ([]vip.Network, error) {
+func startNetworking(c *kubevip.Config, intfMgr *networkinterface.Manager) ([]vip.Network, error) {
 	address := c.VIP
 
 	if c.Address != "" {
@@ -50,7 +78,10 @@ func startNetworking(c *kubevip.Config) ([]vip.Network, error) {
 
 	networks := []vip.Network{}
 	for _, addr := range addresses {
-		network, err := vip.NewConfig(addr, c.Interface, c.LoInterfaceGlobalScope, c.VIPSubnet, c.DDNS, c.RoutingTableID, c.RoutingTableType, c.RoutingProtocol, c.DNSMode, c.LoadBalancerForwardingMethod, c.IptablesBackend)
+		network, err := vip.NewConfig(addr, c.Interface, c.LoInterfaceGlobalScope, c.VIPSubnet, c.DDNS, c.DHCPMode,
+			c.RequireDualStack, c.IsDualStack, c.RoutingTableID, c.RoutingTableType, c.RoutingProtocol, c.DNSMode,
+			c.LoadBalancerForwardingMethod, c.IptablesBackend, c.EnableLoadBalancer, c.LoadBalancerPort,
+			c.EnableServiceSecurity, intfMgr, c.EgressWithNftables, c.SkipDAD)
 		if err != nil {
 			return nil, err
 		}
@@ -64,11 +95,78 @@ func startNetworking(c *kubevip.Config) ([]vip.Network, error) {
 func (cluster *Cluster) Stop() {
 	// Close the stop channel, which will shut down the VIP (if needed)
 	if cluster.stop != nil {
-		cluster.once.Do(func() { // Ensure that the close channel can only ever be called once
-			close(cluster.stop)
-		})
+		close(cluster.stop)
+		cluster.stop = make(chan bool) // recreate channel for future use
+	}
+}
+
+func newHealthCheckHTTPClient(c *kubevip.Config) (*http.Client, error) {
+	defaultTransport, ok := http.DefaultTransport.(*http.Transport)
+	if !ok {
+		return nil, fmt.Errorf("unexpected default HTTP transport type %T", http.DefaultTransport)
 	}
 
-	// Wait until the completed channel is closed, signallign all shutdown tasks completed
-	<-cluster.completed
+	transport := defaultTransport.Clone()
+	if c.ControlPlaneHealthCheck.CAPath != "" {
+		caCert, err := os.ReadFile(c.ControlPlaneHealthCheck.CAPath)
+		if err != nil {
+			return nil, fmt.Errorf("reading health check CA cert %q: %w", c.ControlPlaneHealthCheck.CAPath, err)
+		}
+
+		rootCAs, err := x509.SystemCertPool()
+		if err != nil || rootCAs == nil {
+			rootCAs = x509.NewCertPool()
+		}
+		if !rootCAs.AppendCertsFromPEM(caCert) {
+			return nil, fmt.Errorf("health check CA cert %q contains no valid certificates", c.ControlPlaneHealthCheck.CAPath)
+		}
+
+		tlsConfig := &tls.Config{MinVersion: tls.VersionTLS12}
+		if transport.TLSClientConfig != nil {
+			tlsConfig = transport.TLSClientConfig.Clone()
+		}
+		tlsConfig.RootCAs = rootCAs
+		transport.TLSClientConfig = tlsConfig
+	}
+
+	return &http.Client{
+		Timeout:   time.Duration(c.ControlPlaneHealthCheck.TimeoutSeconds) * time.Second,
+		Transport: transport,
+	}, nil
+}
+
+// cleanupVIPs handles VIP removal based on the PreserveVIPOnLeadershipLoss configuration.
+// When preservation is enabled, IPv6 VIPs are always removed immediately to prevent DAD
+// failures on the new leader, while IPv4 VIPs are intentionally left in place.
+// When preservation is disabled (legacy behavior), all VIPs are removed.
+func (cluster *Cluster) cleanupVIPs(c *kubevip.Config) {
+	for i := range cluster.Network {
+		if c.EnableARP && cluster.arpMgr.Count(cluster.Network[i].ARPName()) > 1 {
+			continue
+		}
+
+		if c.PreserveVIPOnLeadershipLoss {
+			if utils.IsIPv6(cluster.Network[i].IP()) {
+				log.Info("[VIP] Removing IPv6 VIP immediately (required to prevent DAD failures on new leader)", "ip", cluster.Network[i].IP())
+				deleted, err := cluster.Network[i].DeleteIP()
+				if err != nil {
+					log.Warn(err.Error())
+				}
+				if deleted {
+					log.Info("deleted address", "IP", cluster.Network[i].IP(), "interface", cluster.Network[i].Interface())
+				}
+			} else {
+				log.Info("[VIP] Preserving IPv4 VIP address on interface, only stopped ARP broadcasting", "ip", cluster.Network[i].IP())
+			}
+		} else {
+			log.Info("[VIP] Deleting VIP", "ip", cluster.Network[i].IP())
+			deleted, err := cluster.Network[i].DeleteIP()
+			if err != nil {
+				log.Warn(err.Error())
+			}
+			if deleted {
+				log.Info("deleted address", "IP", cluster.Network[i].IP(), "interface", cluster.Network[i].Interface())
+			}
+		}
+	}
 }

pkg/cluster/clusterDDNS.go

@@ -2,6 +2,7 @@ package cluster
 
 import (
 	"context"
+	"sync"
 
 	"github.com/kube-vip/kube-vip/pkg/vip"
 )
@@ -12,16 +13,14 @@ import (
 // during runtime if IP changes, startDDNS don't have to do reconfigure because
 // dnsUpdater already have the functionality to keep trying resolve the IP
 // and update the VIP configuration if it changes
-func (cluster *Cluster) StartDDNS(ctx context.Context) error {
-	for i := range cluster.Network {
-		ddnsMgr := vip.NewDDNSManager(ctx, cluster.Network[i])
-		ip, err := ddnsMgr.Start()
-		if err != nil {
-			return err
-		}
-		if err = cluster.Network[i].SetIP(ip); err != nil {
-			return err
-		}
+func (cluster *Cluster) StartDDNS(ctx context.Context, network vip.Network, backoffAttempts uint, wg *sync.WaitGroup) error {
+	ddnsMgr := vip.NewDDNSManager(network, backoffAttempts)
+	ip, err := ddnsMgr.Start(ctx, wg)
+	if err != nil {
+		return err
+	}
+	if err = network.SetIP(ip); err != nil {
+		return err
 	}
 
 	return nil

pkg/cluster/clusterLeaderElection.go

@@ -3,388 +3,213 @@ package cluster
 import (
 	"context"
 	"fmt"
-	"os"
-	"os/signal"
-	"path/filepath"
-	"syscall"
-	"time"
+	"sync"
 
-	"github.com/davecgh/go-spew/spew"
 	"github.com/kube-vip/kube-vip/pkg/bgp"
-	"github.com/kube-vip/kube-vip/pkg/equinixmetal"
-	"github.com/kube-vip/kube-vip/pkg/etcd"
-	"github.com/kube-vip/kube-vip/pkg/k8s"
+	"github.com/kube-vip/kube-vip/pkg/election"
 	"github.com/kube-vip/kube-vip/pkg/kubevip"
-	"github.com/kube-vip/kube-vip/pkg/loadbalancer"
+	"github.com/kube-vip/kube-vip/pkg/lease"
+	"github.com/kube-vip/kube-vip/pkg/utils"
 
-	"github.com/packethost/packngo"
-
-	log "github.com/sirupsen/logrus"
-	clientv3 "go.etcd.io/etcd/client/v3"
-	v1 "k8s.io/api/core/v1"
-	apierrors "k8s.io/apimachinery/pkg/api/errors"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/apimachinery/pkg/watch"
-	"k8s.io/client-go/kubernetes"
-	"k8s.io/client-go/tools/cache"
-	"k8s.io/client-go/tools/leaderelection"
-	"k8s.io/client-go/tools/leaderelection/resourcelock"
-	watchtools "k8s.io/client-go/tools/watch"
+	log "log/slog"
 )
 
-// Manager degines the manager of the load-balancing services
-type Manager struct {
-	KubernetesClient   *kubernetes.Clientset
-	RetryWatcherClient *kubernetes.Clientset
-	// This channel is used to signal a shutdown
-	SignalChan chan os.Signal
+// StartCluster - Begins a running instance of the Leader Election cluster
+func (cluster *Cluster) StartCluster(ctx context.Context, c *kubevip.Config,
+	em *election.Manager, bgpServer *bgp.Server, leaseMgr *lease.Manager, killFunc func()) error {
 
-	EtcdClient *clientv3.Client
-}
+	ns, leaseName := lease.NamespaceName(c.LeaseName, c)
 
-// NewManager will create a new managing object
-func NewManager(path string, inCluster bool, port int) (*Manager, error) {
-	var hostname string
+	leaseID := lease.NewID(c.LeaderElectionType, ns, leaseName)
 
-	// If inCluster is set then it will likely have started as a static pod or won't have the
-	// VIP up before trying to connect to the API server, we set the API endpoint to this machine to
-	// ensure connectivity. Else if the path passed is empty and not running in the cluster,
-	// attempt to look for a kubeconfig in the default HOME dir.
+	log.Info("cluster membership", "namespace", leaseID.Namespace(), "lock", leaseID.Name(), "id", c.NodeName)
 
-	hostname = fmt.Sprintf("kubernetes:%v", port)
+	objectName := lease.ObjectName(leaseID, "cp")
+	objLease := leaseMgr.Add(ctx, leaseID)
+	isNew := objLease.Add(objectName)
 
-	if len(path) == 0 && !inCluster {
-		path = filepath.Join(os.Getenv("HOME"), ".kube", "config")
+	wg := sync.WaitGroup{}
+	defer wg.Wait()
 
-		// We modify the config so that we can always speak to the correct host
-		id, err := os.Hostname()
-		if err != nil {
-			return nil, err
+	// Start a goroutine that will delete the lease when the service context is cancelled.
+	// This is important for proper cleanup when a service is deleted - it ensures that
+	// the lease context (svcLease.Ctx) gets cancelled, which causes RunOrDie to return.
+	// Without this, RunOrDie would continue running until leadership is naturally lost.
+	wg.Go(func() {
+		<-objLease.Ctx.Done()
+		leaseMgr.Delete(leaseID, objectName, objLease)
+	})
+
+	if !isNew {
+		log.Debug("this election was already done, waiting for it to finish", "lease", leaseName)
+		<-objLease.Ctx.Done()
+		return nil
+	}
+
+	wg.Go(func() {
+		select {
+		case <-cluster.stop:
+		case <-ctx.Done():
 		}
 
-		hostname = fmt.Sprintf("%s:%v", id, port)
-	}
-
-	config, err := k8s.NewRestConfig(path, inCluster, hostname)
-	if err != nil {
-		return nil, fmt.Errorf("failed to create k8s REST config: %w", err)
-	}
-
-	clientset, err := k8s.NewClientset(config)
-	if err != nil {
-		return nil, fmt.Errorf("error creating a new k8s clientset: %v", err)
-	}
-
-	rwConfig, err := k8s.NewRestConfig(path, inCluster, hostname)
-	if err != nil {
-		return nil, fmt.Errorf("failed to create k8s REST config for retryClientSet: %w", err)
-	}
-
-	rwConfig.Timeout = 0 // empty value to disable the timeout
-	rwClientSet, err := k8s.NewClientset(rwConfig)
-	if err != nil {
-		return nil, fmt.Errorf("failed to create k8s client for retry watcher: %w", err)
-	}
-
-	return &Manager{
-		KubernetesClient:   clientset,
-		RetryWatcherClient: rwClientSet,
-	}, nil
-}
-
-// StartCluster - Begins a running instance of the Leader Election cluster
-func (cluster *Cluster) StartCluster(c *kubevip.Config, sm *Manager, bgpServer *bgp.Server) error {
-	var err error
-
-	log.Infof("Beginning cluster membership, namespace [%s], lock name [%s], id [%s]", c.Namespace, c.LeaseName, c.NodeName)
-
-	// use a Go context so we can tell the leaderelection code when we
-	// want to step down
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-
-	// use a Go context so we can tell the arp loop code when we
-	// want to step down
-	ctxArp, cancelArp := context.WithCancel(context.Background())
-	defer cancelArp()
-
-	// use a Go context so we can tell the dns loop code when we
-	// want to step down
-	ctxDNS, cancelDNS := context.WithCancel(context.Background())
-	defer cancelDNS()
-
-	// listen for interrupts or the Linux SIGTERM signal and cancel
-	// our context, which the leader election code will observe and
-	// step down
-	signalChan := make(chan os.Signal, 1)
-	// Add Notification for Userland interrupt
-	signal.Notify(signalChan, syscall.SIGINT)
-
-	// Add Notification for SIGTERM (sent from Kubernetes)
-	signal.Notify(signalChan, syscall.SIGTERM)
-
-	go func() {
-		<-signalChan
 		log.Info("Received termination, signaling cluster shutdown")
-		// Cancel the context, which will in turn cancel the leadership
-		cancel()
-		// Cancel the arp context, which will in turn stop any broadcasts
-	}()
+		// Cancel the leader context, which will in turn cancel the leadership
+		objLease.Cancel()
+	})
 
 	// (attempt to) Remove the virtual IP, in case it already exists
 
 	for i := range cluster.Network {
-		err = cluster.Network[i].DeleteIP()
+		deleted, err := cluster.Network[i].DeleteIP()
 		if err != nil {
-			log.Errorf("could not delete virtualIP: %v", err)
+			log.Error("could not delete virtualIP", "err", err)
+		}
+		if deleted {
+			log.Info("deleted address", "IP", cluster.Network[i].IP(), "interface", cluster.Network[i].Interface())
 		}
 	}
 
-	// Defer a function to check if the bgpServer has been created and if so attempt to close it
+	objLease.Lock()
+
 	defer func() {
-		if bgpServer != nil {
-			bgpServer.Close()
-		}
+		objLease.Unlock()
 	}()
 
-	// If Equinix Metal is enabled then we can begin our preparation work
-	var packetClient *packngo.Client
-	if c.EnableMetal {
-		if c.ProviderConfig != "" {
-			key, project, err := equinixmetal.GetPacketConfig(c.ProviderConfig)
+	// this object is sharing lease with another object
+	if objLease.Elected.Load() {
+		log.Debug("this election was already done, shared lease", "lease", leaseName)
+		// wait for leader election to start or context to be done
+		select {
+		case <-objLease.Started:
+		case <-objLease.Ctx.Done():
+			// Lease was cancelled (e.g., leader election ended), return immediately
+			// This allows the restart loop to create a fresh lease
+			log.Debug("lease context cancelled before leader election started", "lease", leaseName)
+			return fmt.Errorf("lease %q context cancelled before leader election started", leaseName)
+		}
+
+		cluster.OnStartedLeading(c, objLease, em, bgpServer, killFunc, true)
+
+		log.Debug("cluster waiting for leader context done", "lease", leaseName)
+		// wait for leaderelection to be finished
+		<-objLease.Ctx.Done()
+
+		cluster.OnStoppedLeading(c, objLease, bgpServer)
+
+		return nil
+	}
+
+	run := &election.RunConfig{
+		Config:           c,
+		LeaseID:          leaseID,
+		LeaseAnnotations: c.LeaseAnnotations,
+		Mgr:              em,
+		OnStartedLeading: func(context.Context) { //nolint TODO: potential clean code
+			cluster.OnStartedLeading(c, objLease, em, bgpServer, killFunc, false)
+		},
+		OnStoppedLeading: func() {
+			objLease.Elected.Store(false)
+			cluster.OnStoppedLeading(c, objLease, bgpServer)
+		},
+		OnNewLeader: func(identity string) {
+			cluster.OnNewLeader(identity, c)
+		},
+	}
+
+	if err := election.RunOrDie(objLease.Ctx, run, c); err != nil {
+		cluster.Stop()
+		return fmt.Errorf("leaderelection failed: %w", err)
+	}
+
+	return nil
+}
+
+func (cluster *Cluster) OnStartedLeading(c *kubevip.Config, objLease *lease.Lease,
+	em *election.Manager, bgpServer *bgp.Server, killFunc func(), isShared bool) {
+	objLease.Elected.Store(true)
+	objLease.Unlock()
+
+	// When we become leader, ensure we can take over VIPs even if they're preserved on other nodes
+	if !isShared {
+		close(objLease.Started)
+	}
+
+	labels := generateLabelsFromConfig(c.Address, kubevip.HasIP)
+	if err := cluster.nodeLabelMgr.AddLabel(labels); err != nil {
+		log.Error("error adding label to node", "err", err)
+	}
+	cluster.labelAdded = true
+
+	if c.PreserveVIPOnLeadershipLoss {
+		log.Info("Becoming leader with VIP preservation enabled - ensuring VIP takeover")
+		// Force add the VIPs (this will work even if they exist due to the precheck logic)
+		for i := range cluster.Network {
+			added, err := cluster.Network[i].AddIP(true, false)
 			if err != nil {
-				log.Error(err)
+				log.Error("failed to ensure VIP on leader takeover", "vip", cluster.Network[i].IP(), "err", err)
+			} else if added {
+				log.Info("took over VIP as new leader", "IP", cluster.Network[i].IP(), "interface", cluster.Network[i].Interface())
 			} else {
-				// Set the environment variable with the key for the project
-				os.Setenv("PACKET_AUTH_TOKEN", key)
-				// Update the configuration with the project key
-				c.MetalProjectID = project
-			}
-		}
-		packetClient, err = packngo.NewClient()
-		if err != nil {
-			log.Error(err)
-		}
-
-		// We're using Equinix Metal with BGP, populate the Peer information from the API
-		if c.EnableBGP {
-			log.Infoln("Looking up the BGP configuration from Equinix Metal")
-			err = equinixmetal.BGPLookup(packetClient, c)
-			if err != nil {
-				log.Error(err)
+				log.Info("VIP already configured on interface", "IP", cluster.Network[i].IP(), "interface", cluster.Network[i].Interface())
 			}
 		}
 	}
 
-	if c.EnableBGP && bgpServer == nil {
-		// Lets start BGP
-		log.Info("Starting the BGP server to advertise VIP routes to VGP peers")
-		bgpServer, err = bgp.NewBGPServer(&c.BGPConfig, nil)
-		if err != nil {
-			log.Error(err)
-		}
-	}
-
-	run := &runConfig{
-		config:  c,
-		leaseID: c.NodeName,
-		sm:      sm,
-		onStartedLeading: func(ctx context.Context) {
-			// As we're leading lets start the vip service
-			err := cluster.vipService(ctxArp, ctxDNS, c, sm, bgpServer, packetClient)
-			if err != nil {
-				log.Errorf("Error starting the VIP service on the leader [%s]", err)
-			}
-		},
-		onStoppedLeading: func() {
-			// we can do cleanup here
-			log.Info("This node is becoming a follower within the cluster")
-
-			// Stop the dns context
-			cancelDNS()
-			// Stop the Arp context if it is running
-			cancelArp()
-
-			// Stop the BGP server
-			if bgpServer != nil {
-				err := bgpServer.Close()
-				if err != nil {
-					log.Warnf("%v", err)
-				}
-			}
-
-			for i := range cluster.Network {
-				err := cluster.Network[i].DeleteIP()
-				if err != nil {
-					log.Warnf("%v", err)
-				}
-			}
-
-			log.Fatal("lost leadership, restarting kube-vip")
-		},
-		onNewLeader: func(identity string) {
-			// we're notified when new leader elected
-			log.Infof("Node [%s] is assuming leadership of the cluster", identity)
-		},
-	}
-
-	switch c.LeaderElectionType {
-	case "kubernetes", "":
-		cluster.runKubernetesLeaderElectionOrDie(ctx, run)
-	case "etcd":
-		cluster.runEtcdLeaderElectionOrDie(ctx, run)
-	default:
-		log.Info(fmt.Sprintf("LeaderElectionMode %s not supported, exiting", c.LeaderElectionType))
-	}
-
-	return nil
-}
-
-type runConfig struct {
-	config  *kubevip.Config
-	leaseID string
-	sm      *Manager
-
-	// onStartedLeading is called when this member starts leading.
-	onStartedLeading func(context.Context)
-	// onStoppedLeading is called when this member stops leading.
-	onStoppedLeading func()
-	// onNewLeader is called when the client observes a leader that is
-	// not the previously observed leader. This includes the first observed
-	// leader when the client starts.
-	onNewLeader func(identity string)
-}
-
-func (cluster *Cluster) runKubernetesLeaderElectionOrDie(ctx context.Context, run *runConfig) {
-	// we use the Lease lock type since edits to Leases are less common
-	// and fewer objects in the cluster watch "all Leases".
-	lock := &resourcelock.LeaseLock{
-		LeaseMeta: metav1.ObjectMeta{
-			Name:        run.config.LeaseName,
-			Namespace:   run.config.Namespace,
-			Annotations: run.config.LeaseAnnotations,
-		},
-		Client: run.sm.KubernetesClient.CoordinationV1(),
-		LockConfig: resourcelock.ResourceLockConfig{
-			Identity: run.leaseID,
-		},
-	}
-
-	// start the leader election code loop
-	leaderelection.RunOrDie(ctx, leaderelection.LeaderElectionConfig{
-		Lock: lock,
-		// IMPORTANT: you MUST ensure that any code you have that
-		// is protected by the lease must terminate **before**
-		// you call cancel. Otherwise, you could have a background
-		// loop still running and another process could
-		// get elected before your background loop finished, violating
-		// the stated goal of the lease.
-		ReleaseOnCancel: true,
-		LeaseDuration:   time.Duration(run.config.LeaseDuration) * time.Second,
-		RenewDeadline:   time.Duration(run.config.RenewDeadline) * time.Second,
-		RetryPeriod:     time.Duration(run.config.RetryPeriod) * time.Second,
-		Callbacks: leaderelection.LeaderCallbacks{
-			OnStartedLeading: run.onStartedLeading,
-			OnStoppedLeading: run.onStoppedLeading,
-			OnNewLeader:      run.onNewLeader,
-		},
-	})
-}
-
-func (cluster *Cluster) runEtcdLeaderElectionOrDie(ctx context.Context, run *runConfig) {
-	etcd.RunElectionOrDie(ctx, &etcd.LeaderElectionConfig{
-		EtcdConfig:           etcd.ClientConfig{Client: run.sm.EtcdClient},
-		Name:                 run.config.LeaseName,
-		MemberID:             run.leaseID,
-		LeaseDurationSeconds: int64(run.config.LeaseDuration),
-		Callbacks: etcd.LeaderCallbacks{
-			OnStartedLeading: run.onStartedLeading,
-			OnStoppedLeading: run.onStoppedLeading,
-			OnNewLeader:      run.onNewLeader,
-		},
-	})
-}
-
-func (sm *Manager) NodeWatcher(lb *loadbalancer.IPVSLoadBalancer, port int) error {
-	// Use a restartable watcher, as this should help in the event of etcd or timeout issues
-	log.Infof("Kube-Vip is watching nodes for control-plane labels")
-
-	listOptions := metav1.ListOptions{
-		LabelSelector: "node-role.kubernetes.io/control-plane",
-	}
-
-	rw, err := watchtools.NewRetryWatcher("1", &cache.ListWatch{
-		WatchFunc: func(options metav1.ListOptions) (watch.Interface, error) {
-			return sm.RetryWatcherClient.CoreV1().Nodes().Watch(context.Background(), listOptions)
-		},
-	})
+	// As we're leading lets start the vip service
+	err := cluster.StartVipService(objLease.Ctx, c, em, bgpServer, killFunc)
 	if err != nil {
-		return fmt.Errorf("error creating label watcher: %s", err.Error())
+		log.Error("starting VIP service on leader", "err", err)
+		killFunc()
+	}
+}
+
+func (cluster *Cluster) OnStoppedLeading(c *kubevip.Config, objLease *lease.Lease,
+	bgpServer *bgp.Server) {
+	// we can do cleanup here
+	log.Info("This node is becoming a follower within the cluster")
+
+	if cluster.labelAdded {
+		labels := generateLabelsFromConfig(c.Address, kubevip.HasIP)
+		if err := cluster.nodeLabelMgr.RemoveLabel(labels); err != nil {
+			log.Error("error removing label from node", "err", err)
+		}
+		cluster.labelAdded = false
 	}
 
-	go func() {
-		<-sm.SignalChan
-		log.Info("Received termination, signaling shutdown")
-		// Cancel the context
-		rw.Stop()
-	}()
+	// Stop the cluster context if it is running
+	objLease.Cancel()
 
-	ch := rw.ResultChan()
-	// defer rw.Stop()
+	cluster.cleanupVIPs(c)
 
-	for event := range ch {
-		// We need to inspect the event and get ResourceVersion out of it
-		switch event.Type {
-		case watch.Added, watch.Modified:
-			node, ok := event.Object.(*v1.Node)
-			if !ok {
-				return fmt.Errorf("unable to parse Kubernetes Node from Annotation watcher")
+	log.Error("lost leadership, restarting kube-vip")
+}
+
+func (cluster *Cluster) OnNewLeader(identity string, c *kubevip.Config) {
+	// we're notified when new leader elected
+	log.Info("New leader", "leader", identity)
+
+	// If we're not the new leader and we have VIPs preserved from previous leadership,
+	// we need to clean them up to avoid conflicts.
+	if identity != c.NodeName && c.PreserveVIPOnLeadershipLoss {
+		log.Info("Cleaning up preserved VIPs as another node became leader", "new_leader", identity)
+		for i := range cluster.Network {
+			deleted, err := cluster.Network[i].DeleteIP()
+			if err != nil {
+				log.Warn("failed to cleanup preserved VIP", "vip", cluster.Network[i].IP(), "err", err)
 			}
-			// Find the node IP address (this isn't foolproof)
-			for x := range node.Status.Addresses {
-				if node.Status.Addresses[x].Type == v1.NodeInternalIP {
-					err = lb.AddBackend(node.Status.Addresses[x].Address, port)
-					if err != nil {
-						log.Errorf("add IPVS backend [%v]", err)
-					}
-				}
+			if deleted {
+				log.Info("cleaned up preserved VIP to avoid conflict", "IP", cluster.Network[i].IP(),
+					"interface", cluster.Network[i].Interface(), "new_leader", identity)
+			} else {
+				log.Debug("VIP was not present on this node", "IP", cluster.Network[i].IP(),
+					"interface", cluster.Network[i].Interface())
 			}
-		case watch.Deleted:
-			node, ok := event.Object.(*v1.Node)
-			if !ok {
-				return fmt.Errorf("unable to parse Kubernetes Node from Annotation watcher")
-			}
-
-			// Find the node IP address (this isn't foolproof)
-			for x := range node.Status.Addresses {
-				if node.Status.Addresses[x].Type == v1.NodeInternalIP {
-					err = lb.RemoveBackend(node.Status.Addresses[x].Address, port)
-					if err != nil {
-						log.Errorf("Del IPVS backend [%v]", err)
-					}
-				}
-			}
-
-			log.Infof("Node [%s] has been deleted", node.Name)
-
-		case watch.Bookmark:
-			// Un-used
-		case watch.Error:
-			log.Error("Error attempting to watch Kubernetes Nodes")
-
-			// This round trip allows us to handle unstructured status
-			errObject := apierrors.FromObject(event.Object)
-			statusErr, ok := errObject.(*apierrors.StatusError)
-			if !ok {
-				log.Errorf(spew.Sprintf("Received an error which is not *metav1.Status but %#+v", event.Object))
-			}
-
-			status := statusErr.ErrStatus
-			log.Errorf("%v", status)
-		default:
 		}
 	}
-
-	log.Infoln("Exiting Node watcher")
-	return nil
+}
+
+func generateLabelsFromConfig(addr, labelKey string) map[string]string {
+	return map[string]string{
+		labelKey: utils.SanitizeIPForLabel(addr),
+	}
 }

pkg/cluster/cluster_test.go

@@ -0,0 +1,125 @@
+package cluster_test
+
+import (
+	"crypto/ecdsa"
+	"crypto/elliptic"
+	"crypto/rand"
+	"crypto/x509"
+	"encoding/pem"
+	"math/big"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/kube-vip/kube-vip/pkg/cluster"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+)
+
+func TestInitCluster_HealthCheckClientNoCA(t *testing.T) {
+	t.Parallel()
+	cfg := &kubevip.Config{
+		EnableBGP: true,
+		ControlPlaneHealthCheck: kubevip.HealthCheck{
+			Address:        "http://localhost:6443/livez",
+			TimeoutSeconds: 5,
+		},
+	}
+
+	_, err := cluster.InitCluster(cfg, true, nil, nil, nil, nil)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+}
+
+func TestInitCluster_HealthCheckClientValidCA(t *testing.T) {
+	t.Parallel()
+	caPEM := generateTestCACert(t)
+	caFile := filepath.Join(t.TempDir(), "ca.crt")
+	if err := os.WriteFile(caFile, caPEM, 0o600); err != nil {
+		t.Fatal(err)
+	}
+
+	cfg := &kubevip.Config{
+		EnableBGP: true,
+		ControlPlaneHealthCheck: kubevip.HealthCheck{
+			Address:        "https://localhost:6443/livez",
+			TimeoutSeconds: 3,
+			CAPath:         caFile,
+		},
+	}
+
+	_, err := cluster.InitCluster(cfg, true, nil, nil, nil, nil)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+}
+
+func TestInitCluster_HealthCheckClientInvalidCAPath(t *testing.T) {
+	t.Parallel()
+	cfg := &kubevip.Config{
+		EnableBGP: true,
+		ControlPlaneHealthCheck: kubevip.HealthCheck{
+			Address: "https://localhost:6443/livez",
+			CAPath:  "/nonexistent/ca.crt",
+		},
+	}
+
+	_, err := cluster.InitCluster(cfg, true, nil, nil, nil, nil)
+	if err == nil {
+		t.Fatal("expected error for invalid CA path")
+	}
+	if !strings.Contains(err.Error(), "reading health check CA cert") {
+		t.Errorf("expected error about reading CA cert, got: %v", err)
+	}
+}
+
+func TestInitCluster_HealthCheckClientInvalidCAContent(t *testing.T) {
+	t.Parallel()
+	caFile := filepath.Join(t.TempDir(), "bad-ca.crt")
+	if err := os.WriteFile(caFile, []byte("not a certificate"), 0o600); err != nil {
+		t.Fatal(err)
+	}
+
+	cfg := &kubevip.Config{
+		EnableBGP: true,
+		ControlPlaneHealthCheck: kubevip.HealthCheck{
+			Address: "https://localhost:6443/livez",
+			CAPath:  caFile,
+		},
+	}
+
+	_, err := cluster.InitCluster(cfg, true, nil, nil, nil, nil)
+	if err == nil {
+		t.Fatal("expected error for invalid CA content")
+	}
+	if !strings.Contains(err.Error(), "contains no valid certificates") {
+		t.Errorf("expected error about invalid certificates, got: %v", err)
+	}
+}
+
+// generateTestCACert creates a self-signed CA certificate in PEM format for testing.
+func generateTestCACert(t *testing.T) []byte {
+	t.Helper()
+
+	key, err := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	template := &x509.Certificate{
+		SerialNumber:          big.NewInt(1),
+		NotBefore:             time.Now(),
+		NotAfter:              time.Now().Add(time.Hour),
+		IsCA:                  true,
+		BasicConstraintsValid: true,
+	}
+
+	certDER, err := x509.CreateCertificate(rand.Reader, template, template, &key.PublicKey, key)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	return pem.EncodeToMemory(&pem.Block{Type: "CERTIFICATE", Bytes: certDER})
+}

pkg/cluster/service.go

@@ -5,150 +5,141 @@ import (
 	"errors"
 	"fmt"
 	"io/fs"
-	"net"
-	"os"
-	"os/signal"
+	"net/http"
+	"net/url"
+	"strconv"
+	"strings"
+	"sync"
 	"syscall"
 	"time"
 
+	log "log/slog"
+
+	"github.com/kube-vip/kube-vip/pkg/arp"
 	"github.com/kube-vip/kube-vip/pkg/backend"
 	"github.com/kube-vip/kube-vip/pkg/bgp"
-	"github.com/kube-vip/kube-vip/pkg/equinixmetal"
+	"github.com/kube-vip/kube-vip/pkg/election"
 	"github.com/kube-vip/kube-vip/pkg/kubevip"
 	"github.com/kube-vip/kube-vip/pkg/loadbalancer"
+	"github.com/kube-vip/kube-vip/pkg/utils"
 	"github.com/kube-vip/kube-vip/pkg/vip"
-	"github.com/packethost/packngo"
-	log "github.com/sirupsen/logrus"
 	corev1 "k8s.io/api/core/v1"
 	apierrors "k8s.io/apimachinery/pkg/api/errors"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/client-go/kubernetes"
 )
 
-func (cluster *Cluster) vipService(ctxArp, ctxDNS context.Context, c *kubevip.Config, sm *Manager, bgpServer *bgp.Server, packetClient *packngo.Client) error {
+// BGPRouteManager allows to manage the routes announced by the BGP server.
+type BGPRouteManager interface {
+	AddHost(ctx context.Context, addr string, object string) error
+	DelHost(ctx context.Context, addr string, object string) error
+}
+
+func (cluster *Cluster) StartVipService(ctx context.Context, c *kubevip.Config, em *election.Manager,
+	bgpServer BGPRouteManager, killFunc func()) error {
+
 	var err error
 
-	// listen for interrupts or the Linux SIGTERM signal and cancel
-	// our context, which the leader election code will observe and
-	// step down
-	signalChan := make(chan os.Signal, 1)
-	// Add Notification for Userland interrupt
-	signal.Notify(signalChan, syscall.SIGINT)
+	var wg sync.WaitGroup
+	defer wg.Wait()
 
-	// Add Notification for SIGTERM (sent from Kubernetes)
-	signal.Notify(signalChan, syscall.SIGTERM)
+	wg.Go(func() {
+		<-ctx.Done()
+		killFunc()
+	})
+
+	loadbalancers := []*loadbalancer.IPVSLoadBalancer{}
 
 	for i := range cluster.Network {
-		if cluster.Network[i].IsDDNS() {
-			if err := cluster.StartDDNS(ctxDNS); err != nil {
-				log.Error(err)
+		network := cluster.Network[i]
+
+		if network.IsDDNS() {
+			if err := cluster.StartDDNS(ctx, cluster.Network[i], c.DHCPBackoffAttempts, &wg); err != nil {
+				log.Error("failed to start DDNS", "err", err)
 			}
 		}
 
+		if err := network.SetMask(c.VIPSubnet); err != nil {
+			killFunc()
+			return fmt.Errorf("failed to set mask for subnet %q: %w", c.VIPSubnet, err)
+		}
+
 		// start the dns updater if address is dns
-		if cluster.Network[i].IsDNS() {
-			log.Infof("starting the DNS updater for the address %s", cluster.Network[i].DNSName())
-			ipUpdater := vip.NewIPUpdater(cluster.Network[i])
-			ipUpdater.Run(ctxDNS)
+		if network.IsDNS() {
+			log.Info("starting the DNS updater", "address", network.DNSName())
+			ipUpdater := vip.NewIPUpdater(network)
+			wg.Go(func() {
+				ipUpdater.Run(ctx)
+			})
 		}
 
 		if !c.EnableRoutingTable {
-			if c.EnableARP {
-				subnets := vip.Split(c.VIPCIDR)
-				subnet := ""
-				if len(subnets) > 0 {
-					subnet = subnets[0]
-				}
-				if vip.IsIPv6(cluster.Network[i].IP()) && len(subnets) > 1 {
-					subnet = subnets[1]
-				}
-				if subnet == "" {
-					log.Fatalf("no subnet provided for IP %s", cluster.Network[i].IP())
-				}
-				if err = cluster.Network[i].SetMask(subnet); err != nil {
-					log.Fatalf("failed to set mask %s: %s", subnet, err.Error())
-				}
-			}
-			if err = cluster.Network[i].AddIP(false); err != nil {
-				log.Fatalf("%v", err)
-			}
-		}
-
-		if c.EnableMetal {
-			// We're not using Equinix Metal with BGP
-			if !c.EnableBGP {
-				// Attempt to attach the EIP in the standard manner
-				log.Debugf("Attaching the Equinix Metal EIP through the API to this host")
-				err = equinixmetal.AttachEIP(packetClient, c, c.NodeName)
-				if err != nil {
-					log.Error(err)
-				}
+			// Normal VIP addition, use skipDAD=false for normal DAD process
+			if _, err = network.AddIP(false, false); err != nil {
+				log.Error("failed to add IP", "address", network.IP(), "error", err)
 			}
 		}
 
 		if c.EnableBGP {
-			// Lets advertise the VIP over BGP, the host needs to be passed using CIDR notation
-			cidrVip := fmt.Sprintf("%s/%s", cluster.Network[i].IP(), c.VIPCIDR)
-			log.Debugf("Attempting to advertise the address [%s] over BGP", cidrVip)
-
-			err = bgpServer.AddHost(cidrVip)
-			if err != nil {
-				log.Error(err)
+			if c.ControlPlaneHealthCheck.Address != "" {
+				// The health check loop owns route advertisement/withdrawal when configured.
+				wg.Go(func() {
+					cluster.bgpHealthCheckLoop(ctx, c, bgpServer, network.CIDR())
+				})
+			} else {
+				// Lets advertise the VIP over BGP, the host needs to be passed using CIDR notation.
+				log.Debug("Attempting to advertise over BGP", "address", network.CIDR())
+				err = bgpServer.AddHost(ctx, network.CIDR(), c.NodeName)
+				if err != nil {
+					log.Error(err.Error())
+				}
 			}
 		}
 
 		if c.EnableLoadBalancer {
-			log.Infof("Starting IPVS LoadBalancer")
-
-			lb, err := loadbalancer.NewIPVSLB(cluster.Network[i].IP(), c.LoadBalancerPort, c.LoadBalancerForwardingMethod, c.BackendHealthCheckInterval)
+			lb, err := loadbalancer.NewIPVSLB(ctx, network, c.LoadBalancerPort, c.LoadBalancerForwardingMethod,
+				c.BackendHealthCheckInterval, c.EgressWithNftables, killFunc, &wg)
 			if err != nil {
-				log.Errorf("Error creating IPVS LoadBalancer [%s]", err)
+				killFunc()
+				return fmt.Errorf("creating IPVS LoadBalancer: %w", err)
 			}
 
-			go func() {
-				err = sm.NodeWatcher(lb, c.Port)
-				if err != nil {
-					log.Errorf("Error watching node labels [%s]", err)
+			wg.Go(func() {
+				for {
+					select {
+					case <-ctx.Done():
+						return
+					default:
+						err = em.NodeWatcher(ctx, lb, c.Port)
+						if err != nil {
+							log.Error("Error watching node labels", "err", err)
+							if utils.IsPanicError(err) {
+								killFunc()
+								return
+							}
+						}
+					}
 				}
-			}()
-			// Shutdown function that will wait on this signal, unless we call it ourselves
-			go func() {
-				<-signalChan
-				err = lb.RemoveIPVSLB()
-				if err != nil {
-					log.Errorf("Error stopping IPVS LoadBalancer [%s]", err)
-				}
-				log.Info("Stopping IPVS LoadBalancer")
-			}()
+			})
+
+			loadbalancers = append(loadbalancers, lb)
 		}
 
 		if c.EnableARP {
-			go func(ctx context.Context, i int) {
-				ipString := cluster.Network[i].IP()
-				isIPv6 := vip.IsIPv6(ipString)
+			wg.Go(func() {
+				cluster.layer2Update(ctx, network, c)
+			})
+		}
+	}
 
-				var ndp *vip.NdpResponder
-				if isIPv6 {
-					ndp, err = vip.NewNDPResponder(cluster.Network[i].Interface())
-					if err != nil {
-						log.Fatalf("failed to create new NDP Responder")
-					}
-				}
-
-				if ndp != nil {
-					defer ndp.Close()
-				}
-				log.Infof("Gratuitous Arp broadcast will repeat every 3 seconds for [%s/%s]", ipString, cluster.Network[i].Interface())
-				for {
-					select {
-					case <-ctx.Done(): // if cancel() execute
-						return
-					default:
-						cluster.ensureIPAndSendGratuitous(i, ndp)
-					}
-					time.Sleep(3 * time.Second)
-				}
-			}(ctxArp, i)
+	if c.EnableLoadBalancer {
+		// Shutdown function that will wait on this signal, unless we call it ourselves
+		<-ctx.Done()
+		for _, lb := range loadbalancers {
+			err = lb.RemoveIPVSLB()
+			if err != nil {
+				log.Error("Error stopping IPVS LoadBalancer", "err", err)
+			}
 		}
 	}
 
@@ -157,63 +148,51 @@ func (cluster *Cluster) vipService(ctxArp, ctxDNS context.Context, c *kubevip.Co
 		backendMapV6 := backend.Map{}
 		// only check localhost
 
-		nodename := ""
-		if c.NodeName != "" {
-			nodename = c.NodeName
+		// An explicitly configured Kubernetes API address (static-pod
+		// deployments point it at the local API server, whose loopback
+		// listener is often the only certificate-valid local endpoint)
+		// takes precedence over the Node object's addresses: the check
+		// answers "is the local API server healthy" for every VIP family,
+		// regardless of the transport family of the override itself.
+		if entry := kubernetesAddrBackendEntry(c.KubernetesAddr, c.Port); entry != nil {
+			log.Info("using configured Kubernetes address for backend health checks", "address", c.KubernetesAddr)
+			backendMapV4[*entry] = false
+			backendMapV6[*entry] = false
 		} else {
-			nodename = os.Getenv("HOSTNAME")
-		}
-
-		ips := []string{}
-		if nodename != "" {
-			if ips, err = getNodeIPs(ctxArp, nodename, sm.KubernetesClient); err != nil && !apierrors.IsNotFound(err) {
-				log.Error("failed to get IP of control-plane node: %w", err)
-			}
-		}
-
-		if len(ips) == 0 {
-			isV6, err := isV6(cluster.Network[0].IP())
-			if err != nil {
-				return fmt.Errorf("failed to parse IP '%s'", cluster.Network[0].IP())
-			}
-			if !isV6 {
-				ips = append(ips, "127.0.0.1")
-			} else {
-				ips = append(ips, "::1")
-			}
-
-			log.Infof("no IP address found for node - will fallback to use localhost address: %v", ips)
-		}
-
-		for _, ip := range ips {
-			entry := backend.Entry{Addr: ip, Port: c.Port}
-			ipv6, err := isV6(ip)
-			if err != nil {
-				log.Error("failed to check IP type", "IP", ip, "error", err)
-			}
-			if !ipv6 {
-				backendMapV4[entry] = false
-			} else {
-				backendMapV6[entry] = false
-			}
-		}
-
-		stop := make(chan struct{})
-
-		// will wait for system interrupt and will send stop signal to backend watch
-		go func() {
-			<-signalChan
-			stop <- struct{}{}
-		}()
-
-		backend.Watch(func() {
-			for i := range cluster.Network {
-				networkIP := cluster.Network[i].IP()
-				isNetworkV6, err := isV6(networkIP)
-				if err != nil {
-					log.Error("failed to check IP type", "IP", networkIP, "error", err)
-					continue
+			ips := []string{}
+			if c.NodeName != "" {
+				if ips, err = getNodeIPs(ctx, c.NodeName, em.KubernetesClient); err != nil && !apierrors.IsNotFound(err) {
+					log.Error("failed to get IP of control-plane node", "err", err)
 				}
+			}
+
+			if len(ips) == 0 {
+				if !utils.IsIPv6(cluster.Network[0].IP()) {
+					ips = append(ips, "127.0.0.1")
+				} else {
+					ips = append(ips, "::1")
+				}
+
+				log.Info("no IP address found for node - will fallback to use localhost address", "addresses", ips)
+			}
+
+			for _, ip := range ips {
+				entry := backend.Entry{Addr: ip, Port: c.Port}
+				if !utils.IsIPv6(ip) {
+					backendMapV4[entry] = false
+				} else {
+					backendMapV6[entry] = false
+				}
+			}
+		}
+
+		backend.SetKubeConfigPath(c.K8sConfigFile)
+		backend.Watch(ctx, c.BackendHealthCheckInterval, func() {
+			for i := range cluster.Network {
+				network := cluster.Network[i]
+				networkIP := network.IP()
+				isNetworkV6 := utils.IsIPv6(networkIP)
+				log.Debug("current ip to process", "ip", networkIP)
 
 				backendMap := &backendMapV4
 				if isNetworkV6 {
@@ -221,20 +200,50 @@ func (cluster *Cluster) vipService(ctxArp, ctxDNS context.Context, c *kubevip.Co
 				}
 
 				for entry := range *backendMap {
-					if entry.Check() {
-						err = cluster.Network[i].AddIP(true)
+					log.Debug("entry.Check() for entry", "entry", entry)
+					var healthy bool
+					if c.ControlPlaneHealthCheck.Address != "" {
+						req, reqErr := http.NewRequestWithContext(ctx, http.MethodGet, c.ControlPlaneHealthCheck.Address, nil)
+						if reqErr != nil {
+							log.Error("create health check request", "err", reqErr)
+						} else if resp, doErr := cluster.healthCheckHTTPClient.Do(req); doErr != nil {
+							log.Error("health check request failed", "url", c.ControlPlaneHealthCheck.Address, "err", doErr)
+						} else {
+							resp.Body.Close()
+							healthy = resp.StatusCode == http.StatusOK
+							if !healthy {
+								log.Warn("health check returned non-200 status", "url", c.ControlPlaneHealthCheck.Address, "status", resp.StatusCode)
+							}
+						}
+					} else {
+						healthy = entry.Check()
+					}
+					if healthy {
+						log.Debug("entry.Check() true")
+						// Normal VIP addition with precheck, use skipDAD=false for normal DAD process
+						_, err = network.AddIP(true, false)
 						if err != nil {
-							log.Fatalf("error adding IP: %v", err)
+							log.Error("error adding address", "err", err)
 						}
 						if !(*backendMap)[entry] {
-							log.Infof("added IP: %s", cluster.Network[i].IP())
+							log.Info("added backend", "ip", network.IP())
 						}
 
-						err = cluster.Network[i].AddRoute(true)
+						err = cluster.routeMgr.Add(c.NodeName, network, true, false)
 						if err != nil && !errors.Is(err, fs.ErrExist) && !errors.Is(err, syscall.ESRCH) {
-							log.Warnf("%v", err)
+							log.Warn(err.Error())
 						} else if err == nil && !(*backendMap)[entry] {
-							log.Infof("added route: %s", cluster.Network[i].PrepareRoute().String())
+							log.Info("added route", "route", network.PrepareRoute())
+						} else if err == nil || errors.Is(err, fs.ErrExist) {
+							// Re-assert the route on every healthy cycle: routing daemons
+							// (e.g. zebra) can miss the single netlink event for the route,
+							// leaving it unadvertised even though it exists in the kernel.
+							// RouteReplace is idempotent and regenerates that event.
+							if replaceErr := network.ReplaceRoute(); replaceErr != nil {
+								log.Warn("re-asserting route", "err", replaceErr)
+							} else {
+								log.Debug("re-asserted route", "route", network.PrepareRoute())
+							}
 						}
 
 						(*backendMap)[entry] = true
@@ -252,38 +261,127 @@ func (cluster *Cluster) vipService(ctxArp, ctxDNS context.Context, c *kubevip.Co
 				}
 
 				if deleteAddress {
-					err = cluster.Network[i].DeleteRoute()
-					if err != nil && !errors.Is(err, fs.ErrNotExist) && !errors.Is(err, syscall.ESRCH) {
-						log.Warnf("error while deleting route: %v", err)
-					} else if err == nil {
-						log.Infof("deleted route: %s", cluster.Network[i].PrepareRoute().String())
+					err = cluster.routeMgr.Delete(c.NodeName, network)
+					if err != nil {
+						log.Warn("deleting route", "err", err)
 					}
 
-					isSet, err := cluster.Network[i].IsSet()
+					deleted, err := network.DeleteIP()
 					if err != nil {
-						log.Error("failed to check IP address", "error", err)
+						log.Error("error deleting IP", "err", err)
+						killFunc()
+						return
 					}
-					if isSet {
-						err = cluster.Network[i].DeleteIP()
-						if err != nil {
-							log.Fatalf("error deleting IP: %v", err)
-						}
-						log.Infof("deleted IP: %s", cluster.Network[i].IP())
+					if deleted {
+						log.Info("deleted address", "IP", network.IP(), "interface", network.Interface())
 					}
 				}
 			}
-		}, c.BackendHealthCheckInterval, stop)
+		})
+	}
+
+	if c.EnableBGP {
+		<-ctx.Done()
 	}
 
 	return nil
 }
 
-func isV6(ip string) (bool, error) {
-	ipaddr := net.ParseIP(ip)
-	if ipaddr == nil {
-		return false, fmt.Errorf("failed to parse IP '%s'", ip)
+func (cluster *Cluster) bgpHealthCheckLoop(ctx context.Context, c *kubevip.Config, bgpServer BGPRouteManager, vipCIDR string) {
+	period := time.Duration(c.ControlPlaneHealthCheck.PeriodSeconds) * time.Second
+
+	consecutiveFailures := 0
+	routeAnnounced := false
+	ticker := time.NewTicker(period)
+	defer ticker.Stop()
+
+	log.Info("Starting BGP health check",
+		"address", c.ControlPlaneHealthCheck.Address,
+		"cidr", vipCIDR,
+		"period", period,
+		"timeout", cluster.healthCheckHTTPClient.Timeout,
+		"threshold", c.ControlPlaneHealthCheck.FailureThreshold,
+	)
+
+	for {
+		statusCode := 0
+		var healthErr error
+
+		req, err := http.NewRequestWithContext(ctx, http.MethodGet, c.ControlPlaneHealthCheck.Address, nil)
+		if err != nil {
+			healthErr = err
+		} else {
+			resp, err := cluster.healthCheckHTTPClient.Do(req)
+			if err != nil {
+				healthErr = err
+			} else {
+				defer resp.Body.Close()
+				statusCode = resp.StatusCode
+			}
+		}
+
+		healthy := healthErr == nil && statusCode == http.StatusOK
+
+		if healthy {
+			consecutiveFailures = 0
+			if !routeAnnounced {
+				log.Info("BGP health check passed, announcing route", "cidr", vipCIDR)
+				if err := bgpServer.AddHost(ctx, vipCIDR, c.NodeName); err != nil {
+					log.Error("BGP health check: failed to announce route", "cidr", vipCIDR, "err", err)
+				} else {
+					routeAnnounced = true
+				}
+			}
+		} else {
+			consecutiveFailures++
+			if healthErr != nil {
+				log.Warn("BGP health check failed", "address", c.ControlPlaneHealthCheck.Address, "consecutive", consecutiveFailures, "err", healthErr)
+			} else {
+				log.Warn("BGP health check failed", "address", c.ControlPlaneHealthCheck.Address, "consecutive", consecutiveFailures, "status", statusCode)
+			}
+
+			if consecutiveFailures >= c.ControlPlaneHealthCheck.FailureThreshold && routeAnnounced {
+				log.Warn("BGP health check threshold reached, withdrawing route", "failureThreshold", c.ControlPlaneHealthCheck.FailureThreshold, "cidr", vipCIDR)
+				if err := bgpServer.DelHost(ctx, vipCIDR, c.NodeName); err != nil {
+					log.Error("BGP health check: failed to withdraw route", "cidr", vipCIDR, "err", err)
+				} else {
+					routeAnnounced = false
+				}
+			}
+		}
+
+		select {
+		case <-ctx.Done():
+			if routeAnnounced {
+				if err := bgpServer.DelHost(ctx, vipCIDR, c.NodeName); err != nil {
+					log.Error("BGP health check: failed to withdraw route", "cidr", vipCIDR, "err", err)
+				}
+			}
+			return
+		case <-ticker.C:
+		}
 	}
-	return ipaddr.To4() == nil, nil
+}
+
+// kubernetesAddrBackendEntry converts an explicitly configured Kubernetes
+// API address override (config.KubernetesAddr, e.g. "https://127.0.0.1:6443"
+// on static-pod deployments) into a backend health-check entry. Returns nil
+// when no usable override is configured.
+func kubernetesAddrBackendEntry(kubernetesAddr string, defaultPort uint16) *backend.Entry {
+	if kubernetesAddr == "" {
+		return nil
+	}
+	u, err := url.Parse(kubernetesAddr)
+	if err != nil || u.Hostname() == "" {
+		return nil
+	}
+	port := defaultPort
+	if p := u.Port(); p != "" {
+		if parsed, err := strconv.ParseUint(p, 10, 16); err == nil {
+			port = uint16(parsed)
+		}
+	}
+	return &backend.Entry{Addr: u.Hostname(), Port: port}
 }
 
 func getNodeIPs(ctx context.Context, nodename string, client *kubernetes.Clientset) ([]string, error) {
@@ -301,167 +399,180 @@ func getNodeIPs(ctx context.Context, nodename string, client *kubernetes.Clients
 }
 
 // StartLoadBalancerService will start a VIP instance and leave it for kube-proxy to handle
-func (cluster *Cluster) StartLoadBalancerService(c *kubevip.Config, bgp *bgp.Server) {
+func (cluster *Cluster) StartLoadBalancerService(ctx context.Context, c *kubevip.Config, bgp *bgp.Server, name string, wg *sync.WaitGroup) error {
 	// use a Go context so we can tell the arp loop code when we
 	// want to step down
 	//nolint
-	ctxArp, cancelArp := context.WithCancel(context.Background())
+	lbCtx, lbCancel := context.WithCancel(ctx)
 
-	cluster.stop = make(chan bool, 1)
-	cluster.completed = make(chan bool, 1)
+	var lbWg sync.WaitGroup
 
 	for i := range cluster.Network {
 		network := cluster.Network[i]
 
-		err := network.DeleteIP()
-		if err != nil {
-			log.Warnf("Attempted to clean existing VIP => %v", err)
+		if network.IsDDNS() {
+			ddnsReady := make(chan struct{})
+			lbWg.Go(func() {
+				// start the DDNS if requested
+				log.Debug("(svcs) start DDNS", "name", network.DNSName())
+				if err := cluster.StartDDNS(lbCtx, cluster.Network[i], c.DHCPBackoffAttempts, &lbWg); err != nil {
+					log.Error("failed to start DDNS", "err", err)
+				}
+
+				close(ddnsReady)
+				<-lbCtx.Done()
+			})
+			<-ddnsReady
 		}
+
+		log.Debug("current ip to process", "ip", network.IP(), "mask", c.VIPSubnet)
+		if err := network.SetMask(c.VIPSubnet); err != nil {
+			log.Error("failed to set mask", "subnet", c.VIPSubnet, "err", err)
+			lbCancel()
+			return utils.WrapPanicError(err, "failed to set mask for subnet %q", c.VIPSubnet)
+		}
+		_, err := network.DeleteIP()
+		if err != nil {
+			log.Warn("attempted to clean existing VIP", "err", err)
+		}
+		log.Debug("config flags", "enable_routing_table", c.EnableRoutingTable, "enable_leader_election", c.EnableLeaderElection, "enable_services_election", c.EnableServicesElection)
+
 		if c.EnableRoutingTable && (c.EnableLeaderElection || c.EnableServicesElection) {
-			err = network.AddRoute(false)
+			err = cluster.routeMgr.Add(name, network, false, false)
 			if err != nil {
-				log.Warnf("%v", err)
+				log.Warn(err.Error())
+			} else {
+				log.Info("successful add Route")
 			}
-		} else if !c.EnableRoutingTable {
-			if c.EnableARP {
-				subnets := vip.Split(c.VIPCIDR)
-				subnet := ""
-				if len(subnets) > 0 {
-					subnet = subnets[0]
-				}
-				if vip.IsIPv6(cluster.Network[i].IP()) && len(subnets) > 1 {
-					subnet = subnets[1]
-				}
-				if subnet == "" {
-					log.Fatalf("no subnet provided for IP %s", cluster.Network[i].IP())
-				}
-				if err = network.SetMask(subnet); err != nil {
-					log.Fatalf("failed to set mask %s: %s", subnet, err.Error())
-				}
-			}
-			if err = network.AddIP(false); err != nil {
-				log.Warnf("%v", err)
+		}
+
+		if shouldAddServiceIP(c) {
+			// Normal VIP addition, use skipDAD=false for normal DAD process
+			// Note: When WireGuard is enabled, the VIP is added to the tunnel interface
+			// instead of lo, so we skip adding it here.
+			if _, err = network.AddIP(false, false); err != nil {
+				log.Warn(err.Error())
+			} else {
+				log.Info("successful add IP", "address", network.IP())
 			}
 		}
 
 		if c.EnableARP {
-			ipString := network.IP()
-			var ndp *vip.NdpResponder
-			if vip.IsIPv6(ipString) {
-				ndp, err = vip.NewNDPResponder(network.Interface())
-				if err != nil {
-					log.Fatalf("failed to create new NDP Responder")
-				}
-			}
-			go func(ctx context.Context) {
-				if ndp != nil {
-					defer ndp.Close()
-				}
-				log.Debugf("(svcs) broadcasting ARP update for %s via %s, every %dms", ipString, network.Interface(), c.ArpBroadcastRate)
-
-				for {
-					select {
-					case <-ctx.Done(): // if cancel() execute
-						log.Debugf("(svcs) ending ARP update for %s via %s, every %dms", ipString, network.Interface(), c.ArpBroadcastRate)
-						return
-					default:
-						cluster.ensureIPAndSendGratuitous(i, ndp)
-					}
-					if c.ArpBroadcastRate < 500 {
-						log.Errorf("arp broadcast rate is [%d], this shouldn't be lower that 300ms (defaulting to 3000)", c.ArpBroadcastRate)
-						c.ArpBroadcastRate = 3000
-					}
-					time.Sleep(time.Duration(c.ArpBroadcastRate) * time.Millisecond)
-				}
-			}(ctxArp)
+			lbWg.Go(func() {
+				cluster.layer2Update(lbCtx, network, c)
+			})
 		}
 
 		if c.EnableBGP && (c.EnableLeaderElection || c.EnableServicesElection) {
 			// Lets advertise the VIP over BGP, the host needs to be passed using CIDR notation
-			cidrVip := fmt.Sprintf("%s/%s", network.IP(), c.VIPCIDR)
-			log.Debugf("(svcs) attempting to advertise the address [%s] over BGP", cidrVip)
-			err = bgp.AddHost(cidrVip)
+			log.Debug("(svcs) attempting to advertise over BGP", "address", network.CIDR())
+			err = bgp.AddHost(lbCtx, network.CIDR(), name)
 			if err != nil {
-				log.Error(err)
+				log.Error(err.Error())
 			}
 		}
 	}
 
-	go func() {
-		<-cluster.stop
-		// Stop the Arp context if it is running
-		cancelArp()
+	wg.Go(func() {
+		for i := range cluster.Network {
+			network := cluster.Network[i]
 
-		log.Info("[LOADBALANCER] Stopping load balancers")
+			// start the dns updater if address is dns
+			if network.IsDNS() {
+				log.Info("(svcs) starting the DNS updater", "address", network.DNSName(), "ip", network.IP())
+				ipUpdater := vip.NewIPUpdater(network)
+				wg.Go(func() {
+					ipUpdater.Run(lbCtx)
+				})
+			}
+		}
+
+		select {
+		case <-cluster.stop:
+		case <-ctx.Done():
+		}
+
+		// Stop the loadbalancer context if it is running
+		lbCancel()
+
+		lbWg.Wait() // wait for all cluster ARP/NDP to be finished
+
+		log.Info("[LOADBALANCER] Stopping load balancers", "name", name)
 
 		if c.EnableRoutingTable {
 			for i := range cluster.Network {
-				log.Infof("[VIP] Deleting Route for Virtual IP [%s]", cluster.Network[i].IP())
-				if err := cluster.Network[i].DeleteRoute(); err != nil {
-					log.Warnf("%v", err)
+				if err := cluster.routeMgr.Delete(name, cluster.Network[i]); err != nil {
+					log.Warn(err.Error())
 				}
 			}
 
-			close(cluster.completed)
 			return
 		}
-		for i := range cluster.Network {
-			log.Infof("[VIP] Releasing the Virtual IP [%s]", cluster.Network[i].IP())
-			if err := cluster.Network[i].DeleteIP(); err != nil {
-				log.Warnf("%v", err)
-			}
-		}
 
-		close(cluster.completed)
-	}()
+		cluster.cleanupVIPs(c)
+	})
+
+	return nil
 }
 
-// ensureIPAndSendGratuitous - adds IP to the interface if missing, and send
-// either a gratuitous ARP or gratuitous NDP. Re-adds the interface if it is IPv6
-// and in a dadfailed state.
-func (cluster *Cluster) ensureIPAndSendGratuitous(index int, ndp *vip.NdpResponder) {
-	iface := cluster.Network[index].Interface()
-	ipString := cluster.Network[index].IP()
+func shouldAddServiceIP(c *kubevip.Config) bool {
+	return !c.EnableRoutingTable && (!c.EnableBGP || c.BGPAttachIPToInterface) && !c.EnableWireguard
+}
 
-	// Check if IP is dadfailed
-	if cluster.Network[index].IsDADFAIL() {
-		log.Warnf("IP address is in dadfailed state, removing [%s] from interface [%s]", ipString, iface)
-		err := cluster.Network[index].DeleteIP()
-		if err != nil {
-			log.Warnf("%v", err)
-		}
-	}
-
-	// Ensure the address exists on the interface before attempting to ARP
-	set, err := cluster.Network[index].IsSet()
-	if err != nil {
-		log.Warnf("%v", err)
-	}
-	if !set {
-		log.Warnf("Re-applying the VIP configuration [%s] to the interface [%s]", ipString, iface)
-		err = cluster.Network[index].AddIP(false)
-		if err != nil {
-			log.Warnf("%v", err)
-		}
-	}
-
-	if vip.IsIPv6(ipString) {
-		// Gratuitous NDP, will broadcast new MAC <-> IPv6 address
-		if ndp == nil {
-			log.Error("NDP responder was not created")
+// Layer2Update, handles the creation of the
+func (cluster *Cluster) layer2Update(ctx context.Context, network vip.Network, c *kubevip.Config) {
+	var ndp *vip.NdpResponder
+	var err error
+	ipString := network.IP()
+	if utils.IsIPv6(ipString) {
+		if network.IPisLinkLocal() {
+			log.Error("layer2 is link-local can't use NDP", "address", ipString)
 		} else {
-			err := ndp.SendGratuitous(ipString)
+			ndp, err = waitNDPResponder(ctx, network.Interface())
 			if err != nil {
-				log.Warnf("%v", err)
+				log.Error("failed to create new NDP Responder", "error", err)
+			} else {
+				if ndp != nil {
+					defer ndp.Close()
+				}
 			}
 		}
-
-	} else {
-		// Gratuitous ARP, will broadcast to new MAC <-> IPv4 address
-		err := vip.ARPSendGratuitous(ipString, iface)
-		if err != nil {
-			log.Warnf("%v", err)
-		}
 	}
 
+	log.Info("layer 2 broadcaster starting", "IP", network.IP(), "device", network.Interface())
+	log.Debug("layer 2 update", "ip", ipString, "interface", network.Interface(), "ms", c.ArpBroadcastRate)
+
+	arpInstance := arp.NewInstance(network, ndp)
+	cluster.arpMgr.Insert(arpInstance)
+
+	<-ctx.Done() // if cancel() execute
+	log.Debug("ending layer 2 update", "ip", ipString, "interface", network.Interface(), "ms", c.ArpBroadcastRate)
+	cluster.arpMgr.RemoveOnLeadershipLoss(arpInstance)
+}
+
+func waitNDPResponder(ctx context.Context, ifaceName string) (*vip.NdpResponder, error) {
+	ndp, err := vip.NewNDPResponder(ifaceName)
+	if err != nil && strings.Contains(err.Error(), "no such device") {
+		log.Warn("unable to create NDP responder at first try", "interface", ifaceName, "err", err)
+		ndpCreateCtx, cancel := context.WithTimeout(ctx, 2*time.Minute)
+		defer cancel()
+		ticker := time.NewTicker(time.Second)
+
+		for {
+			select {
+			case <-ndpCreateCtx.Done():
+				return nil, fmt.Errorf("failed to create NDP responder for interface %q: %w", ifaceName, ndpCreateCtx.Err())
+			case <-ticker.C:
+				ndp, err = vip.NewNDPResponder(ifaceName)
+				if err != nil {
+					log.Warn("unable to create NDP responder on retry", "interface", ifaceName, "err", err)
+				} else {
+					return ndp, nil
+				}
+			}
+		}
+	} else if err != nil {
+		return nil, fmt.Errorf("unable to create NDP responder for interface %q: %w", ifaceName, err)
+	}
+	return ndp, nil
 }

pkg/cluster/service_config_test.go

@@ -0,0 +1,55 @@
+package cluster
+
+import (
+	"testing"
+
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+)
+
+func TestShouldAddServiceIP(t *testing.T) {
+	tests := []struct {
+		name   string
+		config *kubevip.Config
+		want   bool
+	}{
+		{
+			name:   "BGP default does not attach IP",
+			config: &kubevip.Config{EnableBGP: true},
+			want:   false,
+		},
+		{
+			name: "BGP opt-in attaches IP",
+			config: &kubevip.Config{
+				EnableBGP:              true,
+				BGPAttachIPToInterface: true,
+			},
+			want: true,
+		},
+		{
+			name: "routing table takes precedence",
+			config: &kubevip.Config{
+				EnableBGP:              true,
+				BGPAttachIPToInterface: true,
+				EnableRoutingTable:     true,
+			},
+			want: false,
+		},
+		{
+			name: "WireGuard takes precedence",
+			config: &kubevip.Config{
+				EnableBGP:              true,
+				BGPAttachIPToInterface: true,
+				EnableWireguard:        true,
+			},
+			want: false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := shouldAddServiceIP(tt.config); got != tt.want {
+				t.Fatalf("shouldAddServiceIP() = %t, want %t", got, tt.want)
+			}
+		})
+	}
+}

pkg/cluster/service_internal_test.go

@@ -0,0 +1,61 @@
+package cluster
+
+import (
+	"testing"
+)
+
+func TestKubernetesAddrBackendEntry(t *testing.T) {
+	cases := []struct {
+		name     string
+		addr     string
+		port     uint16
+		wantAddr string
+		wantPort uint16
+		wantNil  bool
+	}{
+		{
+			name:     "explicit v4 loopback with port",
+			addr:     "https://127.0.0.1:6443",
+			port:     9999,
+			wantAddr: "127.0.0.1",
+			wantPort: 6443,
+		},
+		{
+			name:     "hostname without port falls back to config port",
+			addr:     "https://localhost",
+			port:     6443,
+			wantAddr: "localhost",
+			wantPort: 6443,
+		},
+		{
+			name:    "empty override",
+			addr:    "",
+			port:    6443,
+			wantNil: true,
+		},
+		{
+			name:    "garbage override",
+			addr:    "://not-a-url",
+			port:    6443,
+			wantNil: true,
+		},
+	}
+
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			entry := kubernetesAddrBackendEntry(tc.addr, tc.port)
+			if tc.wantNil {
+				if entry != nil {
+					t.Fatalf("expected nil entry, got %+v", entry)
+				}
+				return
+			}
+			if entry == nil {
+				t.Fatal("expected an entry, got nil")
+			}
+			if entry.Addr != tc.wantAddr || entry.Port != tc.wantPort {
+				t.Fatalf("got %+v, want addr %q port %d", entry, tc.wantAddr, tc.wantPort)
+			}
+		})
+	}
+}

pkg/cluster/service_test.go

@@ -0,0 +1,430 @@
+package cluster_test
+
+import (
+	"context"
+	"encoding/pem"
+	"net/http"
+	"net/http/httptest"
+	"os"
+	"path/filepath"
+	"sync"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"github.com/kube-vip/kube-vip/pkg/cluster"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/route"
+	"github.com/kube-vip/kube-vip/pkg/vip"
+	"github.com/vishvananda/netlink"
+	corev1 "k8s.io/api/core/v1"
+)
+
+const testCIDR = "10.0.0.34/32"
+
+func TestBGPHealthCheckLoop_AnnouncesOnHealthy(t *testing.T) {
+	t.Parallel()
+	healthcheck := newTestHealthServer(t, http.StatusOK)
+	t.Cleanup(healthcheck.server.Close)
+
+	bgpManager := newMockBGPRouteManager()
+	startVipService(t, newTestConfig(healthcheck.server.URL, healthcheck.caPath), bgpManager)
+
+	expectEventually(t, func() bool { return bgpManager.isAnnounced() },
+		"route should be announced")
+}
+
+func TestBGPHealthCheckLoop_NoAnnouncementUntilHealthy(t *testing.T) {
+	t.Parallel()
+	healthcheck := newTestHealthServer(t, http.StatusInternalServerError)
+	t.Cleanup(healthcheck.server.Close)
+
+	bgpManager := newMockBGPRouteManager()
+	startVipService(t, newTestConfig(healthcheck.server.URL, healthcheck.caPath), bgpManager)
+
+	expectConsistently(t, func() bool { return !bgpManager.isAnnounced() },
+		2*time.Second, "route should not be announced while unhealthy")
+
+	healthcheck.setStatus(http.StatusOK)
+	expectEventually(t, func() bool { return bgpManager.isAnnounced() },
+		"route should be announced after recovery")
+}
+
+func TestBGPHealthCheckLoop_WithdrawsAfterThreshold(t *testing.T) {
+	t.Parallel()
+	healthcheck := newTestHealthServer(t, http.StatusOK)
+	t.Cleanup(healthcheck.server.Close)
+
+	bgpManager := newMockBGPRouteManager()
+	cfg := newTestConfig(healthcheck.server.URL, healthcheck.caPath)
+	cfg.ControlPlaneHealthCheck.FailureThreshold = 3
+	startVipService(t, cfg, bgpManager)
+
+	expectEventually(t, func() bool { return bgpManager.isAnnounced() },
+		"route should be announced")
+
+	healthcheck.setStatus(http.StatusServiceUnavailable)
+
+	expectConsistently(t, func() bool { return bgpManager.isAnnounced() },
+		1500*time.Millisecond, "route should stay announced before threshold is reached")
+
+	expectEventually(t, func() bool { return !bgpManager.isAnnounced() },
+		"route should be withdrawn after threshold")
+}
+
+func TestBGPHealthCheckLoop_ReAnnouncesOnRecovery(t *testing.T) {
+	t.Parallel()
+	healthcheck := newTestHealthServer(t, http.StatusOK)
+	t.Cleanup(healthcheck.server.Close)
+
+	bgpManager := newMockBGPRouteManager()
+	cfg := newTestConfig(healthcheck.server.URL, healthcheck.caPath)
+	cfg.ControlPlaneHealthCheck.FailureThreshold = 1
+	startVipService(t, cfg, bgpManager)
+
+	expectEventually(t, func() bool { return bgpManager.isAnnounced() },
+		"route should be announced")
+
+	healthcheck.setStatus(http.StatusServiceUnavailable)
+	expectEventually(t, func() bool { return !bgpManager.isAnnounced() },
+		"route should be withdrawn")
+
+	healthcheck.setStatus(http.StatusOK)
+	expectEventually(t, func() bool { return bgpManager.isAnnounced() },
+		"route should be re-announced")
+}
+
+func TestBGPHealthCheckLoop_StopsOnContextCancel(t *testing.T) {
+	t.Parallel()
+	healthcheck := newTestHealthServer(t, http.StatusOK)
+	t.Cleanup(healthcheck.server.Close)
+
+	bgpManager := newMockBGPRouteManager()
+	cancelContext, vipServiceDone := startVipService(t, newTestConfig(healthcheck.server.URL, healthcheck.caPath), bgpManager)
+
+	expectEventually(t, func() bool { return bgpManager.isAnnounced() },
+		"route should be announced")
+
+	cancelContext()
+
+	select {
+	case <-vipServiceDone:
+	case <-time.After(5 * time.Second):
+		t.Fatal("vipService did not stop after context cancellation")
+	}
+}
+
+func TestBGPHealthCheckLoop_RetriesAddHostOnFailure(t *testing.T) {
+	t.Parallel()
+	healthcheck := newTestHealthServer(t, http.StatusOK)
+	t.Cleanup(healthcheck.server.Close)
+
+	bgpManager := newMockBGPRouteManager()
+	bgpManager.setAddErr(errTestAddHost)
+	startVipService(t, newTestConfig(healthcheck.server.URL, healthcheck.caPath), bgpManager)
+
+	expectConsistently(t, func() bool { return !bgpManager.isAnnounced() },
+		2*time.Second, "route should not be announced while AddHost errors")
+
+	bgpManager.setAddErr(nil)
+	expectEventually(t, func() bool { return bgpManager.isAnnounced() },
+		"route should be announced after clearing AddHost error")
+}
+
+func TestBGPHealthCheckLoop_RetriesDelHostOnFailure(t *testing.T) {
+	t.Parallel()
+	healthcheck := newTestHealthServer(t, http.StatusOK)
+	t.Cleanup(healthcheck.server.Close)
+
+	bgpManager := newMockBGPRouteManager()
+	cfg := newTestConfig(healthcheck.server.URL, healthcheck.caPath)
+	cfg.ControlPlaneHealthCheck.FailureThreshold = 1
+	startVipService(t, cfg, bgpManager)
+
+	expectEventually(t, func() bool { return bgpManager.isAnnounced() },
+		"route should be announced")
+
+	bgpManager.setDelErr(errTestDelHost)
+	healthcheck.setStatus(http.StatusServiceUnavailable)
+
+	expectConsistently(t, func() bool { return bgpManager.isAnnounced() },
+		1500*time.Millisecond, "route should stay announced while DelHost errors")
+
+	bgpManager.setDelErr(nil)
+	expectEventually(t, func() bool { return !bgpManager.isAnnounced() },
+		"route should be withdrawn after clearing DelHost error")
+}
+
+func TestRoutingTableHealthCheck_AddsVIPWhenHealthy(t *testing.T) {
+	t.Parallel()
+	healthcheck := newTestHealthServer(t, http.StatusOK)
+	t.Cleanup(healthcheck.server.Close)
+
+	network := &mockNetwork{ip: "10.0.0.1", cidr: testCIDR}
+	startRoutingTableVipService(t, newRoutingTableConfig(healthcheck.server.URL, healthcheck.caPath), network)
+
+	expectEventually(t, network.isPresent,
+		"VIP should be added while health check is healthy")
+}
+
+func TestRoutingTableHealthCheck_RemovesVIPWhenUnhealthy(t *testing.T) {
+	t.Parallel()
+	healthcheck := newTestHealthServer(t, http.StatusOK)
+	t.Cleanup(healthcheck.server.Close)
+
+	network := &mockNetwork{ip: "10.0.0.1", cidr: testCIDR}
+	startRoutingTableVipService(t, newRoutingTableConfig(healthcheck.server.URL, healthcheck.caPath), network)
+
+	expectEventually(t, network.isPresent,
+		"VIP should be added while health check is healthy")
+
+	healthcheck.setStatus(http.StatusServiceUnavailable)
+	expectEventually(t, func() bool { return !network.isPresent() },
+		"VIP should be removed once health check becomes unhealthy")
+}
+
+var (
+	errTestAddHost = &testError{msg: "mock AddHost error"}
+	errTestDelHost = &testError{msg: "mock DelHost error"}
+)
+
+type testError struct{ msg string }
+
+func (e *testError) Error() string { return e.msg }
+
+// startVipService launches vipService in a goroutine with a mock network and
+// registers a cleanup to cancel the context and wait for it to finish.
+// Uses InitCluster so the real code parses certs for the BGP health check client.
+func startVipService(t *testing.T, cfg *kubevip.Config, bgpManager *mockBGPRouteManager) (context.CancelFunc, <-chan struct{}) {
+	t.Helper()
+
+	c, err := cluster.InitCluster(cfg, true, nil, nil, nil, nil)
+	if err != nil {
+		t.Fatalf("InitCluster: %v", err)
+	}
+	c.Network = []vip.Network{&mockNetwork{ip: "10.0.0.1", cidr: testCIDR}}
+
+	ctx, cancel := context.WithCancel(context.Background())
+	done := make(chan struct{})
+
+	go func() {
+		_ = c.StartVipService(ctx, cfg, nil, bgpManager, func() {})
+		close(done)
+	}()
+
+	t.Cleanup(func() {
+		cancel()
+		<-done
+	})
+
+	return cancel, done
+}
+
+// startRoutingTableVipService launches vipService in routing-table mode with a
+// mock network and a real route.Manager (which only drives the mock network's
+// route methods, so no netlink calls happen). Registers cleanup to stop it.
+func startRoutingTableVipService(t *testing.T, cfg *kubevip.Config, network *mockNetwork) {
+	t.Helper()
+
+	c, err := cluster.InitCluster(cfg, true, nil, nil, route.NewManager(), nil)
+	if err != nil {
+		t.Fatalf("InitCluster: %v", err)
+	}
+	c.Network = []vip.Network{network}
+
+	ctx, cancel := context.WithCancel(context.Background())
+	done := make(chan struct{})
+
+	go func() {
+		_ = c.StartVipService(ctx, cfg, nil, nil, func() {})
+		close(done)
+	}()
+
+	t.Cleanup(func() {
+		cancel()
+		<-done
+	})
+}
+
+func newRoutingTableConfig(url, caPath string) *kubevip.Config {
+	cfg := newTestConfig(url, caPath)
+	cfg.EnableBGP = false
+	cfg.EnableRoutingTable = true
+	cfg.BackendHealthCheckInterval = 1
+	return cfg
+}
+
+func newTestConfig(url, caPath string) *kubevip.Config {
+	return &kubevip.Config{
+		EnableBGP: true,
+		ControlPlaneHealthCheck: kubevip.HealthCheck{
+			Address:          url,
+			CAPath:           caPath,
+			PeriodSeconds:    1,
+			TimeoutSeconds:   2,
+			FailureThreshold: 1,
+		},
+	}
+}
+
+// mockBGPRouteManager tracks announced addresses as a set.
+// AddHost adds, DelHost removes. Errors prevent state changes.
+type mockBGPRouteManager struct {
+	mu        sync.Mutex
+	announced map[string]bool
+	addErr    error
+	delErr    error
+}
+
+func newMockBGPRouteManager() *mockBGPRouteManager {
+	return &mockBGPRouteManager{announced: make(map[string]bool)}
+}
+
+func (m *mockBGPRouteManager) AddHost(_ context.Context, addr string, _ string) error {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.addErr != nil {
+		return m.addErr
+	}
+	m.announced[addr] = true
+	return nil
+}
+
+func (m *mockBGPRouteManager) DelHost(_ context.Context, addr string, _ string) error {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.delErr != nil {
+		return m.delErr
+	}
+	delete(m.announced, addr)
+	return nil
+}
+
+func (m *mockBGPRouteManager) isAnnounced() bool {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	return m.announced[testCIDR]
+}
+
+func (m *mockBGPRouteManager) setAddErr(err error) {
+	m.mu.Lock()
+	m.addErr = err
+	m.mu.Unlock()
+}
+
+func (m *mockBGPRouteManager) setDelErr(err error) {
+	m.mu.Lock()
+	m.delErr = err
+	m.mu.Unlock()
+}
+
+// mockNetwork implements vip.Network with no-op operations.
+type mockNetwork struct {
+	ip   string
+	cidr string
+
+	mu      sync.Mutex
+	present bool
+}
+
+func (m *mockNetwork) AddIP(bool, bool, ...int) (bool, error) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	m.present = true
+	return true, nil
+}
+func (m *mockNetwork) DeleteIP() (bool, error) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	m.present = false
+	return false, nil
+}
+func (m *mockNetwork) isPresent() bool {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	return m.present
+}
+func (m *mockNetwork) AddRoute(bool) (bool, error)     { return false, nil }
+func (m *mockNetwork) ReplaceRoute() error             { return nil }
+func (m *mockNetwork) DeleteRoute() error              { return nil }
+func (m *mockNetwork) UpdateRoutes() (bool, error)     { return false, nil }
+func (m *mockNetwork) IsSet() (*netlink.Addr, error)   { return nil, nil }
+func (m *mockNetwork) IP() string                      { return m.ip }
+func (m *mockNetwork) CIDR() string                    { return m.cidr }
+func (m *mockNetwork) IPisLinkLocal() bool             { return false }
+func (m *mockNetwork) PrepareRoute() *netlink.Route    { return nil }
+func (m *mockNetwork) RouteHash() string               { return "" }
+func (m *mockNetwork) SetIP(string) error              { return nil }
+func (m *mockNetwork) SetServicePorts(*corev1.Service) {}
+func (m *mockNetwork) Interface() string               { return "eth0" }
+func (m *mockNetwork) IsDADFAIL() bool                 { return false }
+func (m *mockNetwork) IsDNS() bool                     { return false }
+func (m *mockNetwork) IsDDNS() bool                    { return false }
+func (m *mockNetwork) DDNSHostName() string            { return "" }
+func (m *mockNetwork) DNSName() string                 { return "" }
+func (m *mockNetwork) SetMask(string) error            { return nil }
+func (m *mockNetwork) SetHasEndpoints(bool)            {}
+func (m *mockNetwork) HasEndpoints() bool              { return false }
+func (m *mockNetwork) ARPName() string                 { return "" }
+func (m *mockNetwork) GetPossibleSubnets() string      { return "" }
+func (m *mockNetwork) DHCPFamily() string              { return "" }
+func (m *mockNetwork) IPVSMark() uint32                { return 0 }
+
+// testHealthServer wraps an HTTPS httptest.Server with an atomic status code.
+// caPath is the path to the server's CA cert for client verification.
+type testHealthServer struct {
+	server     *httptest.Server
+	statusCode atomic.Int64
+	caPath     string
+}
+
+func newTestHealthServer(t *testing.T, status int) *testHealthServer {
+	t.Helper()
+	healthcheck := &testHealthServer{}
+	healthcheck.statusCode.Store(int64(status))
+	healthcheck.server = httptest.NewTLSServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
+		w.WriteHeader(int(healthcheck.statusCode.Load()))
+	}))
+
+	cert := healthcheck.server.Certificate()
+	if cert == nil {
+		t.Fatal("TLS server has no certificate")
+	}
+	caPEM := pem.EncodeToMemory(&pem.Block{Type: "CERTIFICATE", Bytes: cert.Raw})
+	caFile := filepath.Join(t.TempDir(), "ca.crt")
+	if err := os.WriteFile(caFile, caPEM, 0o600); err != nil {
+		t.Fatal(err)
+	}
+	healthcheck.caPath = caFile
+	return healthcheck
+}
+
+func (ths *testHealthServer) setStatus(code int) {
+	ths.statusCode.Store(int64(code))
+}
+
+// expectConsistently continuously checks that condition remains true for the given duration.
+// Fails immediately if the condition becomes false at any point.
+func expectConsistently(t *testing.T, condition func() bool, duration time.Duration, msg string) {
+	t.Helper()
+	deadline := time.Now().Add(duration)
+	for time.Now().Before(deadline) {
+		if !condition() {
+			t.Fatalf("condition violated: %s", msg)
+		}
+		time.Sleep(50 * time.Millisecond)
+	}
+}
+
+// expectEventually polls condition until it returns true or 5s timeout is reached.
+func expectEventually(t *testing.T, condition func() bool, msg string) {
+	t.Helper()
+	deadline := time.Now().Add(5 * time.Second)
+	for time.Now().Before(deadline) {
+		if condition() {
+			return
+		}
+		time.Sleep(50 * time.Millisecond)
+	}
+	t.Fatalf("timeout: %s", msg)
+}

pkg/cluster/singleNode.go

@@ -1,79 +0,0 @@
-package cluster
-
-import (
-	"context"
-
-	"github.com/packethost/packngo"
-	log "github.com/sirupsen/logrus"
-
-	"github.com/kube-vip/kube-vip/pkg/bgp"
-	"github.com/kube-vip/kube-vip/pkg/kubevip"
-	"github.com/kube-vip/kube-vip/pkg/vip"
-)
-
-// StartSingleNode will start a single node cluster
-func (cluster *Cluster) StartSingleNode(c *kubevip.Config, disableVIP bool) error {
-	// Start kube-vip as a single node server
-
-	// TODO - Split all this code out as a separate function
-	log.Infoln("Starting kube-vip as a single node cluster")
-
-	log.Info("This node is assuming leadership of the cluster")
-
-	cluster.stop = make(chan bool, 1)
-	cluster.completed = make(chan bool, 1)
-
-	for i := range cluster.Network {
-		if !disableVIP {
-			err := cluster.Network[i].DeleteIP()
-			if err != nil {
-				log.Warnf("Attempted to clean existing VIP => %v", err)
-			}
-
-			err = cluster.Network[i].AddIP(false)
-			if err != nil {
-				log.Warnf("%v", err)
-			}
-
-		}
-
-		if c.EnableARP {
-			// Gratuitous ARP, will broadcast to new MAC <-> IP
-			err := vip.ARPSendGratuitous(cluster.Network[i].IP(), c.Interface)
-			if err != nil {
-				log.Warnf("%v", err)
-			}
-		}
-	}
-
-	go func() {
-		<-cluster.stop
-
-		if !disableVIP {
-			for i := range cluster.Network {
-				log.Infof("[VIP] Releasing the Virtual IP [%s]", cluster.Network[i].IP())
-				err := cluster.Network[i].DeleteIP()
-				if err != nil {
-					log.Warnf("%v", err)
-				}
-			}
-		}
-		close(cluster.completed)
-	}()
-	log.Infoln("Started Load Balancer and Virtual IP")
-	return nil
-}
-
-func (cluster *Cluster) StartVipService(c *kubevip.Config, sm *Manager, bgp *bgp.Server, packetClient *packngo.Client) error {
-	// use a Go context so we can tell the arp loop code when we
-	// want to step down
-	ctxArp, cancelArp := context.WithCancel(context.Background())
-	defer cancelArp()
-
-	// use a Go context so we can tell the dns loop code when we
-	// want to step down
-	ctxDNS, cancelDNS := context.WithCancel(context.Background())
-	defer cancelDNS()
-
-	return cluster.vipService(ctxArp, ctxDNS, c, sm, bgp, packetClient)
-}

pkg/debouncer/debouncer.go

@@ -0,0 +1,252 @@
+package debouncer
+
+import (
+	"context"
+	"fmt"
+	log "log/slog"
+	"sync"
+	"sync/atomic"
+	"time"
+
+	v1 "k8s.io/api/core/v1"
+	discoveryv1 "k8s.io/api/discovery/v1"
+	"k8s.io/apimachinery/pkg/watch"
+)
+
+const (
+	DefaultTime = "0s"
+	minimalTime = time.Millisecond * 200
+)
+
+type debouncer struct {
+	input    <-chan watch.Event
+	output   chan watch.Event
+	stopChan chan any
+	stopOnce sync.Once
+	// events holds event per namespace
+	namespaces   sync.Map
+	debounceTime time.Duration
+}
+
+type ns struct {
+	sync.Map
+	cnt atomic.Int64
+}
+
+func (n *ns) get(name string) (*object, bool) {
+	value, exists := n.Load(name)
+	if !exists {
+		return nil, false
+	}
+	i, ok := value.(*object)
+	if !ok {
+		return nil, false
+	}
+	return i, true
+}
+
+func (n *ns) add(name string, output chan<- watch.Event) *object {
+	i := newObject(output)
+	n.Store(name, i)
+	n.cnt.Add(1)
+	return i
+}
+
+func (n *ns) del(name string) {
+	if _, exists := n.Load(name); exists {
+		n.Delete(name)
+		n.cnt.Add(-1)
+	}
+}
+
+func New(input <-chan watch.Event, debounceTime string) (*debouncer, error) {
+	dt, err := time.ParseDuration(debounceTime)
+	if err != nil {
+		// debouncer was configured with invalid unparsable value, return error
+		return nil, fmt.Errorf("failed to parse debounce time configuration: %w", err)
+	}
+	if dt < minimalTime {
+		if dt > 0 {
+			log.Warn("configured debounce time is less than the minimal threshold of 200ms, debouncer will remain disabled", "config value", dt.String())
+		}
+		return nil, nil
+	}
+	return &debouncer{
+		input:        input,
+		output:       make(chan watch.Event),
+		stopChan:     make(chan any),
+		debounceTime: dt,
+	}, nil
+}
+
+func (d *debouncer) Start(ctx context.Context) error {
+	wg := sync.WaitGroup{}
+	debouncerCtx, cancel := context.WithCancel(ctx)
+	defer func() {
+		cancel()
+		wg.Wait()
+		close(d.output)
+	}()
+
+	for {
+		select {
+		case <-debouncerCtx.Done():
+			// return if debouncer context was cancelled
+			return nil
+		case <-d.stopChan:
+			// return if Stop() was called
+			return nil
+		case tmp := <-d.input:
+			// event has no type, probably error
+			if tmp.Type == "" {
+				return fmt.Errorf("get undefined object (input channel probably closed)")
+			}
+
+			var namespace, name string
+
+			// type switch event object
+			switch v := tmp.Object.(type) {
+			case *discoveryv1.EndpointSlice:
+				namespace = v.Namespace
+				name = v.Name
+			case *v1.Endpoints: //nolint:staticcheck
+				namespace = v.Namespace
+				name = v.Name
+			case *v1.Service:
+				namespace = v.Namespace
+				name = v.Name
+			default:
+				return fmt.Errorf("objects of type %T are not supported", v)
+			}
+
+			eventNs, exists := d.getNs(namespace)
+			if !exists {
+				// if not, create new map for the namespace
+				eventNs = d.addNs(namespace)
+			}
+
+			// check if the object was previously reconciled
+			eventObject, exists := eventNs.get(name)
+
+			// if not and the event is not of type 'Deleted', create new object
+			if !exists && tmp.Type != watch.Deleted {
+				eventObject = eventNs.add(name, d.output)
+
+				wg.Go(func() {
+					// start deboucing events for this object
+					eventObject.start(debouncerCtx, d.debounceTime)
+					// if debouncer for the object ended - e.g. object was deleted - clean the map of objects
+					eventObject = nil
+					eventNs.del(name)
+					// if namespace is empty, delete the namespace map
+					if eventNs.cnt.Load() == 0 {
+						d.delNs(namespace)
+					}
+				})
+			}
+
+			if eventObject != nil {
+				// pass the watch event to the debouncer object
+				eventObject.input <- tmp
+			}
+		}
+	}
+}
+
+func (d *debouncer) Stop() {
+	d.stopOnce.Do(func() {
+		close(d.stopChan)
+	})
+}
+
+func (d *debouncer) Output() chan watch.Event {
+	return d.output
+}
+
+func (d *debouncer) getNs(namespace string) (*ns, bool) {
+	value, exists := d.namespaces.Load(namespace)
+	if !exists {
+		return nil, false
+	}
+	n, ok := value.(*ns)
+	if !ok {
+		return nil, false
+	}
+	return n, true
+}
+
+func (d *debouncer) addNs(namespace string) *ns {
+	n := ns{}
+	d.namespaces.Store(namespace, &n)
+	return &n
+}
+
+func (d *debouncer) delNs(namespace string) {
+	d.namespaces.Delete(namespace)
+}
+
+type object struct {
+	input    chan watch.Event
+	output   chan<- watch.Event
+	stopChan chan any
+	stopOnce sync.Once
+}
+
+func newObject(output chan<- watch.Event) *object {
+	return &object{
+		input:    make(chan watch.Event),
+		output:   output,
+		stopChan: make(chan any),
+	}
+}
+
+func (o *object) start(ctx context.Context, debounceTime time.Duration) {
+	t := time.NewTicker(debounceTime)
+
+	var last *watch.Event
+
+	defer func() {
+		if last != nil {
+			o.output <- *last
+			last = nil
+		}
+	}()
+
+	for {
+		select {
+		case <-ctx.Done():
+			// if context is done, return
+			return
+		case <-o.stopChan:
+			// return if Stop() was called
+			return
+		case tmp := <-o.input:
+			// if last event is known, but an event of another type arrived,
+			// send out the previous event
+			if last != nil && last.Type != tmp.Type {
+				o.output <- *last
+			}
+			// save current event as the last event
+			last = &tmp
+			// reset the ticker to wait for more events
+			t.Reset(debounceTime)
+		case <-t.C:
+			if last != nil {
+				// on tick, if we have an event, send it out
+				o.output <- *last
+				// if the event is of type 'Deleted', stop the debouncer for the object
+				if last.Type == watch.Deleted {
+					o.stop()
+				}
+				// reset last known event, so it won't be send out twice
+				last = nil
+			}
+		}
+	}
+}
+
+func (o *object) stop() {
+	o.stopOnce.Do(func() {
+		close(o.stopChan)
+	})
+}

pkg/debouncer/debouncer_test.go

@@ -0,0 +1,394 @@
+package debouncer
+
+import (
+	"context"
+	"fmt"
+	"strconv"
+	"sync"
+	"testing"
+	"time"
+
+	v1 "k8s.io/api/core/v1"
+	discoveryv1 "k8s.io/api/discovery/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/watch"
+)
+
+func TestTimeSetting(t *testing.T) {
+	tcs := []struct {
+		name       string
+		configured string
+		expected   string
+	}{
+		{
+			name:       "configured proper value 10s",
+			configured: "10s",
+			expected:   "10s",
+		},
+		{
+			name:       "configured value less than 200ms",
+			configured: "0s",
+			expected:   "disabled",
+		},
+		{
+			name:       "configured proper value 1s",
+			configured: "1s",
+			expected:   "1s",
+		},
+		{
+			name:       "configured proper value 1500ms",
+			configured: "1500ms",
+			expected:   "1.5s",
+		},
+		{
+			name:       "configured to value greater than 0s but lower than 200ms",
+			configured: "150ms",
+			expected:   "disabled",
+		},
+		{
+			name:       "configured invalid value that cannot be parsed",
+			configured: "invalid",
+			expected:   "error",
+		},
+		{
+			name:       "configured negative value",
+			configured: "-1s",
+			expected:   "disabled",
+		},
+	}
+
+	input := make(chan watch.Event)
+	defer close(input)
+
+	for _, tc := range tcs {
+		t.Run(tc.name, func(t *testing.T) {
+			d, err := New(input, tc.configured)
+
+			switch tc.expected {
+			case "disabled":
+				if err != nil {
+					t.Fatalf("failed to create debouncer with debounce time %q", tc.configured)
+				}
+				if d != nil {
+					t.Fatalf("debouncer was created but should be disabled for value %q", tc.configured)
+				}
+			case "error":
+				if err == nil {
+					t.Fatalf("debouncer was created but should error for value %q", tc.configured)
+				}
+			default:
+				if d == nil {
+					t.Fatalf("debouncer was not created for value %q", tc.configured)
+				}
+
+				if d.debounceTime.String() != tc.expected {
+					t.Fatalf("invalid debounce time %q was configured instead of expected %q", d.debounceTime.String(), tc.expected)
+				}
+			}
+
+		})
+	}
+}
+
+func TestStartStop(t *testing.T) {
+	t.Run("Run and stop the debouncer without issues", func(t *testing.T) {
+		input := make(chan watch.Event)
+		defer close(input)
+
+		expected := "200ms"
+
+		d, err := New(input, expected)
+
+		if err != nil {
+			t.Fatalf("failed to create debouncer with debounce time %q", expected)
+		}
+
+		if d.debounceTime.String() != expected {
+			t.Fatalf("invalid debounce time %q was configured instead of expected %q", d.debounceTime.String(), expected)
+		}
+
+		ctx, cancel := context.WithCancel(context.Background())
+
+		wg := sync.WaitGroup{}
+
+		wg.Go(func() {
+			if err := d.Start(ctx); err != nil {
+				t.Fatalf("debouncer error: %s", err.Error())
+			}
+		})
+
+		cancel()
+
+		timedOut := waitTimeout(&wg, time.Second*3)
+
+		if timedOut {
+			t.Fatal("debouncer was not closed before timeout")
+		}
+	})
+}
+
+func TestDebouncing(t *testing.T) {
+	tcs := []string{"endpointslices", "endpoints", "services"}
+
+	for _, tc := range tcs {
+		t.Run(fmt.Sprintf("Get the newest event as the only one when using %s", tc), func(t *testing.T) {
+			expected := "500ms"
+
+			fw := watch.NewFake()
+			defer fw.Stop()
+
+			d, err := New(fw.ResultChan(), expected)
+
+			if err != nil {
+				t.Fatalf("failed to create debouncer with debounce time %q", expected)
+			}
+
+			if d.debounceTime.String() != expected {
+				t.Fatalf("invalid debounce time %q was configured instead of expected %q", d.debounceTime.String(), expected)
+			}
+
+			ctx, cancel := context.WithCancel(context.Background())
+
+			wg := sync.WaitGroup{}
+
+			wg.Go(func() {
+				if err := d.Start(ctx); err != nil {
+					t.Fatalf("debouncer error: %s", err.Error())
+				}
+			})
+
+			numOfUpdates := 100
+
+			switch tc {
+			case "endpointslices":
+				epslice := &discoveryv1.EndpointSlice{
+					ObjectMeta: metav1.ObjectMeta{
+						Name:      "test",
+						Namespace: "test",
+					},
+					Endpoints: make([]discoveryv1.Endpoint, 1),
+				}
+
+				addrEpslices := []string{}
+
+				for i := range numOfUpdates {
+					addrEpslices = append(addrEpslices, strconv.Itoa(i))
+					epslice.Endpoints[0].Addresses = addrEpslices
+					fw.Add(epslice)
+				}
+			case "endpoints":
+				ep := &v1.Endpoints{ //nolint:staticcheck
+					ObjectMeta: metav1.ObjectMeta{
+						Name:      "test",
+						Namespace: "test",
+					},
+					Subsets: make([]v1.EndpointSubset, 1), //nolint:staticcheck
+				}
+
+				addrEp := []v1.EndpointAddress{}
+
+				for i := range numOfUpdates {
+					addrEp = append(addrEp, v1.EndpointAddress{IP: strconv.Itoa(i)})
+					ep.Subsets[0].Addresses = addrEp
+					fw.Add(ep)
+				}
+			case "services":
+				svc := &v1.Service{
+					ObjectMeta: metav1.ObjectMeta{
+						Name:      "test",
+						Namespace: "test",
+					},
+				}
+
+				svcPorts := []v1.ServicePort{}
+
+				for i := range numOfUpdates {
+					svcPorts = append(svcPorts, v1.ServicePort{Port: int32(i)})
+					svc.Spec.Ports = svcPorts
+					fw.Add(svc)
+				}
+
+			default:
+				t.Fatal("unknown test", "type", tc)
+			}
+
+			out := <-d.output
+
+			switch tc {
+			case "endpointslices":
+				outEps, ok := out.Object.(*discoveryv1.EndpointSlice)
+				if !ok {
+					t.Fatal("got different type of object than EndpointSlice, failed to cast")
+				}
+
+				if len(outEps.Endpoints[0].Addresses) != numOfUpdates {
+					t.Fatalf("expected to aggregate %d events, but got %d", numOfUpdates, len(outEps.Endpoints[0].Addresses))
+				}
+			case "endpoints":
+				outEps, ok := out.Object.(*v1.Endpoints) //nolint:staticcheck
+				if !ok {
+					t.Fatal("got different type of object than EndpointSlice, failed to cast")
+				}
+
+				if len(outEps.Subsets[0].Addresses) != numOfUpdates {
+					t.Fatalf("expected to aggregate %d events, but got %d", numOfUpdates, len(outEps.Subsets[0].Addresses))
+				}
+			case "services":
+				outSvc, ok := out.Object.(*v1.Service) //nolint:staticcheck
+				if !ok {
+					t.Fatal("got different type of object than EndpointSlice, failed to cast")
+				}
+
+				if len(outSvc.Spec.Ports) != numOfUpdates {
+					t.Fatalf("expected to aggregate %d events, but got %d", numOfUpdates, len(outSvc.Spec.Ports))
+				}
+			default:
+				t.Fatal("unknown test", "type", tc)
+			}
+
+			cancel()
+
+			timedOut := waitTimeout(&wg, time.Second*3)
+
+			if timedOut {
+				t.Fatal("debouncer was not closed before timeout")
+			}
+		})
+	}
+}
+
+func TestTypeChange(t *testing.T) {
+	tcs := []string{"endpointslices", "endpoints", "services"}
+
+	for _, tc := range tcs {
+		t.Run(fmt.Sprintf("Get the newest event as the only one when using %s", tc), func(t *testing.T) {
+			expected := "500ms"
+
+			fw := watch.NewFake()
+			defer fw.Stop()
+
+			d, err := New(fw.ResultChan(), expected)
+
+			if err != nil {
+				t.Fatalf("failed to create debouncer with debounce time %q", expected)
+			}
+
+			if d.debounceTime.String() != expected {
+				t.Fatalf("invalid debounce time %q was configured instead of expected %q", d.debounceTime.String(), expected)
+			}
+
+			ctx, cancel := context.WithCancel(context.Background())
+
+			wg := sync.WaitGroup{}
+
+			wg.Go(func() {
+				if err := d.Start(ctx); err != nil {
+					t.Fatalf("debouncer error: %s", err.Error())
+				}
+			})
+
+			numOfUpdates := 100
+
+			switch tc {
+			case "endpointslices":
+				epslice := &discoveryv1.EndpointSlice{
+					ObjectMeta: metav1.ObjectMeta{
+						Name:      "test",
+						Namespace: "test",
+					},
+					Endpoints: make([]discoveryv1.Endpoint, 1),
+				}
+
+				addrEpslices := []string{}
+
+				for i := range numOfUpdates {
+					addrEpslices = append(addrEpslices, strconv.Itoa(i))
+					epslice.Endpoints[0].Addresses = addrEpslices
+					if i < numOfUpdates-1 {
+						fw.Add(epslice)
+					} else {
+						fw.Delete(epslice)
+					}
+				}
+			case "endpoints":
+				ep := &v1.Endpoints{ //nolint:staticcheck
+					ObjectMeta: metav1.ObjectMeta{
+						Name:      "test",
+						Namespace: "test",
+					},
+					Subsets: make([]v1.EndpointSubset, 1), //nolint:staticcheck
+				}
+
+				addrEp := []v1.EndpointAddress{}
+
+				for i := range numOfUpdates {
+					addrEp = append(addrEp, v1.EndpointAddress{IP: strconv.Itoa(i)})
+					ep.Subsets[0].Addresses = addrEp
+					if i < numOfUpdates-1 {
+						fw.Add(ep)
+					} else {
+						fw.Delete(ep)
+					}
+				}
+			case "services":
+				svc := &v1.Service{
+					ObjectMeta: metav1.ObjectMeta{
+						Name:      "test",
+						Namespace: "test",
+					},
+				}
+
+				svcPorts := []v1.ServicePort{}
+
+				for i := range numOfUpdates {
+					svcPorts = append(svcPorts, v1.ServicePort{Port: int32(i)})
+					svc.Spec.Ports = svcPorts
+					if i < numOfUpdates-1 {
+						fw.Add(svc)
+					} else {
+						fw.Delete(svc)
+					}
+				}
+
+			default:
+				t.Fatal("unknown test", "type", tc)
+			}
+
+			out := <-d.output
+
+			if out.Type != watch.Added {
+				t.Fatalf("expected to get add event, but got %s event", out.Type)
+			}
+
+			out = <-d.output
+
+			if out.Type != watch.Deleted {
+				t.Fatalf("expected to get delete event, but got %s event", out.Type)
+			}
+
+			cancel()
+
+			timedOut := waitTimeout(&wg, time.Second*3)
+
+			if timedOut {
+				t.Fatal("debouncer was not closed before timeout")
+			}
+		})
+	}
+}
+
+// waitTimeout waits for the waitgroup for the specified max timeout.
+// Returns true if waiting timed out.
+func waitTimeout(wg *sync.WaitGroup, timeout time.Duration) bool {
+	c := make(chan struct{})
+	go func() {
+		defer close(c)
+		wg.Wait()
+	}()
+	select {
+	case <-c:
+		return false // completed normally
+	case <-time.After(timeout):
+		return true // timed out
+	}
+}

pkg/detector/interfaces.go

@@ -27,7 +27,7 @@ func FindIPAddress(addrName string) (string, string, error) {
 					// If we're not searching for a specific adapter return the first one
 					if addrName == "" {
 						return iface.Name, address, nil
-					} else
+					}
 					// If this is the correct adapter return the details
 					if iface.Name == addrName {
 						return iface.Name, address, nil
@@ -37,5 +37,5 @@ func FindIPAddress(addrName string) (string, string, error) {
 		}
 
 	}
-	return "", "", fmt.Errorf("Unknown interface [%s]", addrName)
+	return "", "", fmt.Errorf("unknown interface [%s]", addrName)
 }

pkg/egress/egress.go

@@ -0,0 +1,98 @@
+package egress
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/kube-vip/kube-vip/pkg/iptables"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/nftables"
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	"github.com/kube-vip/kube-vip/pkg/vip"
+)
+
+func Teardown(podIP, vipIP, namespace, serviceUUID string, annotations map[string]string, useNftables bool) error {
+	// Look up the destination ports from the annotations on the service
+	destinationPorts := annotations[kubevip.EgressDestinationPorts]
+	deniedNetworks := annotations[kubevip.EgressDeniedNetworks]
+	allowedNetworks := annotations[kubevip.EgressAllowedNetworks]
+	internalEgress := annotations[kubevip.EgressInternal]
+
+	protocol := iptables.ProtocolIPv4
+	if utils.IsIPv6(podIP) {
+		protocol = iptables.ProtocolIPv6
+	}
+
+	// Use the internal egress implementation
+	if internalEgress != "" || useNftables {
+		return nftables.DeleteSNATFromAllTables(serviceUUID)
+	}
+
+	i, err := vip.CreateIptablesClient(useNftables, namespace, protocol)
+	if err != nil {
+		return fmt.Errorf("error Creating iptables client [%s]", err)
+	}
+
+	if deniedNetworks != "" {
+		networks := strings.Split(deniedNetworks, ",")
+		for x := range networks {
+			err = i.DeleteMangleReturnForNetwork(vip.MangleChainName, networks[x])
+			if err != nil {
+				return fmt.Errorf("error deleting rules in mangle chain [%s], error [%s]", vip.MangleChainName, err)
+			}
+		}
+	}
+
+	if allowedNetworks != "" {
+		networks := strings.Split(allowedNetworks, ",")
+		for x := range networks {
+			err = i.DeleteMangleMarkingForNetwork(podIP, vip.MangleChainName, networks[x])
+			if err != nil {
+				return fmt.Errorf("error deleting rules in mangle chain [%s], error [%s]", vip.MangleChainName, err)
+			}
+		}
+	} else {
+		// Remove the marking of egress packets
+		err = i.DeleteMangleMarking(podIP, vip.MangleChainName)
+		if err != nil {
+			return fmt.Errorf("error changing iptables rules for egress [%s]", err)
+		}
+	}
+
+	// Clear up SNAT rules
+	if destinationPorts != "" {
+		fixedPorts := strings.Split(destinationPorts, ",")
+
+		for _, fixedPort := range fixedPorts {
+			var proto, port string
+
+			data := strings.Split(fixedPort, ":")
+			if len(data) == 0 {
+				continue
+			} else if len(data) == 1 {
+				proto = "tcp"
+				port = data[0]
+			} else {
+				proto = data[0]
+				port = data[1]
+			}
+
+			err = i.DeleteSourceNatForDestinationPort(podIP, vipIP, port, proto)
+			if err != nil {
+				return fmt.Errorf("error changing iptables rules for egress [%s]", err)
+			}
+
+		}
+	} else {
+		err = i.DeleteSourceNat(podIP, vipIP)
+		if err != nil {
+			return fmt.Errorf("error changing iptables rules for egress [%s]", err)
+		}
+	}
+
+	err = vip.DeleteExistingSessions(podIP, false, destinationPorts, "")
+	if err != nil {
+		return fmt.Errorf("error changing iptables rules for egress [%s]", err)
+	}
+	return nil
+}

pkg/election/election.go

@@ -0,0 +1,261 @@
+package election
+
+import (
+	"context"
+	"fmt"
+	"sync"
+	"time"
+
+	log "log/slog"
+
+	"github.com/kube-vip/kube-vip/pkg/etcd"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/lease"
+	"github.com/kube-vip/kube-vip/pkg/loadbalancer"
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	clientv3 "go.etcd.io/etcd/client/v3"
+	v1 "k8s.io/api/core/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/watch"
+	"k8s.io/client-go/kubernetes"
+	"k8s.io/client-go/tools/cache"
+	"k8s.io/client-go/tools/leaderelection"
+	"k8s.io/client-go/tools/leaderelection/resourcelock"
+	watchtools "k8s.io/client-go/tools/watch"
+)
+
+type Manager struct {
+	KubernetesClient   *kubernetes.Clientset
+	RetryWatcherClient *kubernetes.Clientset
+	// This channel is used to signal a shutdown
+
+	EtcdClient *clientv3.Client
+}
+
+// NewManager will create a new managing object
+func NewManager(config *kubevip.Config, k8sClientset, rwClientset *kubernetes.Clientset) (*Manager, error) {
+	m := &Manager{}
+
+	switch config.LeaderElectionType {
+	case "kubernetes", "":
+		if k8sClientset == nil || rwClientset == nil {
+			return nil, fmt.Errorf("provided nil clientset")
+		}
+		m.KubernetesClient = k8sClientset
+		m.RetryWatcherClient = rwClientset
+	case "etcd":
+		client, err := etcd.NewClient(config)
+		if err != nil {
+			return nil, err
+		}
+		m.EtcdClient = client
+	default:
+		return nil, fmt.Errorf("invalid LeaderElectionMode %s not supported", config.LeaderElectionType)
+	}
+
+	return m, nil
+}
+
+func RunOrDie(ctx context.Context, run *RunConfig, c *kubevip.Config) error {
+	switch c.LeaderElectionType {
+	case "kubernetes", "":
+		runKubernetesLeaderElectionOrDie(ctx, run)
+	case "etcd":
+		if err := runEtcdLeaderElectionOrDie(ctx, run); err != nil {
+			return err
+		}
+	default:
+		log.Info("LeaderElectionMode not supported, exiting", "mode", c.LeaderElectionType)
+	}
+
+	return nil
+}
+
+func runKubernetesLeaderElectionOrDie(ctx context.Context, run *RunConfig) {
+	// we use the Lease lock type since edits to Leases are less common
+	// and fewer objects in the cluster watch "all Leases".
+	lock := &resourcelock.LeaseLock{
+		LeaseMeta: metav1.ObjectMeta{
+			Name:        run.LeaseID.Name(),
+			Namespace:   run.LeaseID.Namespace(),
+			Annotations: run.LeaseAnnotations,
+		},
+		Client: run.Mgr.KubernetesClient.CoordinationV1(),
+		LockConfig: resourcelock.ResourceLockConfig{
+			Identity: run.Config.NodeName,
+		},
+	}
+
+	// start the leader election code loop
+	leaderelection.RunOrDie(ctx, leaderelection.LeaderElectionConfig{
+		Lock: lock,
+		// IMPORTANT: you MUST ensure that any code you have that
+		// is protected by the lease must terminate **before**
+		// you call cancel. Otherwise, you could have a background
+		// loop still running and another process could
+		// get elected before your background loop finished, violating
+		// the stated goal of the lease.
+		ReleaseOnCancel: true,
+		LeaseDuration:   time.Duration(run.Config.LeaseDuration) * time.Second,
+		RenewDeadline:   time.Duration(run.Config.RenewDeadline) * time.Second,
+		RetryPeriod:     time.Duration(run.Config.RetryPeriod) * time.Second,
+		Callbacks: leaderelection.LeaderCallbacks{
+			OnStartedLeading: run.OnStartedLeading,
+			OnStoppedLeading: run.OnStoppedLeading,
+			OnNewLeader:      run.OnNewLeader,
+		},
+	})
+}
+
+func runEtcdLeaderElectionOrDie(ctx context.Context, run *RunConfig) error {
+	if err := etcd.RunElectionOrDie(ctx, &etcd.LeaderElectionConfig{
+		EtcdConfig:           etcd.ClientConfig{Client: run.Mgr.EtcdClient},
+		Name:                 run.LeaseID.NamespacedName(),
+		MemberID:             run.Config.NodeName,
+		LeaseDurationSeconds: int64(run.Config.LeaseDuration),
+		Callbacks: etcd.LeaderCallbacks{
+			OnStartedLeading: run.OnStartedLeading,
+			OnStoppedLeading: run.OnStoppedLeading,
+			OnNewLeader:      run.OnNewLeader,
+		},
+	}); err != nil {
+		return fmt.Errorf("etcd leaderelection: %w", err)
+	}
+	return nil
+}
+
+type Actions interface {
+	OnStartedLeading(ctx context.Context)
+	OnStoppedLeading()
+	OnNewLeader(identity string)
+}
+
+type RunConfig struct {
+	Config           *kubevip.Config
+	LeaseID          lease.ID
+	Mgr              *Manager
+	LeaseAnnotations map[string]string
+
+	// onStartedLeading is called when this member starts leading.
+	OnStartedLeading func(context.Context)
+	// onStoppedLeading is called when this member stops leading.
+	OnStoppedLeading func()
+	// onNewLeader is called when the client observes a leader that is
+	// not the previously observed leader. This includes the first observed
+	// leader when the client starts.
+	OnNewLeader func(identity string)
+}
+
+func (em *Manager) NodeWatcher(ctx context.Context, lb *loadbalancer.IPVSLoadBalancer, port uint16) error {
+	// Use a restartable watcher, as this should help in the event of etcd or timeout issues
+	log.Info("Kube-Vip is watching nodes for control-plane labels")
+
+	listOptions := metav1.ListOptions{
+		LabelSelector: "node-role.kubernetes.io/control-plane",
+	}
+
+	wg := sync.WaitGroup{}
+	defer wg.Wait()
+
+	watchCtx, watchCancel := context.WithCancel(ctx)
+	defer watchCancel()
+
+	rw, err := watchtools.NewRetryWatcherWithContext(watchCtx, "1", &cache.ListWatch{
+		WatchFunc: func(_ metav1.ListOptions) (watch.Interface, error) {
+			return utils.WatchWithAuthRetry(ctx, func(ctx context.Context) (watch.Interface, error) {
+				return em.RetryWatcherClient.CoreV1().Nodes().Watch(watchCtx, listOptions)
+			})
+		},
+	})
+	if err != nil {
+		return fmt.Errorf("error creating label watcher: %w", err)
+	}
+
+	wg.Go(func() {
+		<-watchCtx.Done()
+		log.Info("Node watcher context cancelled, stopping")
+		// Stop the retrywatcher
+		rw.Stop()
+	})
+
+	ch := rw.ResultChan()
+
+	var watchErr error
+	for event := range ch {
+		// We need to inspect the event and get ResourceVersion out of it
+		switch event.Type {
+		case watch.Added, watch.Modified:
+			node, ok := event.Object.(*v1.Node)
+			if !ok {
+				return fmt.Errorf("unable to parse Kubernetes Node from Annotation watcher")
+			}
+			// Find the node IP address (this isn't foolproof)
+			for x := range node.Status.Addresses {
+				if node.Status.Addresses[x].Type == v1.NodeInternalIP {
+					if checkIfNodeIsReady(node) {
+						err = lb.AddBackend(node.Status.Addresses[x].Address, port)
+						if err != nil {
+							log.Error("adding node to load balancer", "node", node.Name, "ip", node.Status.Addresses[x].Address, "err", err)
+							if utils.IsPanicError(err) {
+								return fmt.Errorf("add IPVS backend: %w", err)
+							}
+						}
+					} else {
+						err = lb.RemoveBackend(node.Status.Addresses[x].Address, port)
+						if err != nil {
+							log.Error("removing node from load balancer", "node", node.Name, "ip", node.Status.Addresses[x].Address, "err", err)
+						}
+					}
+				}
+			}
+		case watch.Deleted:
+			node, ok := event.Object.(*v1.Node)
+			if !ok {
+				return fmt.Errorf("unable to parse Kubernetes Node from Annotation watcher")
+			}
+
+			// Find the node IP address (this isn't foolproof)
+			for x := range node.Status.Addresses {
+				if node.Status.Addresses[x].Type == v1.NodeInternalIP {
+					err = lb.RemoveBackend(node.Status.Addresses[x].Address, port)
+					if err != nil {
+						log.Error("removing node from load balancer", "node", node.Name, "ip", node.Status.Addresses[x].Address, "err", err)
+					}
+				}
+			}
+
+			log.Info("Node deleted", "name", node.Name)
+
+		case watch.Bookmark:
+			// Un-used
+		case watch.Error:
+			log.Error("Error attempting to watch Kubernetes Nodes")
+			watchErr = fmt.Errorf("node watcher error: %w", utils.WatchError(event.Object))
+			log.Error("watcher", "err", watchErr)
+		default:
+		}
+	}
+
+	log.Info("Exiting Node watcher")
+	if watchErr != nil {
+		return watchErr
+	}
+	if ctx.Err() != nil {
+		return nil
+	}
+	return utils.NewPanicError("node watcher channel closed unexpectedly")
+}
+
+func checkIfNodeIsReady(node *v1.Node) bool {
+	if node == nil {
+		return false
+	}
+	for _, condition := range node.Status.Conditions {
+		if condition.Type == v1.NodeReady {
+			if condition.Status == v1.ConditionTrue {
+				return true
+			}
+		}
+	}
+	return false
+}

pkg/endpoints/endpoints.go

@@ -0,0 +1,368 @@
+package endpoints
+
+import (
+	"context"
+	"fmt"
+	"net"
+	"strings"
+	"sync"
+	"time"
+
+	log "log/slog"
+
+	"github.com/kube-vip/kube-vip/pkg/bgp"
+	"github.com/kube-vip/kube-vip/pkg/endpoints/providers"
+	"github.com/kube-vip/kube-vip/pkg/instance"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/lease"
+	"github.com/kube-vip/kube-vip/pkg/metrics"
+	"github.com/kube-vip/kube-vip/pkg/route"
+	"github.com/kube-vip/kube-vip/pkg/servicecontext"
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	"github.com/kube-vip/kube-vip/pkg/wireguard"
+	v1 "k8s.io/api/core/v1"
+	"k8s.io/apimachinery/pkg/watch"
+	"k8s.io/client-go/kubernetes"
+)
+
+type Processor struct {
+	config    *kubevip.Config
+	provider  providers.Provider
+	bgpServer *bgp.Server
+	worker    endpointWorker
+	instances *[]*instance.Instance
+	leaseMgr  *lease.Manager
+}
+
+func NewEndpointProcessor(config *kubevip.Config, provider providers.Provider, bgpServer *bgp.Server,
+	instances *[]*instance.Instance, leaseMgr *lease.Manager, tunnelMgr *wireguard.TunnelManager, routeMgr *route.Manager) *Processor {
+	return &Processor{
+		config:    config,
+		provider:  provider,
+		bgpServer: bgpServer,
+		instances: instances,
+		leaseMgr:  leaseMgr,
+		worker:    newEndpointWorker(config, provider, bgpServer, instances, leaseMgr, tunnelMgr, routeMgr),
+	}
+}
+
+// Reconcile applies a watch event to the provider and reconciles the service
+// against the endpoints that remain afterwards. A deleted object is only one of
+// potentially several backing the service, so deletions are recomputed rather
+// than assumed to empty it. It reports whether the caller should skip this event
+// and wait for the next one.
+func (p *Processor) Reconcile(svcCtx *servicecontext.Context, event watch.Event,
+	lastKnownGoodEndpoint *string, service *v1.Service, id string,
+	serviceFunc func(*servicecontext.Context, *v1.Service, *sync.WaitGroup, bool) error, wg *sync.WaitGroup,
+	clientSet *kubernetes.Clientset,
+	egressUpdateFunc func(context.Context, *v1.Service) error) (bool, error) {
+
+	if err := p.applyEvent(svcCtx, event); err != nil {
+		return false, err
+	}
+
+	endpoints, err := p.worker.getEndpoints(service, id)
+	if err != nil {
+		return false, fmt.Errorf("[%s] error getting endpoints: %w", p.provider.GetLabel(), err)
+	}
+
+	if err := p.worker.setInstanceEndpointsStatus(svcCtx.Ctx, service, endpoints); err != nil {
+		log.Error("updating instance", "err", err)
+	}
+
+	allowReconcileWithoutEndpoints := shouldAllowReconcileWithoutEndpoints(service)
+
+	// Find out if we have any local endpoints
+	// if out endpoint is empty then populate it
+	// if not, go through the endpoints and see if ours still exists
+	// If we have a local endpoint then begin the leader Election, unless it's already running
+	//
+
+	// Check that we have local endpoints
+	if len(endpoints) != 0 {
+		// Ignore IPv4
+		if service.Annotations[kubevip.EgressIPv6] == "true" && !hasV6(endpoints) {
+			return true, nil
+		}
+
+		p.updateLastKnownGoodEndpoint(lastKnownGoodEndpoint, endpoints, service)
+
+		if err := p.startServiceHandlingIfNeeded(svcCtx, service, serviceFunc, wg); err != nil {
+			return true, err
+		}
+
+		svcCtx.SignalReadiness()
+
+		if p.shouldProcessInstance() {
+			if err := p.worker.processInstance(svcCtx, service); err != nil {
+				return false, fmt.Errorf("failed to process non-empty instance: %w", err)
+			}
+		}
+	} else {
+		if allowReconcileWithoutEndpoints {
+			// Explicit opt-in for controllers that create LoadBalancer services without endpoints
+			if err := p.startServiceHandlingIfNeeded(svcCtx, service, serviceFunc, wg); err != nil {
+				return true, err
+			}
+			svcCtx.SignalReadiness()
+
+			if p.shouldProcessInstance() {
+				if err := p.worker.processInstance(svcCtx, service); err != nil {
+					return false, fmt.Errorf("failed to process endpointless instance: %w", err)
+				}
+			}
+		} else if svcCtx.Signalled.Load() {
+			p.handleNoEndpoints(svcCtx, service, lastKnownGoodEndpoint)
+		}
+	}
+
+	// Set the service accordingly
+	p.updateAnnotations(service, lastKnownGoodEndpoint, clientSet, egressUpdateFunc)
+
+	log.Debug("watcher", "provider",
+		p.provider.GetLabel(), "service name", service.Name, "namespace", service.Namespace, "endpoints", len(endpoints), "last endpoint", *lastKnownGoodEndpoint)
+
+	return false, nil
+}
+
+// applyEvent updates the provider's view of the objects backing this service.
+func (p *Processor) applyEvent(svcCtx *servicecontext.Context, event watch.Event) error {
+	if event.Type == watch.Deleted {
+		if err := p.provider.DeleteObject(event.Object); err != nil {
+			return fmt.Errorf("[%s] error deleting k8s object: %w", p.provider.GetLabel(), err)
+		}
+		return nil
+	}
+
+	if err := p.provider.LoadObject(event.Object, svcCtx.Cancel); err != nil {
+		return fmt.Errorf("[%s] error loading k8s object: %w", p.provider.GetLabel(), err)
+	}
+	return nil
+}
+
+// shouldProcessInstance reports whether this node has to program the datapath
+// itself, rather than waiting to be told to by a leader election callback.
+// WireGuard always reprograms, because its DNAT rules are per-endpoint.
+func (p *Processor) shouldProcessInstance() bool {
+	return (!p.config.EnableServicesElection && !p.config.EnableLeaderElection) || p.config.EnableWireguard
+}
+
+// handleNoEndpoints tears down everything backing a service that no longer has
+// any usable endpoints.
+func (p *Processor) handleNoEndpoints(svcCtx *servicecontext.Context, service *v1.Service, lastKnownGoodEndpoint *string) {
+	svcCtx.ResetReadiness()
+	p.worker.clear(svcCtx, lastKnownGoodEndpoint, service)
+	if p.config.EnableARP && !p.config.EnableServicesElection && p.instances != nil {
+		if i := instance.FindServiceInstance(service, *p.instances); i != nil {
+			for _, c := range i.Clusters {
+				c.Stop()
+			}
+		}
+	}
+}
+
+func (p *Processor) updateLastKnownGoodEndpoint(lastKnownGoodEndpoint *string, endpoints []string, service *v1.Service) {
+	// if we haven't populated one, then do so
+	family := utils.IPv4Family
+	if service.Annotations[kubevip.EgressIPv6] == "true" {
+		family = utils.IPv6Family
+	}
+
+	ep := getEndpoint(endpoints, family)
+
+	if *lastKnownGoodEndpoint == "" {
+		*lastKnownGoodEndpoint = ep
+		return
+	}
+
+	// check out previous endpoint exists
+	stillExists := false
+
+	for x := range endpoints {
+		if endpoints[x] == *lastKnownGoodEndpoint {
+			stillExists = true
+		}
+	}
+	// If the last endpoint no longer exists, we cancel our leader Election, and set another endpoint as last known good
+	if !stillExists {
+		ip := net.ParseIP(*lastKnownGoodEndpoint)
+		if (ip.To4() != nil && service.Annotations[kubevip.Egress] == "true") ||
+			(ip.To4() == nil && service.Annotations[kubevip.EgressIPv6] == "true") {
+			p.worker.removeEgress(service, lastKnownGoodEndpoint)
+		}
+		// Set our active endpoint to an existing one
+		*lastKnownGoodEndpoint = ep
+	}
+}
+
+func (p *Processor) updateAnnotations(service *v1.Service, lastKnownGoodEndpoint *string,
+	clientSet *kubernetes.Clientset,
+	egressUpdateFunc func(context.Context, *v1.Service) error) {
+	// Set the service accordingly
+	if service.Annotations[kubevip.Egress] == "true" {
+		ip := net.ParseIP(*lastKnownGoodEndpoint)
+
+		// Store old values from ServiceSnapshot to detect if annotation actually changed
+		// We use the ServiceSnapshot instead of the service parameter because the service parameter
+		// may have stale annotations if the last update failed
+		var oldEndpoint, oldEndpointIPv6 string
+		if p.instances != nil {
+			serviceInstance := instance.FindServiceInstance(service, *p.instances)
+			if serviceInstance != nil {
+				oldEndpoint = serviceInstance.ServiceSnapshot.Annotations[kubevip.ActiveEndpoint]
+				oldEndpointIPv6 = serviceInstance.ServiceSnapshot.Annotations[kubevip.ActiveEndpointIPv6]
+			}
+		}
+		// Fall back to service annotations if we couldn't find the instance
+		if oldEndpoint == "" && oldEndpointIPv6 == "" {
+			oldEndpoint = service.Annotations[kubevip.ActiveEndpoint]
+			oldEndpointIPv6 = service.Annotations[kubevip.ActiveEndpointIPv6]
+		}
+
+		// Determine which annotation to update based on IP version
+		var endpoint, endpointIPv6 string
+		if ip.To4() == nil && !p.config.EnableEndpoints {
+			// IPv6
+			endpointIPv6 = *lastKnownGoodEndpoint
+			endpoint = oldEndpoint // Preserve existing IPv4 if any
+		} else {
+			// IPv4
+			endpoint = *lastKnownGoodEndpoint
+			endpointIPv6 = oldEndpointIPv6 // Preserve existing IPv6 if any
+		}
+
+		// Check if annotation actually changed
+		annotationChanged := (oldEndpoint != endpoint) || (oldEndpointIPv6 != endpointIPv6)
+		if !annotationChanged {
+			return // Nothing to do
+		}
+
+		// Persist to Kubernetes
+		ctx := context.Background()
+
+		if err := p.provider.UpdateServiceAnnotation(ctx, endpoint, endpointIPv6, service, clientSet); err != nil {
+			log.Warn("failed to update service annotation", "service", service.Name, "namespace", service.Namespace, "err", err)
+			return
+		}
+
+		log.Debug("updated active endpoint annotation", "service", service.Name, "namespace", service.Namespace, "endpoint", *lastKnownGoodEndpoint)
+
+		// Trigger egress reconfiguration
+		// For services with leader election, the service watcher doesn't process Modified events
+		// after initial setup, so we need to directly call the update function
+		if egressUpdateFunc != nil {
+			// Create a copy of service with updated annotations
+			svcCopy := service.DeepCopy()
+			svcCopy.Annotations[kubevip.ActiveEndpoint] = endpoint
+			svcCopy.Annotations[kubevip.ActiveEndpointIPv6] = endpointIPv6
+
+			if err := egressUpdateFunc(ctx, svcCopy); err != nil {
+				log.Error("failed to reconfigure egress", "service", service.Name, "namespace", service.Namespace, "err", err)
+			}
+		}
+	}
+}
+
+func (p *Processor) startServiceHandlingIfNeeded(svcCtx *servicecontext.Context, service *v1.Service,
+	serviceFunc func(*servicecontext.Context, *v1.Service, *sync.WaitGroup, bool) error, wg *sync.WaitGroup) error {
+	if p.config.EnableServicesElection {
+		// startLeaderElection restarts itself until the service context is cancelled,
+		// so start it only once instead of on every endpoint event.
+		svcCtx.StartLeaderElectionOnce(func() {
+			wg.Go(func() {
+				p.startLeaderElection(svcCtx, service, serviceFunc, wg)
+			})
+		})
+		return nil
+	}
+
+	if p.config.EnableARP || (p.config.EnableRoutingTable && p.config.EnableLeaderElection) {
+		if !svcCtx.Signalled.Load() {
+			inst := instance.FindServiceInstance(service, *p.instances)
+			if inst == nil {
+				return fmt.Errorf("[%s] failed to find an instance for service %s/%s", p.provider.GetLabel(), service.Namespace, service.Name)
+			}
+			for x := range inst.VIPConfigs {
+				log.Debug("starting loadbalancer for service", "provider", p.provider.GetLabel(), "name", service.Name, "namespace", service.Namespace, "uid", service.UID)
+				if err := inst.Clusters[x].StartLoadBalancerService(svcCtx.Ctx, inst.VIPConfigs[x], p.bgpServer, lease.ServiceNamespacedName(service), wg); err != nil {
+					return fmt.Errorf("failed to start lb: %w", err)
+				}
+			}
+		}
+	}
+
+	return nil
+}
+
+func (p *Processor) startLeaderElection(svcCtx *servicecontext.Context, service *v1.Service, serviceFunc func(*servicecontext.Context, *v1.Service, *sync.WaitGroup, bool) error, wg *sync.WaitGroup) {
+	// Track this loop for the lifetime of the goroutine. There has to be at most
+	// one per service, so a value above 1 means loops leaked.
+	loops := metrics.ServiceElectionLoops.WithLabelValues(service.Namespace, service.Name)
+	loops.Inc()
+	defer loops.Dec()
+
+	attempts := metrics.ServiceElectionAttemptsTotal.WithLabelValues(service.Namespace, service.Name)
+
+	// This is a blocking function, that will restart (in the event of failure)
+	for {
+		select {
+		case <-svcCtx.Ctx.Done():
+			return
+		default:
+			leaseNamespace, serviceLease := lease.ServiceName(service)
+			id := lease.NewID(p.config.LeaderElectionType, leaseNamespace, serviceLease)
+			// The lease is retired once its last service is gone, so an absent one means
+			// this loop has nothing left to elect for.
+			l := p.leaseMgr.Get(id)
+			if l == nil {
+				return
+			}
+			l.Lock()
+
+			if !l.Elected.Load() {
+				l.Unlock()
+				attempts.Inc()
+				err := serviceFunc(svcCtx, service, wg, true)
+				if err != nil {
+					log.Error(err.Error())
+				}
+			} else {
+				l.Unlock()
+				time.Sleep(time.Millisecond * 200)
+			}
+		}
+	}
+}
+
+func shouldAllowReconcileWithoutEndpoints(service *v1.Service) bool {
+	if service == nil || service.Spec.ExternalTrafficPolicy != v1.ServiceExternalTrafficPolicyTypeCluster {
+		return false
+	}
+
+	return strings.EqualFold(service.Annotations[kubevip.AllowReconcileWithoutEndpoints], "true")
+}
+
+func hasV6(endpoints []string) bool {
+	for _, e := range endpoints {
+		ip := net.ParseIP(e)
+		if ip != nil {
+			if ip.To4() == nil {
+				return true
+			}
+		}
+	}
+	return false
+}
+
+func getEndpoint(endpoints []string, family string) string {
+	for _, e := range endpoints {
+		ip := net.ParseIP(e)
+		if family == utils.IPv4Family && ip.To4() != nil {
+			return e
+		}
+		if family == utils.IPv6Family && ip.To4() == nil {
+			return e
+		}
+	}
+	return ""
+}

pkg/endpoints/endpoints_bgp.go

@@ -0,0 +1,105 @@
+package endpoints
+
+import (
+	"context"
+	log "log/slog"
+
+	"github.com/kube-vip/kube-vip/pkg/bgp"
+	"github.com/kube-vip/kube-vip/pkg/instance"
+	"github.com/kube-vip/kube-vip/pkg/lease"
+	"github.com/kube-vip/kube-vip/pkg/servicecontext"
+	v1 "k8s.io/api/core/v1"
+)
+
+type BGP struct {
+	generic
+	bgpServer *bgp.Server
+}
+
+func newBGP(generic generic, bgpServer *bgp.Server) endpointWorker {
+	return &BGP{
+		generic:   generic,
+		bgpServer: bgpServer,
+	}
+}
+
+func (b *BGP) processInstance(svcCtx *servicecontext.Context, service *v1.Service) error {
+	if instance := instance.FindServiceInstance(service, *b.instances); instance != nil {
+		for _, cluster := range instance.Clusters {
+			for i := range cluster.Network {
+				if !svcCtx.IsNetworkConfigured(cluster.Network[i].IP()) {
+					log.Debug("attempting to advertise BGP service", "provider", b.provider.GetLabel(), "ip", cluster.Network[i].IP())
+					err := b.bgpServer.AddHost(svcCtx.Ctx, cluster.Network[i].CIDR(), lease.ServiceNamespacedName(service))
+					if err != nil {
+						log.Error("error adding BGP host", "provider", b.provider.GetLabel(), "err", err)
+					} else {
+						log.Info("added BGP host", "provider",
+							b.provider.GetLabel(), "ip", cluster.Network[i].CIDR(), "service name", service.Name, "namespace", service.Namespace)
+						svcCtx.ConfiguredNetworks.Store(cluster.Network[i].IP(), true)
+					}
+				}
+			}
+		}
+	}
+	return nil
+}
+
+func (b *BGP) clear(svcCtx *servicecontext.Context, lastKnownGoodEndpoint *string, service *v1.Service) {
+	if !b.config.EnableServicesElection && !b.config.EnableLeaderElection {
+		// If BGP mode is enabled - routes should be deleted
+		if instance := instance.FindServiceInstance(service, *b.instances); instance != nil {
+			for _, cluster := range instance.Clusters {
+				for i := range cluster.Network {
+					err := b.bgpServer.DelHost(svcCtx.Ctx, cluster.Network[i].CIDR(), lease.ServiceNamespacedName(service))
+					if err != nil {
+						log.Error("deleting BGP host", "provider", b.provider.GetLabel(), "ip", cluster.Network[i].IP(), "err", err)
+					} else {
+						log.Info("deleted BGP host", "provider",
+							b.provider.GetLabel(), "ip", cluster.Network[i].IP(), "service name", service.Name, "namespace", service.Namespace)
+						svcCtx.ConfiguredNetworks.Delete(cluster.Network[i].IP())
+					}
+				}
+			}
+
+		}
+	}
+
+	b.clearEgress(lastKnownGoodEndpoint, service)
+
+	if svcCtx.LeaderCancel != nil {
+		svcCtx.LeaderCancel()
+	}
+}
+
+func (b *BGP) getEndpoints(service *v1.Service, id string) ([]string, error) {
+	return b.getAllEndpoints(service, id)
+}
+
+func (b *BGP) setInstanceEndpointsStatus(_ context.Context, _ *v1.Service, _ []string) error {
+	return nil
+}
+
+func ClearBGPHosts(ctx context.Context, service *v1.Service, instances *[]*instance.Instance, bgpServer *bgp.Server) {
+	if instance := instance.FindServiceInstance(service, *instances); instance != nil {
+		ClearBGPHostsByInstance(ctx, instance, bgpServer)
+	}
+}
+
+func ClearBGPHostsByInstance(ctx context.Context, instance *instance.Instance, bgpServer *bgp.Server) {
+	if instance == nil {
+		log.Error("failed to clear BGP host for nil instance")
+		return
+	}
+	for _, cluster := range instance.Clusters {
+		for i := range cluster.Network {
+			network := cluster.Network[i]
+			err := bgpServer.DelHost(ctx, network.CIDR(), lease.ServiceNamespacedName(instance.ServiceSnapshot))
+			if err != nil {
+				log.Error("[endpoint] error deleting BGP host", "err", err)
+			} else {
+				log.Debug("[endpoint] deleted BGP host", "ip",
+					network.CIDR(), "service name", instance.ServiceSnapshot.Name, "namespace", instance.ServiceSnapshot.Namespace)
+			}
+		}
+	}
+}

pkg/endpoints/endpoints_generic.go

@@ -0,0 +1,109 @@
+package endpoints
+
+import (
+	"context"
+	"fmt"
+	log "log/slog"
+
+	"github.com/kube-vip/kube-vip/pkg/bgp"
+	"github.com/kube-vip/kube-vip/pkg/egress"
+	"github.com/kube-vip/kube-vip/pkg/endpoints/providers"
+	"github.com/kube-vip/kube-vip/pkg/instance"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/lease"
+	"github.com/kube-vip/kube-vip/pkg/route"
+	"github.com/kube-vip/kube-vip/pkg/servicecontext"
+	"github.com/kube-vip/kube-vip/pkg/wireguard"
+	v1 "k8s.io/api/core/v1"
+)
+
+type endpointWorker interface {
+	processInstance(svcCtx *servicecontext.Context, service *v1.Service) error
+	clear(svcCtx *servicecontext.Context, lastKnownGoodEndpoint *string, service *v1.Service)
+	getEndpoints(service *v1.Service, id string) ([]string, error)
+	removeEgress(service *v1.Service, lastKnownGoodEndpoint *string)
+	setInstanceEndpointsStatus(ctx context.Context, service *v1.Service, endpoints []string) error
+}
+
+func newEndpointWorker(config *kubevip.Config, provider providers.Provider, bgpServer *bgp.Server, instances *[]*instance.Instance,
+	leaseMgr *lease.Manager, tunnelMgr *wireguard.TunnelManager, routeMgr *route.Manager) endpointWorker {
+	generic := newGeneric(config, provider, instances, leaseMgr)
+
+	if config.EnableWireguard {
+		return newWireguardWorker(config, provider, bgpServer, instances, leaseMgr, tunnelMgr)
+	}
+	if config.EnableRoutingTable {
+		return newRoutingTable(generic, routeMgr)
+	}
+	if config.EnableBGP {
+		return newBGP(generic, bgpServer)
+	}
+
+	return &generic
+}
+
+type generic struct {
+	config    *kubevip.Config
+	provider  providers.Provider
+	instances *[]*instance.Instance
+	leaseMgr  *lease.Manager
+}
+
+func newGeneric(config *kubevip.Config, provider providers.Provider, instances *[]*instance.Instance, leaseMgr *lease.Manager) generic {
+	return generic{
+		config:    config,
+		provider:  provider,
+		instances: instances,
+		leaseMgr:  leaseMgr,
+	}
+}
+
+func (g *generic) processInstance(_ *servicecontext.Context, _ *v1.Service) error {
+	return nil
+}
+
+func (g *generic) clear(svcCtx *servicecontext.Context, lastKnownGoodEndpoint *string, service *v1.Service) {
+	g.clearEgress(lastKnownGoodEndpoint, service)
+	if svcCtx.LeaderCancel != nil {
+		svcCtx.LeaderCancel()
+	}
+}
+
+func (g *generic) clearEgress(lastKnownGoodEndpoint *string, service *v1.Service) {
+	if *lastKnownGoodEndpoint != "" {
+		log.Warn("existing endpoint has been removed, no remaining endpoints for leaderElection", "provider", g.provider.GetLabel(), "endpoint", lastKnownGoodEndpoint)
+		if err := egress.Teardown(*lastKnownGoodEndpoint, service.Spec.LoadBalancerIP, service.Namespace, string(service.UID), service.Annotations, g.config.EgressWithNftables); err != nil {
+			log.Error("error removing redundant egress rules", "err", err)
+		}
+
+		*lastKnownGoodEndpoint = "" // reset endpoint
+	}
+}
+
+func (g *generic) getEndpoints(service *v1.Service, id string) ([]string, error) {
+	return g.getAllEndpoints(service, id)
+}
+
+func (g *generic) getAllEndpoints(service *v1.Service, id string) ([]string, error) {
+	// Build endpoints
+	var err error
+	var endpoints []string
+	if service.Spec.ExternalTrafficPolicy == v1.ServiceExternalTrafficPolicyTypeCluster {
+		if endpoints, err = g.provider.GetAllEndpoints(); err != nil {
+			return nil, fmt.Errorf("[%s] error getting all endpoints: %w", g.provider.GetLabel(), err)
+		}
+	} else {
+		if endpoints, err = g.provider.GetLocalEndpoints(id, g.config); err != nil {
+			return nil, fmt.Errorf("[%s] error getting local endpoints: %w", g.provider.GetLabel(), err)
+		}
+	}
+
+	return endpoints, nil
+}
+
+func (g *generic) removeEgress(_ *v1.Service, _ *string) {
+}
+
+func (g *generic) setInstanceEndpointsStatus(_ context.Context, _ *v1.Service, _ []string) error {
+	return nil
+}

pkg/endpoints/endpoints_routing_table.go

@@ -0,0 +1,139 @@
+package endpoints
+
+import (
+	"context"
+	"fmt"
+	"net"
+	"sync"
+
+	log "log/slog"
+
+	"github.com/kube-vip/kube-vip/pkg/egress"
+	"github.com/kube-vip/kube-vip/pkg/instance"
+	"github.com/kube-vip/kube-vip/pkg/lease"
+	"github.com/kube-vip/kube-vip/pkg/route"
+	"github.com/kube-vip/kube-vip/pkg/servicecontext"
+	v1 "k8s.io/api/core/v1"
+)
+
+type RoutingTable struct {
+	generic
+	mtx      sync.Mutex
+	routeMgr *route.Manager
+}
+
+func newRoutingTable(generic generic, routeMgr *route.Manager) endpointWorker {
+	return &RoutingTable{
+		generic:  generic,
+		routeMgr: routeMgr,
+	}
+}
+
+func (rt *RoutingTable) processInstance(svcCtx *servicecontext.Context, service *v1.Service) error {
+	inst := instance.FindServiceInstance(service, *rt.instances)
+	if inst != nil {
+		for _, cluster := range inst.Clusters {
+			for i := range cluster.Network {
+				if !svcCtx.IsNetworkConfigured(cluster.Network[i].IP()) && cluster.Network[i].HasEndpoints() {
+					if err := rt.routeMgr.Add(lease.ServiceNamespacedName(service), cluster.Network[i], false, true); err != nil {
+						return fmt.Errorf("[%s] error adding route: %s", rt.provider.GetLabel(), err.Error())
+					} else {
+						log.Info("added route", "provider",
+							rt.provider.GetLabel(), "ip", cluster.Network[i].IP(), "service name", service.Name, "namespace",
+							service.Namespace, "interface", cluster.Network[i].Interface(), "tableID", rt.config.RoutingTableID)
+						svcCtx.ConfiguredNetworks.Store(cluster.Network[i].IP(), true)
+					}
+				}
+			}
+		}
+	}
+
+	return nil
+}
+
+func (rt *RoutingTable) clear(svcCtx *servicecontext.Context, lastKnownGoodEndpoint *string, service *v1.Service) {
+	rt.mtx.Lock()
+	defer rt.mtx.Unlock()
+	if !rt.config.EnableServicesElection {
+		if errs := ClearRoutes(service, rt.instances, rt.routeMgr); len(errs) == 0 {
+			svcCtx.ConfiguredNetworks.Clear()
+		} else {
+			for _, err := range errs {
+				log.Error("error while clearing routes", "err", err)
+			}
+		}
+	}
+
+	rt.clearEgress(lastKnownGoodEndpoint, service)
+
+	if svcCtx.LeaderCancel != nil {
+		svcCtx.LeaderCancel()
+	}
+}
+
+func (rt *RoutingTable) getEndpoints(service *v1.Service, id string) ([]string, error) {
+	return rt.getAllEndpoints(service, id)
+}
+
+func (rt *RoutingTable) removeEgress(service *v1.Service, lastKnownGoodEndpoint *string) {
+	if err := egress.Teardown(*lastKnownGoodEndpoint, service.Spec.LoadBalancerIP,
+		service.Namespace, string(service.UID), service.Annotations, rt.config.EgressWithNftables); err != nil {
+		log.Warn("removing redundant egress rules", "err", err)
+	}
+}
+
+func (rt *RoutingTable) setInstanceEndpointsStatus(ctx context.Context, service *v1.Service, endpoints []string) error {
+	inst := instance.FindServiceInstance(service, *rt.instances)
+	if inst == nil {
+		log.Error("failed to find the instance", "namespace", service.Namespace, "name", service.Name, "uid", service.UID, "provider", rt.provider.GetLabel())
+	} else {
+		for _, c := range inst.Clusters {
+			for n := range c.Network {
+				// if there are no endpoints set HasEndpoints false just in case
+				if len(endpoints) < 1 {
+					c.Network[n].SetHasEndpoints(false)
+				} else {
+					// check if endpoint are available and are of same IP family as service
+					for _, ep := range endpoints {
+						if (net.ParseIP(c.Network[n].IP()).To4() == nil) == (net.ParseIP(ep).To4() == nil) {
+							c.Network[n].SetHasEndpoints(true)
+							break
+						}
+					}
+				}
+			}
+		}
+	}
+
+	return nil
+}
+
+func ClearRoutes(service *v1.Service, instances *[]*instance.Instance, routeMgr *route.Manager) []error {
+	errs := []error{}
+	if svcInst := instance.FindServiceInstance(service, *instances); svcInst != nil {
+		clearErrs := ClearRoutesByInstance(service, svcInst, instances, routeMgr)
+		errs = append(errs, clearErrs...)
+	}
+	return errs
+}
+
+func ClearRoutesByInstance(service *v1.Service, svcInst *instance.Instance, instances *[]*instance.Instance, routeMgr *route.Manager) []error {
+	if svcInst == nil {
+		return []error{fmt.Errorf("failed to remove routes for nil instance of service %s/%s, uid: %s", service.Namespace, service.Name, service.UID)}
+	}
+	errs := []error{}
+	for _, cluster := range svcInst.Clusters {
+		for i := range cluster.Network {
+			err := routeMgr.Delete(lease.ServiceNamespacedName(service), cluster.Network[i])
+			if err != nil {
+				log.Error("failed to delete route", "ip", cluster.Network[i].IP(), "err", err)
+				errs = append(errs, err)
+			}
+			log.Debug("deleted route", "ip",
+				cluster.Network[i].IP(), "service name", service.Name, "namespace", service.Namespace, "interface", cluster.Network[i].Interface())
+
+		}
+	}
+
+	return errs
+}

pkg/endpoints/endpoints_test.go

@@ -0,0 +1,305 @@
+package endpoints
+
+import (
+	"context"
+	"sync"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"github.com/kube-vip/kube-vip/pkg/endpoints/providers"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/lease"
+	"github.com/kube-vip/kube-vip/pkg/metrics"
+	"github.com/kube-vip/kube-vip/pkg/servicecontext"
+	"github.com/prometheus/client_golang/prometheus/testutil"
+	v1 "k8s.io/api/core/v1"
+	discoveryv1 "k8s.io/api/discovery/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/watch"
+)
+
+func TestShouldAllowReconcileWithoutEndpoints(t *testing.T) {
+	if shouldAllowReconcileWithoutEndpoints(nil) {
+		t.Fatal("nil service should not be allowed")
+	}
+
+	clusterOptIn := &v1.Service{
+		Spec:       v1.ServiceSpec{ExternalTrafficPolicy: v1.ServiceExternalTrafficPolicyTypeCluster},
+		ObjectMeta: metav1.ObjectMeta{Annotations: map[string]string{kubevip.AllowReconcileWithoutEndpoints: "true"}},
+	}
+	if !shouldAllowReconcileWithoutEndpoints(clusterOptIn) {
+		t.Fatal("cluster service with opt-in annotation should be allowed")
+	}
+
+	localOptIn := &v1.Service{
+		Spec:       v1.ServiceSpec{ExternalTrafficPolicy: v1.ServiceExternalTrafficPolicyTypeLocal},
+		ObjectMeta: metav1.ObjectMeta{Annotations: map[string]string{kubevip.AllowReconcileWithoutEndpoints: "true"}},
+	}
+	if shouldAllowReconcileWithoutEndpoints(localOptIn) {
+		t.Fatal("local service should not be allowed")
+	}
+}
+
+type fakeWorker struct {
+	endpoints     []string
+	clearCalled   bool
+	processCalled bool
+}
+
+func (f *fakeWorker) processInstance(_ *servicecontext.Context, _ *v1.Service) error {
+	f.processCalled = true
+	return nil
+}
+
+func (f *fakeWorker) clear(_ *servicecontext.Context, _ *string, _ *v1.Service) {
+	f.clearCalled = true
+}
+
+func (f *fakeWorker) getEndpoints(_ *v1.Service, _ string) ([]string, error) { return f.endpoints, nil }
+func (f *fakeWorker) removeEgress(_ *v1.Service, _ *string)                  {}
+func (f *fakeWorker) setInstanceEndpointsStatus(_ context.Context, _ *v1.Service, _ []string) error {
+	return nil
+}
+
+// TestReconcile_RecomputesRemainingEndpoints asserts that deleting one EndpointSlice
+// reconciles against the endpoints that remain, instead of assuming the service
+// lost all of them.
+func TestReconcile_RecomputesRemainingEndpoints(t *testing.T) {
+	t.Parallel()
+
+	tests := []struct {
+		name              string
+		remaining         []string
+		lastKnown         string
+		expectReady       bool
+		expectClear       bool
+		expectProcess     bool
+		expectedLastKnown string
+	}{
+		{
+			name:              "remaining endpoints keep the service up",
+			remaining:         []string{"10.0.0.2"},
+			lastKnown:         "10.0.0.2",
+			expectReady:       true,
+			expectProcess:     true,
+			expectedLastKnown: "10.0.0.2",
+		},
+		{
+			name:              "stale last known endpoint moves to a survivor",
+			remaining:         []string{"10.0.0.2"},
+			lastKnown:         "10.0.0.1",
+			expectReady:       true,
+			expectProcess:     true,
+			expectedLastKnown: "10.0.0.2",
+		},
+		{
+			name:        "last endpoint removed tears the service down",
+			remaining:   nil,
+			lastKnown:   "10.0.0.1",
+			expectReady: false,
+			expectClear: true,
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			worker := &fakeWorker{endpoints: test.remaining}
+			p := &Processor{
+				config:   &kubevip.Config{},
+				provider: providers.NewEndpointslices(),
+				worker:   worker,
+			}
+
+			svcCtx := servicecontext.New(context.Background())
+			svcCtx.SignalReadiness()
+
+			lastKnown := test.lastKnown
+			restart, err := p.Reconcile(
+				svcCtx,
+				watch.Event{
+					Type:   watch.Deleted,
+					Object: &discoveryv1.EndpointSlice{ObjectMeta: metav1.ObjectMeta{Name: "slice-1"}},
+				},
+				&lastKnown,
+				&v1.Service{Spec: v1.ServiceSpec{ExternalTrafficPolicy: v1.ServiceExternalTrafficPolicyTypeLocal}},
+				"node-1",
+				func(*servicecontext.Context, *v1.Service, *sync.WaitGroup, bool) error { return nil },
+				&sync.WaitGroup{},
+				nil,
+				nil,
+			)
+			if err != nil {
+				t.Fatalf("Reconcile returned error: %v", err)
+			}
+			if restart {
+				t.Fatal("Reconcile unexpectedly requested restart")
+			}
+
+			if ready := svcCtx.Signalled.Load(); ready != test.expectReady {
+				t.Fatalf("readiness mismatch: expected %v, got %v", test.expectReady, ready)
+			}
+			if worker.clearCalled != test.expectClear {
+				t.Fatalf("clearCalled mismatch: expected %v, got %v", test.expectClear, worker.clearCalled)
+			}
+			if worker.processCalled != test.expectProcess {
+				t.Fatalf("processCalled mismatch: expected %v, got %v", test.expectProcess, worker.processCalled)
+			}
+			if test.expectedLastKnown != "" && lastKnown != test.expectedLastKnown {
+				t.Fatalf("lastKnownGoodEndpoint mismatch: expected %q, got %q", test.expectedLastKnown, lastKnown)
+			}
+		})
+	}
+}
+
+func TestReconcile_ZeroEndpointsBehavior(t *testing.T) {
+	t.Parallel()
+
+	run := func(t *testing.T, service *v1.Service, presetSignalled bool, expectReady bool, expectClear bool, expectProcess bool) {
+		t.Helper()
+
+		worker := &fakeWorker{endpoints: []string{}}
+		p := &Processor{
+			config:   &kubevip.Config{},
+			provider: providers.NewEndpointslices(),
+			worker:   worker,
+		}
+
+		svcCtx := servicecontext.New(context.Background())
+		if presetSignalled {
+			svcCtx.SignalReadiness()
+		}
+
+		restart, err := p.Reconcile(
+			svcCtx,
+			watch.Event{Type: watch.Modified, Object: &discoveryv1.EndpointSlice{}},
+			new(string),
+			service,
+			"node-1",
+			func(*servicecontext.Context, *v1.Service, *sync.WaitGroup, bool) error { return nil },
+			&sync.WaitGroup{},
+			nil,
+			nil,
+		)
+		if err != nil {
+			t.Fatalf("Reconcile returned error: %v", err)
+		}
+		if restart {
+			t.Fatal("Reconcile unexpectedly requested restart")
+		}
+
+		if ready := svcCtx.Signalled.Load(); ready != expectReady {
+			t.Fatalf("readiness mismatch: expected %v, got %v", expectReady, ready)
+		}
+		if worker.clearCalled != expectClear {
+			t.Fatalf("clearCalled mismatch: expected %v, got %v", expectClear, worker.clearCalled)
+		}
+		if worker.processCalled != expectProcess {
+			t.Fatalf("processCalled mismatch: expected %v, got %v", expectProcess, worker.processCalled)
+		}
+	}
+
+	t.Run("cluster opt-in keeps readiness and skips clear", func(t *testing.T) {
+		service := &v1.Service{
+			ObjectMeta: metav1.ObjectMeta{Annotations: map[string]string{kubevip.AllowReconcileWithoutEndpoints: "true"}},
+			Spec:       v1.ServiceSpec{ExternalTrafficPolicy: v1.ServiceExternalTrafficPolicyTypeCluster},
+		}
+		run(t, service, false, true, false, true)
+	})
+
+	t.Run("cluster without opt-in resets and clears when pre-signalled", func(t *testing.T) {
+		service := &v1.Service{
+			ObjectMeta: metav1.ObjectMeta{Annotations: map[string]string{}},
+			Spec:       v1.ServiceSpec{ExternalTrafficPolicy: v1.ServiceExternalTrafficPolicyTypeCluster},
+		}
+		run(t, service, true, false, true, false)
+	})
+
+	t.Run("local opt-in still resets and clears when pre-signalled", func(t *testing.T) {
+		service := &v1.Service{
+			ObjectMeta: metav1.ObjectMeta{Annotations: map[string]string{kubevip.AllowReconcileWithoutEndpoints: "true"}},
+			Spec:       v1.ServiceSpec{ExternalTrafficPolicy: v1.ServiceExternalTrafficPolicyTypeLocal},
+		}
+		run(t, service, true, false, true, false)
+	})
+}
+
+// TestReconcile_ServicesElectionStartsOnce asserts that repeated endpoint events
+// for the same service start the leader-election restart loop exactly once.
+//
+// Reconcile runs on every EndpointSlice add/modify/resync event, and the loop it
+// starts only returns once the service context is cancelled. Starting it per event
+// therefore accumulates duplicate goroutines that all contend on the same lease.
+//
+// See https://github.com/kube-vip/kube-vip/issues/1665.
+func TestReconcile_ServicesElectionStartsOnce(t *testing.T) {
+	config := &kubevip.Config{
+		EnableServicesElection: true,
+		LeaderElectionType:     "kubernetes",
+	}
+
+	service := &v1.Service{
+		ObjectMeta: metav1.ObjectMeta{Name: "test-svc", Namespace: "default", UID: "test-uid"},
+		Spec:       v1.ServiceSpec{Type: v1.ServiceTypeLoadBalancer},
+	}
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	leaseMgr := lease.NewManager()
+	leaseNamespace, serviceLease := lease.ServiceName(service)
+	svcLease := leaseMgr.Add(ctx, lease.NewID(config.LeaderElectionType, leaseNamespace, serviceLease))
+
+	svcCtx := servicecontext.New(svcLease.Ctx)
+
+	// The started loops only return once the service context is cancelled, so it has
+	// to be cancelled before waiting on them.
+	wg := &sync.WaitGroup{}
+	defer wg.Wait()
+	defer svcCtx.Cancel()
+
+	p := &Processor{
+		config:   config,
+		provider: providers.NewEndpointslices(),
+		worker:   &fakeWorker{endpoints: []string{"10.0.0.1"}},
+		leaseMgr: leaseMgr,
+	}
+
+	// starts counts the restart loops. The real StartServicesLeaderElection blocks
+	// until the service context is cancelled, so each loop parks in a single call.
+	var starts atomic.Int64
+	serviceFunc := func(svcCtx *servicecontext.Context, _ *v1.Service, _ *sync.WaitGroup, _ bool) error {
+		starts.Add(1)
+		<-svcCtx.Ctx.Done()
+		return nil
+	}
+
+	// Three endpoint events, as a flapping backend pod would produce.
+	for range 3 {
+		restart, err := p.Reconcile(svcCtx, watch.Event{Type: watch.Modified, Object: &discoveryv1.EndpointSlice{}},
+			new(string), service, "node-1", serviceFunc, wg, nil, nil)
+		if err != nil {
+			t.Fatalf("Reconcile returned error: %v", err)
+		}
+		if restart {
+			t.Fatal("Reconcile unexpectedly requested restart")
+		}
+	}
+
+	// Give every loop that is going to start a chance to reach serviceFunc.
+	for deadline := time.Now().Add(2 * time.Second); time.Now().Before(deadline); {
+		if starts.Load() > 1 {
+			break
+		}
+		time.Sleep(10 * time.Millisecond)
+	}
+
+	if got := starts.Load(); got != 1 {
+		t.Errorf("leader election started %d times, want 1", got)
+	}
+
+	// The gauge the e2e fault tests assert on has to agree with the call count.
+	if got := testutil.ToFloat64(metrics.ServiceElectionLoops.WithLabelValues(service.Namespace, service.Name)); got != 1 {
+		t.Errorf("kube_vip_service_election_loops is %v, want 1", got)
+	}
+}

pkg/endpoints/endpoints_wireguard.go

@@ -0,0 +1,281 @@
+package endpoints
+
+import (
+	"context"
+	"fmt"
+
+	log "log/slog"
+
+	"github.com/kube-vip/kube-vip/pkg/bgp"
+	"github.com/kube-vip/kube-vip/pkg/endpoints/providers"
+	"github.com/kube-vip/kube-vip/pkg/instance"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/lease"
+	"github.com/kube-vip/kube-vip/pkg/nftables"
+	"github.com/kube-vip/kube-vip/pkg/servicecontext"
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	"github.com/kube-vip/kube-vip/pkg/wireguard"
+	v1 "k8s.io/api/core/v1"
+)
+
+// wireguardWorker handles endpoint changes for WireGuard-based services
+type wireguardWorker struct {
+	config    *kubevip.Config
+	provider  providers.Provider
+	bgpServer *bgp.Server
+	instances *[]*instance.Instance
+	leaseMgr  *lease.Manager
+	tunnelMgr *wireguard.TunnelManager
+}
+
+func newWireguardWorker(config *kubevip.Config, provider providers.Provider, bgpServer *bgp.Server,
+	instances *[]*instance.Instance, leaseMgr *lease.Manager, tunnelMgr *wireguard.TunnelManager) *wireguardWorker {
+	return &wireguardWorker{
+		config:    config,
+		provider:  provider,
+		bgpServer: bgpServer,
+		instances: instances,
+		leaseMgr:  leaseMgr,
+		tunnelMgr: tunnelMgr,
+	}
+}
+
+// processInstance updates nftables DNAT rules when endpoints change
+// This is called by the endpoint watcher when endpoints are added/modified
+func (w *wireguardWorker) processInstance(svcCtx *servicecontext.Context, service *v1.Service) error {
+	log.Debug("[wireguard] processing instance for endpoint change", "service", service.Name, "namespace", service.Namespace)
+
+	// Get the target endpoint for this service
+	// For ExternalTrafficPolicy=Local, only use local endpoints
+	// For ExternalTrafficPolicy=Cluster, use all endpoints
+	var endpoints []string
+	var err error
+	if service.Spec.ExternalTrafficPolicy == v1.ServiceExternalTrafficPolicyTypeLocal {
+		endpoints, err = w.provider.GetLocalEndpoints(w.config.NodeName, w.config)
+	} else {
+		endpoints, err = w.provider.GetAllEndpoints()
+	}
+	if err != nil {
+		return fmt.Errorf("failed to get endpoints: %w", err)
+	}
+
+	if len(endpoints) == 0 {
+		log.Debug("[wireguard] no endpoints available", "service", service.Name)
+		w.clear(svcCtx, nil, service)
+		return nil
+	}
+
+	// Find the service processor to call updateServiceWireguardEndpoints
+	// Note: This requires access to the service processor which we don't have here
+	// So we'll recreate the DNAT rules directly
+
+	// First, clear existing rules
+	w.clear(svcCtx, nil, service)
+
+	// Get service VIPs
+	serviceIPs, err := utils.FetchServiceIPs(service)
+	if err != nil {
+		return fmt.Errorf("failed to get service IPs: %w", err)
+	}
+
+	// Create service identifier
+	serviceID := utils.SanitizeServiceID(fmt.Sprintf("%s_%s", service.Namespace, service.Name))
+
+	log.Info("[wireguard] updating DNAT rules for endpoint change",
+		"service", service.Name,
+		"namespace", service.Namespace,
+		"endpoints", endpoints,
+		"vips", serviceIPs)
+
+	// Update DNAT rules for each port
+	for _, port := range service.Spec.Ports {
+		// Determine target port (resolve named ports if necessary)
+		targetPort := w.provider.ResolvePort(port)
+		log.Info("[wireguard] resolved port", "service", service.Name, "servicePort", port.Port, "targetPort", targetPort, "targetPortName", port.TargetPort.StrVal)
+
+		// Build targets list from all endpoints
+		targets := make([]nftables.DNATTarget, len(endpoints))
+		for i, ep := range endpoints {
+			targets[i] = nftables.DNATTarget{
+				IP:   ep,
+				Port: uint16(targetPort), //nolint:gosec // Port range validated by Kubernetes
+			}
+		}
+
+		for _, vip := range serviceIPs {
+			// Strip CIDR notation if present
+			vipAddr := utils.StripCIDR(vip)
+
+			// Get WireGuard interface name from TunnelManager for this VIP
+			if w.tunnelMgr == nil {
+				log.Error("[wireguard] TunnelManager not configured; cannot update DNAT rules",
+					"service", service.Name,
+					"namespace", service.Namespace)
+				return fmt.Errorf("TunnelManager not configured")
+			}
+			tunnelConfig := w.tunnelMgr.GetConfigForVIP(vipAddr)
+			if tunnelConfig == nil {
+				log.Error("[wireguard] WireGuard interface name not configured; cannot update DNAT rules",
+					"service", service.Name,
+					"namespace", service.Namespace,
+					"vip", vipAddr)
+				return fmt.Errorf("wireguard interface name not configured for VIP %s", vipAddr)
+			}
+			wgInterface := tunnelConfig.InterfaceName
+
+			portServiceID := fmt.Sprintf("%s_p%d", serviceID, port.Port)
+
+			log.Info("[wireguard] applying DNAT rule with load balancing",
+				"service", service.Name,
+				"vip", vipAddr,
+				"interface", wgInterface,
+				"sourcePort", port.Port,
+				"targets", targets,
+				"chainID", portServiceID)
+
+			// Apply the DNAT rule with load balancing across all endpoints
+			// localEndpoint=true when using ExternalTrafficPolicy=Local, which preserves client source IP
+			isLocalEndpoint := service.Spec.ExternalTrafficPolicy == v1.ServiceExternalTrafficPolicyTypeLocal
+			err := nftables.ApplyDNAT(
+				wgInterface,
+				vipAddr,
+				uint16(port.Port), //nolint:gosec // Port range validated by Kubernetes
+				targets,
+				portServiceID,
+				port.Protocol,
+				isLocalEndpoint,
+				tunnelConfig.ListenPort,
+			)
+			if err != nil {
+				log.Error("[wireguard] failed to update DNAT rule",
+					"service", service.Name,
+					"vip", vipAddr,
+					"port", port.Port,
+					"err", err)
+				continue
+			}
+
+			log.Debug("[wireguard] DNAT rule updated successfully",
+				"service", service.Name,
+				"vip", vipAddr,
+				"port", port.Port,
+				"targetCount", len(targets))
+		}
+	}
+
+	return nil
+}
+
+// clear removes DNAT rules when no endpoints are available
+func (w *wireguardWorker) clear(svcCtx *servicecontext.Context, lastKnownGoodEndpoint *string, service *v1.Service) {
+	log.Info("[wireguard] clearing DNAT rules (no endpoints)", "service", service.Name, "namespace", service.Namespace)
+
+	serviceID := utils.SanitizeServiceID(fmt.Sprintf("%s_%s", service.Namespace, service.Name))
+
+	// Get service IPs to determine IPv4 vs IPv6
+	serviceIPs, _ := utils.FetchServiceIPs(service)
+
+	// Delete DNAT chains for each port
+	for _, port := range service.Spec.Ports {
+		if port.Protocol != v1.ProtocolTCP && port.Protocol != v1.ProtocolUDP {
+			continue
+		}
+
+		portServiceID := fmt.Sprintf("%s_p%d", serviceID, port.Port)
+
+		// Determine if we have IPv4 or IPv6
+		hasIPv4, hasIPv6 := false, false
+		for _, vip := range serviceIPs {
+			if isIPv6Address(vip) {
+				hasIPv6 = true
+			} else {
+				hasIPv4 = true
+			}
+		}
+
+		if hasIPv4 {
+			if err := nftables.DeleteIngressChains(false, portServiceID); err != nil {
+				log.Warn("[wireguard] failed to delete IPv4 DNAT chains",
+					"service", service.Name,
+					"port", port.Port,
+					"err", err)
+			}
+		}
+
+		if hasIPv6 {
+			if err := nftables.DeleteIngressChains(true, portServiceID); err != nil {
+				log.Warn("[wireguard] failed to delete IPv6 DNAT chains",
+					"service", service.Name,
+					"port", port.Port,
+					"err", err)
+			}
+		}
+	}
+
+	if svcCtx.LeaderCancel != nil {
+		svcCtx.LeaderCancel()
+	}
+}
+
+// getEndpoints retrieves the list of endpoints for a service
+// For ExternalTrafficPolicy=Local, only local endpoints are returned
+// For ExternalTrafficPolicy=Cluster, all endpoints are returned
+func (w *wireguardWorker) getEndpoints(service *v1.Service, id string) ([]string, error) {
+	var endpoints []string
+	var err error
+	if service.Spec.ExternalTrafficPolicy == v1.ServiceExternalTrafficPolicyTypeLocal {
+		endpoints, err = w.provider.GetLocalEndpoints(id, w.config)
+	} else {
+		endpoints, err = w.provider.GetAllEndpoints()
+	}
+	if err != nil {
+		return nil, fmt.Errorf("[wireguard] failed to get endpoints: %w", err)
+	}
+
+	log.Debug("[wireguard] retrieved endpoints", "service", service.Name, "count", len(endpoints), "endpoints", endpoints)
+	return endpoints, nil
+}
+
+// removeEgress is a no-op for WireGuard since egress is handled separately
+func (w *wireguardWorker) removeEgress(service *v1.Service, lastKnownGoodEndpoint *string) {
+	// WireGuard doesn't use egress in the same way as other modes
+	log.Debug("[wireguard] removeEgress called (no-op)", "service", service.Name)
+}
+
+// setInstanceEndpointsStatus updates the endpoint status on the service instance
+func (w *wireguardWorker) setInstanceEndpointsStatus(_ context.Context, service *v1.Service, endpoints []string) error {
+	hasEndpoints := len(endpoints) > 0
+
+	log.Debug("[wireguard] setting instance endpoint status",
+		"service", service.Name,
+		"hasEndpoints", hasEndpoints,
+		"endpointCount", len(endpoints))
+
+	// Find the service instance
+	for _, inst := range *w.instances {
+		if inst.ServiceSnapshot == nil {
+			continue
+		}
+		if inst.ServiceSnapshot.UID == service.UID {
+			// Update the network status for all clusters
+			for _, cluster := range inst.Clusters {
+				for i := range cluster.Network {
+					cluster.Network[i].SetHasEndpoints(hasEndpoints)
+				}
+			}
+			log.Debug("[wireguard] updated instance endpoint status",
+				"service", service.Name,
+				"hasEndpoints", hasEndpoints)
+			return nil
+		}
+	}
+
+	log.Debug("[wireguard] instance not found for endpoint status update", "service", service.Name)
+	return nil
+}
+
+func isIPv6Address(ip string) bool {
+	// Strip CIDR notation if present before checking
+	addr := utils.StripCIDR(ip)
+	return utils.IsIPv6(addr)
+}

pkg/endpoints/providers/endpoints.go

@@ -0,0 +1,158 @@
+package providers
+
+import (
+	"context"
+	"fmt"
+	"strings"
+
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	v1 "k8s.io/api/core/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/fields"
+	"k8s.io/apimachinery/pkg/runtime"
+	"k8s.io/apimachinery/pkg/watch"
+	"k8s.io/client-go/kubernetes"
+	"k8s.io/client-go/tools/cache"
+	watchtools "k8s.io/client-go/tools/watch"
+	"k8s.io/client-go/util/retry"
+
+	log "log/slog"
+)
+
+type Endpoints struct {
+	label string
+	//nolint:staticcheck // SA1019 endpoints are moving to an opt-in only
+	endpoints *v1.Endpoints
+}
+
+func NewEndpoints() Provider {
+	return &Endpoints{
+		label: "endpoints",
+	}
+}
+
+func (ep *Endpoints) CreateRetryWatcher(ctx context.Context, clientSet *kubernetes.Clientset,
+	service *v1.Service) (*watchtools.RetryWatcher, error) {
+	opts := metav1.ListOptions{
+		FieldSelector: fields.OneTermEqualSelector("metadata.name", service.Name).String(),
+	}
+
+	rw, err := watchtools.NewRetryWatcherWithContext(ctx, "1", &cache.ListWatch{
+		WatchFunc: func(_ metav1.ListOptions) (watch.Interface, error) {
+			return clientSet.CoreV1().Endpoints(service.Namespace).Watch(ctx, opts)
+		},
+	})
+
+	if err != nil {
+		return nil, fmt.Errorf("error creating endpoint watcher: %s", err.Error())
+	}
+
+	return rw, nil
+}
+
+func (ep *Endpoints) LoadObject(endpoints runtime.Object, cancel context.CancelFunc) error {
+	//nolint:staticcheck // SA1019 endpoints have to be explicitly requested now
+	eps, ok := endpoints.(*v1.Endpoints)
+	if !ok {
+		cancel()
+		return fmt.Errorf("[%s] unable to parse Kubernetes services from API watcher", ep.GetLabel())
+	}
+	ep.endpoints = eps
+	return nil
+}
+
+// DeleteObject drops the tracked object. A service is backed by exactly one
+// v1.Endpoints object, so there is nothing to match on and the cache is reset.
+func (ep *Endpoints) DeleteObject(endpoints runtime.Object) error {
+	//nolint:staticcheck // SA1019 endpoints have to be explicitly requested now
+	if _, ok := endpoints.(*v1.Endpoints); !ok {
+		return fmt.Errorf("[%s] unable to parse Kubernetes object", ep.GetLabel())
+	}
+	//nolint:staticcheck // SA1019 endpoints have to be explicitly requested now
+	ep.endpoints = &v1.Endpoints{}
+	return nil
+}
+
+func (ep *Endpoints) GetAllEndpoints() ([]string, error) {
+	result := []string{}
+	for subset := range ep.endpoints.Subsets {
+		for address := range ep.endpoints.Subsets[subset].Addresses {
+			addr := strings.Split(ep.endpoints.Subsets[subset].Addresses[address].IP, "/")
+			result = append(result, addr[0])
+		}
+	}
+
+	return result, nil
+}
+
+func (ep *Endpoints) GetLocalEndpoints(id string, _ *kubevip.Config) ([]string, error) {
+	var localEndpoints []string
+
+	for _, subset := range ep.endpoints.Subsets {
+		for _, address := range subset.Addresses {
+			log.Debug("processing endpoint", "label", ep.label, "ip", address.IP)
+
+			// 1. Compare the Nodename
+			if address.NodeName != nil && id == *address.NodeName {
+				log.Debug("found local endpoint", "label", ep.label, "ip", address.IP, "hostname", address.Hostname, "nodename", *address.NodeName)
+				localEndpoints = append(localEndpoints, address.IP)
+				continue
+			}
+			// 2. Compare the Hostname (only useful if address.NodeName is not available)
+			if id == address.Hostname {
+				log.Debug("found local endpoint", "label", ep.label, "ip", address.IP, "hostname", address.Hostname)
+				localEndpoints = append(localEndpoints, address.IP)
+				continue
+			}
+		}
+	}
+	return localEndpoints, nil
+}
+
+func (ep *Endpoints) UpdateServiceAnnotation(ctx context.Context, endpoint string, _ string, service *v1.Service, clientSet *kubernetes.Clientset) error {
+	retryErr := retry.RetryOnConflict(retry.DefaultRetry, func() error {
+		// Retrieve the latest version of Deployment before attempting update
+		// RetryOnConflict uses exponential backoff to avoid exhausting the apiserver
+		currentService, err := clientSet.CoreV1().Services(service.Namespace).Get(ctx, service.Name, metav1.GetOptions{})
+		if err != nil {
+			return err
+		}
+
+		currentServiceCopy := currentService.DeepCopy()
+		if currentServiceCopy.Annotations == nil {
+			currentServiceCopy.Annotations = make(map[string]string)
+		}
+
+		currentServiceCopy.Annotations[kubevip.ActiveEndpoint] = endpoint
+
+		_, err = clientSet.CoreV1().Services(currentService.Namespace).Update(ctx, currentServiceCopy, metav1.UpdateOptions{})
+		if err != nil {
+			log.Error("error updating Service Spec", "label", ep.GetLabel(), "name", currentServiceCopy.Name, "err", err)
+			return err
+		}
+		return nil
+	})
+
+	if retryErr != nil {
+		log.Error("failed to set Services", "label", ep.GetLabel(), "err", retryErr)
+		return retryErr
+	}
+	return nil
+}
+
+func (ep *Endpoints) GetLabel() string {
+	return ep.label
+}
+
+func (ep *Endpoints) ResolvePort(servicePort v1.ServicePort) int32 {
+	return ResolvePortWithLookup(servicePort, func(name string) int32 {
+		for _, subset := range ep.endpoints.Subsets {
+			for _, p := range subset.Ports {
+				if p.Name == name {
+					return p.Port
+				}
+			}
+		}
+		return 0
+	})
+}

pkg/endpoints/providers/endpointslices.go

@@ -0,0 +1,179 @@
+package providers
+
+import (
+	"context"
+	"fmt"
+
+	log "log/slog"
+
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	v1 "k8s.io/api/core/v1"
+	discoveryv1 "k8s.io/api/discovery/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/labels"
+	"k8s.io/apimachinery/pkg/runtime"
+	"k8s.io/apimachinery/pkg/watch"
+	"k8s.io/client-go/kubernetes"
+	"k8s.io/client-go/tools/cache"
+	watchtools "k8s.io/client-go/tools/watch"
+	"k8s.io/client-go/util/retry"
+)
+
+type Endpointslices struct {
+	label  string
+	slices map[string]*discoveryv1.EndpointSlice
+}
+
+func NewEndpointslices() Provider {
+	return &Endpointslices{
+		label:  "endpointslices",
+		slices: make(map[string]*discoveryv1.EndpointSlice),
+	}
+}
+
+func (ep *Endpointslices) CreateRetryWatcher(ctx context.Context, clientSet *kubernetes.Clientset,
+	service *v1.Service) (*watchtools.RetryWatcher, error) {
+	labelSelector := metav1.LabelSelector{MatchLabels: map[string]string{"kubernetes.io/service-name": service.Name}}
+
+	opts := metav1.ListOptions{
+		LabelSelector: labels.Set(labelSelector.MatchLabels).String(),
+	}
+
+	rw, err := watchtools.NewRetryWatcherWithContext(ctx, "1", &cache.ListWatch{
+		WatchFunc: func(_ metav1.ListOptions) (watch.Interface, error) {
+			return clientSet.DiscoveryV1().EndpointSlices(service.Namespace).Watch(ctx, opts)
+		},
+	})
+	if err != nil {
+		return nil, fmt.Errorf("[%s] error creating endpointslices watcher: %s", ep.label, err.Error())
+	}
+
+	return rw, nil
+}
+
+func (ep *Endpointslices) LoadObject(endpoints runtime.Object, cancel context.CancelFunc) error {
+	eps, ok := endpoints.(*discoveryv1.EndpointSlice)
+	if !ok {
+		cancel()
+		return fmt.Errorf("[%s] error casting endpoints to v1.Endpoints struct", ep.label)
+	}
+
+	if ep.slices == nil {
+		ep.slices = make(map[string]*discoveryv1.EndpointSlice)
+	}
+	ep.slices[eps.Name] = eps.DeepCopy()
+
+	return nil
+}
+
+func (ep *Endpointslices) DeleteObject(endpoints runtime.Object) error {
+	eps, ok := endpoints.(*discoveryv1.EndpointSlice)
+	if !ok {
+		return fmt.Errorf("[%s] unable to parse Kubernetes object", ep.GetLabel())
+	}
+	delete(ep.slices, eps.Name)
+	return nil
+}
+
+// isServing reports whether an endpoint should receive traffic. Per the
+// EndpointConditions godoc a nil Serving defers to Ready, and a nil Ready is an
+// unknown state that consumers should interpret as ready.
+func isServing(conditions discoveryv1.EndpointConditions) bool {
+	serving := conditions.Serving
+	if serving == nil {
+		serving = conditions.Ready
+	}
+	return serving == nil || *serving
+}
+
+func (ep *Endpointslices) GetAllEndpoints() ([]string, error) {
+	result := []string{}
+	for _, eps := range ep.slices {
+		for _, e := range eps.Endpoints {
+			if !isServing(e.Conditions) {
+				continue
+			}
+			result = append(result, e.Addresses...)
+		}
+	}
+	return result, nil
+}
+
+func (ep *Endpointslices) GetLocalEndpoints(id string, _ *kubevip.Config) ([]string, error) {
+	var localEndpoints []string
+	for _, eps := range ep.slices {
+		for _, endpoint := range eps.Endpoints {
+			if !isServing(endpoint.Conditions) {
+				continue
+			}
+			for _, address := range endpoint.Addresses {
+				// 1. Compare the Nodename
+				if endpoint.NodeName != nil && id == *endpoint.NodeName {
+					if endpoint.Hostname != nil {
+						log.Debug("found endpoint", "provider", ep.label, "ip", address, "hostname", *endpoint.Hostname, "nodename", *endpoint.NodeName)
+					} else {
+						log.Debug("found endpoint", "provider", ep.label, "ip", address, "nodename", *endpoint.NodeName)
+					}
+					localEndpoints = append(localEndpoints, address)
+					continue
+				}
+
+				// 2. Compare the Hostname (only useful if endpoint.NodeName is not available)
+				if endpoint.Hostname != nil && id == *endpoint.Hostname {
+					log.Debug("found endpoint", "provider", ep.label, "ip", address, "hostname", *endpoint.Hostname)
+					localEndpoints = append(localEndpoints, address)
+				}
+			}
+		}
+	}
+	return localEndpoints, nil
+}
+
+func (ep *Endpointslices) UpdateServiceAnnotation(ctx context.Context, endpoint, endpointIPv6 string, service *v1.Service, clientSet *kubernetes.Clientset) error {
+	retryErr := retry.RetryOnConflict(retry.DefaultRetry, func() error {
+		// Retrieve the latest version of Deployment before attempting update
+		// RetryOnConflict uses exponential backoff to avoid exhausting the apiserver
+		currentService, err := clientSet.CoreV1().Services(service.Namespace).Get(ctx, service.Name, metav1.GetOptions{})
+		if err != nil {
+			return err
+		}
+
+		currentServiceCopy := currentService.DeepCopy()
+		if currentServiceCopy.Annotations == nil {
+			currentServiceCopy.Annotations = make(map[string]string)
+		}
+
+		currentServiceCopy.Annotations[kubevip.ActiveEndpoint] = endpoint
+		currentServiceCopy.Annotations[kubevip.ActiveEndpointIPv6] = endpointIPv6
+
+		_, err = clientSet.CoreV1().Services(currentService.Namespace).Update(ctx, currentServiceCopy, metav1.UpdateOptions{})
+		if err != nil {
+			log.Error("error updating Service Spec", "provider", ep.label, "service name", currentServiceCopy.Name, "err", err)
+			return err
+		}
+		return nil
+	})
+
+	if retryErr != nil {
+		log.Error("failed to set Services", "provider", ep.label, "err", retryErr)
+		return retryErr
+	}
+	return nil
+}
+
+func (ep *Endpointslices) GetLabel() string {
+	return ep.label
+}
+
+func (ep *Endpointslices) ResolvePort(servicePort v1.ServicePort) int32 {
+	return ResolvePortWithLookup(servicePort, func(name string) int32 {
+		for _, eps := range ep.slices {
+			for _, p := range eps.Ports {
+				if p.Name != nil && *p.Name == name && p.Port != nil {
+					return *p.Port
+				}
+			}
+		}
+		return 0
+	})
+}

pkg/endpoints/providers/endpointslices_test.go

@@ -0,0 +1,149 @@
+package providers
+
+import (
+	"context"
+	"testing"
+
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	discoveryv1 "k8s.io/api/discovery/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+)
+
+func TestEndpointslicesTracksAndDeletesSlices(t *testing.T) {
+	provider := NewEndpointslices().(*Endpointslices)
+	serving := true
+	nodeName := "node-1"
+
+	slice1 := &discoveryv1.EndpointSlice{
+		ObjectMeta:  metav1.ObjectMeta{Name: "slice-1"},
+		AddressType: discoveryv1.AddressTypeIPv4,
+		Endpoints: []discoveryv1.Endpoint{{
+			Addresses:  []string{"10.0.0.1"},
+			Conditions: discoveryv1.EndpointConditions{Serving: &serving},
+			NodeName:   &nodeName,
+		}},
+	}
+	slice2 := &discoveryv1.EndpointSlice{
+		ObjectMeta:  metav1.ObjectMeta{Name: "slice-2"},
+		AddressType: discoveryv1.AddressTypeIPv4,
+		Endpoints: []discoveryv1.Endpoint{{
+			Addresses:  []string{"10.0.0.2"},
+			Conditions: discoveryv1.EndpointConditions{Serving: &serving},
+			NodeName:   &nodeName,
+		}},
+	}
+
+	for _, slice := range []*discoveryv1.EndpointSlice{slice1, slice2} {
+		if err := provider.LoadObject(slice, func() {}); err != nil {
+			t.Fatalf("LoadObject returned error: %v", err)
+		}
+	}
+
+	assertEndpoints(t, provider, []string{"10.0.0.1", "10.0.0.2"})
+	assertLocalEndpoints(t, provider, nodeName, []string{"10.0.0.1", "10.0.0.2"})
+
+	if err := provider.DeleteObject(slice1); err != nil {
+		t.Fatalf("DeleteObject returned error: %v", err)
+	}
+	assertEndpoints(t, provider, []string{"10.0.0.2"})
+	assertLocalEndpoints(t, provider, nodeName, []string{"10.0.0.2"})
+
+	if err := provider.DeleteObject(slice2); err != nil {
+		t.Fatalf("DeleteObject returned error: %v", err)
+	}
+	assertEndpoints(t, provider, nil)
+	assertLocalEndpoints(t, provider, nodeName, nil)
+}
+
+func TestEndpointslicesReplacingSliceUpdatesState(t *testing.T) {
+	provider := NewEndpointslices().(*Endpointslices)
+	first := &discoveryv1.EndpointSlice{
+		ObjectMeta:  metav1.ObjectMeta{Name: "slice-1"},
+		AddressType: discoveryv1.AddressTypeIPv4,
+		Endpoints:   []discoveryv1.Endpoint{{Addresses: []string{"10.0.0.1"}}},
+	}
+	replacement := first.DeepCopy()
+	replacement.Endpoints[0].Addresses = []string{"10.0.0.2"}
+
+	if err := provider.LoadObject(first, context.CancelFunc(func() {})); err != nil {
+		t.Fatalf("LoadObject returned error: %v", err)
+	}
+	if err := provider.LoadObject(replacement, context.CancelFunc(func() {})); err != nil {
+		t.Fatalf("LoadObject returned error: %v", err)
+	}
+
+	assertEndpoints(t, provider, []string{"10.0.0.2"})
+}
+
+func TestEndpointslicesEndpointConditions(t *testing.T) {
+	yes, no := true, false
+	nodeName := "node-1"
+
+	tests := []struct {
+		name       string
+		conditions discoveryv1.EndpointConditions
+		want       []string
+	}{
+		{"serving true", discoveryv1.EndpointConditions{Serving: &yes}, []string{"10.0.0.1"}},
+		{"serving false", discoveryv1.EndpointConditions{Serving: &no}, nil},
+		{"serving false overrides ready true", discoveryv1.EndpointConditions{Serving: &no, Ready: &yes}, nil},
+		{"nil serving defers to ready true", discoveryv1.EndpointConditions{Ready: &yes}, []string{"10.0.0.1"}},
+		{"nil serving defers to ready false", discoveryv1.EndpointConditions{Ready: &no}, nil},
+		{"both nil is treated as ready", discoveryv1.EndpointConditions{}, []string{"10.0.0.1"}},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			provider := NewEndpointslices().(*Endpointslices)
+			slice := &discoveryv1.EndpointSlice{
+				ObjectMeta:  metav1.ObjectMeta{Name: "slice-1"},
+				AddressType: discoveryv1.AddressTypeIPv4,
+				Endpoints: []discoveryv1.Endpoint{{
+					Addresses:  []string{"10.0.0.1"},
+					Conditions: test.conditions,
+					NodeName:   &nodeName,
+				}},
+			}
+			if err := provider.LoadObject(slice, func() {}); err != nil {
+				t.Fatalf("LoadObject returned error: %v", err)
+			}
+			// Cluster and Local policy have to agree on which endpoints are usable.
+			assertEndpoints(t, provider, test.want)
+			assertLocalEndpoints(t, provider, nodeName, test.want)
+		})
+	}
+}
+
+func assertEndpoints(t *testing.T, provider *Endpointslices, want []string) {
+	t.Helper()
+	got, err := provider.GetAllEndpoints()
+	if err != nil {
+		t.Fatalf("GetAllEndpoints returned error: %v", err)
+	}
+	assertStringSet(t, got, want)
+}
+
+func assertLocalEndpoints(t *testing.T, provider *Endpointslices, nodeName string, want []string) {
+	t.Helper()
+	got, err := provider.GetLocalEndpoints(nodeName, &kubevip.Config{})
+	if err != nil {
+		t.Fatalf("GetLocalEndpoints returned error: %v", err)
+	}
+	assertStringSet(t, got, want)
+}
+
+func assertStringSet(t *testing.T, got, want []string) {
+	t.Helper()
+	counts := map[string]int{}
+	for _, value := range got {
+		counts[value]++
+	}
+	for _, value := range want {
+		counts[value]--
+	}
+	for value, count := range counts {
+		if count != 0 {
+			t.Fatalf("endpoint set mismatch for %q: got %v, want %v", value, got, want)
+		}
+	}
+}

pkg/endpoints/providers/interface.go

@@ -0,0 +1,40 @@
+package providers
+
+import (
+	"context"
+
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	v1 "k8s.io/api/core/v1"
+	"k8s.io/apimachinery/pkg/runtime"
+	"k8s.io/client-go/kubernetes"
+	watchtools "k8s.io/client-go/tools/watch"
+)
+
+type Provider interface {
+	CreateRetryWatcher(context.Context, *kubernetes.Clientset,
+		*v1.Service) (*watchtools.RetryWatcher, error)
+	GetAllEndpoints() ([]string, error)
+	GetLocalEndpoints(string, *kubevip.Config) ([]string, error)
+	GetLabel() string
+	UpdateServiceAnnotation(context.Context, string, string, *v1.Service, *kubernetes.Clientset) error
+	LoadObject(runtime.Object, context.CancelFunc) error
+	DeleteObject(runtime.Object) error
+	// ResolvePort resolves a service port to the actual target port.
+	// For named ports, it looks up the port number from the endpoint.
+	// For numeric ports, it returns the port as-is.
+	ResolvePort(servicePort v1.ServicePort) int32
+}
+
+// ResolvePortWithLookup is a helper that resolves a service port using a lookup function
+// for named ports. This consolidates the common resolution logic.
+func ResolvePortWithLookup(servicePort v1.ServicePort, lookupNamedPort func(string) int32) int32 {
+	if servicePort.TargetPort.IntVal != 0 {
+		return servicePort.TargetPort.IntVal
+	}
+	if servicePort.TargetPort.StrVal != "" {
+		if port := lookupNamedPort(servicePort.TargetPort.StrVal); port != 0 {
+			return port
+		}
+	}
+	return servicePort.Port
+}

pkg/equinixmetal/bgp.go

@@ -1,63 +0,0 @@
-package equinixmetal
-
-import (
-	"fmt"
-
-	"github.com/kube-vip/kube-vip/pkg/bgp"
-	"github.com/kube-vip/kube-vip/pkg/kubevip"
-	"github.com/packethost/packngo"
-	log "github.com/sirupsen/logrus"
-)
-
-// BGPLookup will use the Equinix Metal API functions to populate the BGP information
-func BGPLookup(c *packngo.Client, k *kubevip.Config) error {
-	var thisDevice *packngo.Device
-	if k.MetalProjectID == "" {
-		proj := findProject(k.MetalProject, c)
-		if proj == nil {
-			return fmt.Errorf("Unable to find Project [%s]", k.MetalProject)
-		}
-		thisDevice = findSelf(c, proj.ID)
-	} else {
-		thisDevice = findSelf(c, k.MetalProjectID)
-	}
-	if thisDevice == nil {
-		return fmt.Errorf("Unable to find local/this device in Equinix Metal API")
-	}
-
-	log.Infof("Querying BGP settings for [%s]", thisDevice.Hostname)
-	neighbours, _, err := c.Devices.ListBGPNeighbors(thisDevice.ID, &packngo.ListOptions{})
-	if err != nil {
-		return err
-	}
-	// Ensure neighbours exist (and it's enabled)
-	if len(neighbours) == 0 {
-		return fmt.Errorf("The server [%s]/[%s] has no BGP neighbours, ensure BGP is enabled", thisDevice.Hostname, thisDevice.ID)
-	}
-
-	// Add a warning (TODO)
-	if len(neighbours) > 1 {
-		log.Warnf("There are [%d] neighbours, only designed to manage one", len(neighbours))
-	}
-
-	// Ensure a peer exists
-	if len(neighbours[0].PeerIps) == 0 {
-		return fmt.Errorf("The server [%s]/[%s] has no BGP peers, ensure BGP is enabled", thisDevice.Hostname, thisDevice.ID)
-	}
-
-	k.BGPConfig.RouterID = neighbours[0].CustomerIP
-	k.BGPConfig.AS = uint32(neighbours[0].CustomerAs)
-
-	// Add the peer(s)
-	for x := range neighbours[0].PeerIps {
-		peer := bgp.Peer{
-			Address:  neighbours[0].PeerIps[x],
-			AS:       uint32(neighbours[0].PeerAs),
-			MultiHop: neighbours[0].Multihop,
-			Password: neighbours[0].Md5Password,
-		}
-		k.BGPConfig.Peers = append(k.BGPConfig.Peers, peer)
-	}
-
-	return nil
-}

pkg/equinixmetal/eip.go

@@ -1,65 +0,0 @@
-package equinixmetal
-
-import (
-	"fmt"
-	"path"
-
-	"github.com/kube-vip/kube-vip/pkg/kubevip"
-	"github.com/packethost/packngo"
-	log "github.com/sirupsen/logrus"
-)
-
-// AttachEIP will use the Equinix Metal APIs to move an EIP and attach to a host
-func AttachEIP(c *packngo.Client, k *kubevip.Config, _ string) error {
-	// Use MetalProjectID if it is defined
-	projID := k.MetalProjectID
-
-	if projID == "" {
-		// Fallback to attempting to find the project by name
-		proj := findProject(k.MetalProject, c)
-		if proj == nil {
-			return fmt.Errorf("unable to find Project [%s]", k.MetalProject)
-		}
-
-		projID = proj.ID
-	}
-
-	// Prefer Address over VIP
-	vip := k.Address
-	if vip == "" {
-		vip = k.VIP
-	}
-
-	ips, _, _ := c.ProjectIPs.List(projID, &packngo.ListOptions{})
-	for _, ip := range ips {
-		// Find the device id for our EIP
-		if ip.Address == vip {
-			log.Infof("Found EIP ->%s ID -> %s\n", ip.Address, ip.ID)
-			// If attachments already exist then remove them
-			if len(ip.Assignments) != 0 {
-				hrefID := path.Base(ip.Assignments[0].Href)
-				_, err := c.DeviceIPs.Unassign(hrefID)
-				if err != nil {
-					return fmt.Errorf("unable to unassign deviceIP %q: %v", hrefID, err)
-				}
-			}
-		}
-	}
-
-	// Lookup this server through the Equinix Metal API
-	thisDevice := findSelf(c, projID)
-	if thisDevice == nil {
-		return fmt.Errorf("unable to find local/this device in Equinix Metal API")
-	}
-
-	// Assign the EIP to this device
-	log.Infof("Assigning EIP to -> %s\n", thisDevice.Hostname)
-	_, _, err := c.DeviceIPs.Assign(thisDevice.ID, &packngo.AddressStruct{
-		Address: vip,
-	})
-	if err != nil {
-		return err
-	}
-
-	return nil
-}

pkg/equinixmetal/utils.go

@@ -1,59 +0,0 @@
-package equinixmetal
-
-import (
-	"encoding/json"
-	"fmt"
-	"os"
-
-	"github.com/packethost/packngo"
-	log "github.com/sirupsen/logrus"
-)
-
-func findProject(project string, c *packngo.Client) *packngo.Project {
-	l := &packngo.ListOptions{Includes: []string{project}}
-	ps, _, err := c.Projects.List(l)
-	if err != nil {
-		log.Error(err)
-	}
-	for _, p := range ps {
-
-		// Find our project
-		if p.Name == project {
-			return &p
-		}
-	}
-	return nil
-}
-
-func findSelf(c *packngo.Client, projectID string) *packngo.Device {
-	// Go through devices
-	dev, _, _ := c.Devices.List(projectID, &packngo.ListOptions{})
-	for _, d := range dev {
-		// TODO do we need to replace os.Hostname with config.NodeName here?
-		me, _ := os.Hostname()
-		if me == d.Hostname {
-			return &d
-		}
-	}
-	return nil
-}
-
-// GetPacketConfig will lookup the configuration from a file path
-func GetPacketConfig(providerConfig string) (string, string, error) {
-	var config struct {
-		AuthToken string `json:"apiKey"`
-		ProjectID string `json:"projectId"`
-	}
-	// get our token and project
-	if providerConfig != "" {
-		configBytes, err := os.ReadFile(providerConfig)
-		if err != nil {
-			return "", "", fmt.Errorf("failed to get read configuration file at path %s: %v", providerConfig, err)
-		}
-		err = json.Unmarshal(configBytes, &config)
-		if err != nil {
-			return "", "", fmt.Errorf("failed to process json of configuration file at path %s: %v", providerConfig, err)
-		}
-	}
-	return config.AuthToken, config.ProjectID, nil
-}

pkg/etcd/election.go

@@ -2,11 +2,14 @@ package etcd
 
 import (
 	"context"
+	"fmt"
 	"hash/fnv"
+	"sync"
 	"time"
 
+	log "log/slog"
+
 	"github.com/pkg/errors"
-	log "github.com/sirupsen/logrus"
 	pb "go.etcd.io/etcd/api/v3/etcdserverpb"
 	clientv3 "go.etcd.io/etcd/client/v3"
 	"go.etcd.io/etcd/client/v3/concurrency"
@@ -31,7 +34,7 @@ type LeaderElectionConfig struct {
 
 	// MemberUniqueID is the int equivalent to MemberID that allows to override the default conversion
 	// from string to int using hashing.
-	MemberUniqueID *int64
+	MemberUniqueID *uint64
 
 	// LeaseDurationSeconds is the duration that non-leader candidates will
 	// wait to force acquire leadership.
@@ -63,17 +66,18 @@ type ClientConfig struct {
 }
 
 // RunElectionOrDie behaves the same way as RunElection but panics if there is an error.
-func RunElectionOrDie(ctx context.Context, config *LeaderElectionConfig) {
+func RunElectionOrDie(ctx context.Context, config *LeaderElectionConfig) error {
 	if err := RunElection(ctx, config); err != nil {
-		panic(err)
+		return fmt.Errorf("leaderelection error: %w", err)
 	}
+	return nil
 }
 
 // RunElection starts a client with the provided config or panics.
 // RunElection blocks until leader election loop is
 // stopped by ctx or it has stopped holding the leader lease.
 func RunElection(ctx context.Context, config *LeaderElectionConfig) error {
-	var memberID int64
+	var memberID uint64
 	if config.MemberUniqueID != nil {
 		memberID = *config.MemberUniqueID
 	} else {
@@ -81,11 +85,11 @@ func RunElection(ctx context.Context, config *LeaderElectionConfig) error {
 		if _, err := h.Write(append([]byte(config.Name), []byte(config.MemberID)...)); err != nil {
 			return err
 		}
-		memberID = int64(h.Sum64())
+		memberID = h.Sum64()
 	}
 
 	ttl := config.LeaseDurationSeconds
-	r := &pb.LeaseGrantRequest{TTL: ttl, ID: memberID}
+	r := &pb.LeaseGrantRequest{TTL: ttl, ID: int64(memberID)} //nolint
 	lease, err := clientv3.RetryLeaseClient(
 		config.EtcdConfig.Client,
 	).LeaseGrant(ctx, r)
@@ -115,7 +119,12 @@ func RunElection(ctx context.Context, config *LeaderElectionConfig) error {
 		leaseTTL:       lease.TTL,
 	}
 
-	go m.tryToBeLeader(ctx)
+	wg := sync.WaitGroup{}
+	defer wg.Wait()
+
+	wg.Go(func() {
+		m.tryToBeLeader(ctx, &wg)
+	})
 	m.watchLeaderChanges(ctx)
 
 	return nil
@@ -147,9 +156,9 @@ watcher:
 
 			m.isLeader = true
 			m.key = m.election.Key() // by this time, this should already be set, since Campaign has already returned
-			log.Debugf("[%s] Marking self as leader with key %s\n", m.memberID, m.key)
+			log.Debug("Marking self as leader with key", "id", m.memberID, "key", m.key)
 		case response := <-changes:
-			log.Debugf("[%s] Leader Changes: %+v\n", m.memberID, response)
+			log.Debug("Leader Changes", "id", m.memberID, "response", response)
 			if len(response.Kvs) == 0 {
 				// There is a race condition where just after we stop being the leader
 				// if there are no more leaders, we might get a response with no key-values
@@ -181,15 +190,15 @@ watcher:
 		m.callbacks.OnStoppedLeading()
 	}
 
-	log.Debugf("[%s] Exiting watcher\n", m.memberID)
+	log.Debug("Exiting watcher", "id", m.memberID)
 }
 
-func (m *member) tryToBeLeader(ctx context.Context) {
+func (m *member) tryToBeLeader(ctx context.Context, wg *sync.WaitGroup) {
 	if err := m.election.Campaign(ctx, m.memberID); err != nil {
-		log.Errorf("Failed trying to become the leader: %s", err)
+		log.Error("Failed trying to become the leader", "err", err)
 		// Resign just in case we acquired leadership just before failing
 		if err := m.election.Resign(m.client.Ctx()); err != nil {
-			log.Warnf("Failed to resign after we failed becoming the leader, this might not be a problem if we were never the leader: %s", err)
+			log.Warn("Failed to resign after we failed becoming the leader, this might not be a problem if we were never the leader", "err", err)
 		}
 		return
 		// TODO: what to do here?
@@ -203,13 +212,15 @@ func (m *member) tryToBeLeader(ctx context.Context) {
 	m.weAreTheLeader <- struct{}{}
 
 	// Once we are the leader, start the routine to resign if context is canceled
-	go m.resignOnCancel(ctx)
+	wg.Go(func() {
+		m.resignOnCancel(ctx)
+	})
 
 	// After becoming the leader, we wait for at least a lease TTL to wait for
 	// the previous leader to detect the new leadership (if there was one) and
 	// stop its processes
 	// TODO: is this too cautious?
-	log.Debugf("[%s] Waiting %d seconds before running OnStartedLeading", m.memberID, m.leaseTTL)
+	log.Debug("timeout before OnStartedLeading", "id", m.memberID, "timeout", m.leaseTTL)
 	time.Sleep(time.Second * time.Duration(m.leaseTTL))
 
 	// We are the leader, execute our code
@@ -220,7 +231,7 @@ func (m *member) tryToBeLeader(ctx context.Context) {
 
 func (m *member) resignOnCancel(ctx context.Context) {
 	<-ctx.Done()
-	if err := m.election.Resign(m.client.Ctx()); err != nil {
-		log.Errorf("Failed to resign after the context was canceled: %s", err)
+	if err := m.election.Resign(m.client.Ctx()); err != nil && !errors.Is(err, context.Canceled) {
+		log.Error("Failed to resign after the context was canceled", "err", err)
 	}
 }

pkg/etcd/election_test.go

@@ -20,19 +20,26 @@ import (
 func TestRunElectionWithMemberIDCollision(t *testing.T) {
 	t.Parallel()
 	g := NewWithT(t)
-	ctx := context.Background()
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
 	cli := client(g)
 	defer cli.Close()
 
 	electionName := randomElectionNameForTest("memberIDConflict")
 	log.Printf("Election name %s\n", electionName)
 	memberCtx, cancelMember1 := context.WithCancel(ctx)
+
+	// Use a channel to signal when the first member has observed a new leader
+	// This ensures proper ordering without relying on sleep timing
+	firstMemberObservedLeader := make(chan struct{})
+	var firstMemberObservedOnce sync.Once
+
 	config := &etcd.LeaderElectionConfig{
 		EtcdConfig: etcd.ClientConfig{
 			Client: cli,
 		},
 		Name:                 electionName,
-		MemberID:             "my-host",
+		MemberID:             randomElectionNameForTest("my-host"),
 		LeaseDurationSeconds: 1,
 		Callbacks: etcd.LeaderCallbacks{
 			OnStartedLeading: func(ctx context.Context) {
@@ -42,6 +49,11 @@ func TestRunElectionWithMemberIDCollision(t *testing.T) {
 			},
 			OnNewLeader: func(identity string) {
 				log.Printf("New leader: %s\n", identity)
+				// Signal that the first member has observed a leader
+				// This means the lease has been created
+				firstMemberObservedOnce.Do(func() {
+					close(firstMemberObservedLeader)
+				})
 			},
 			OnStoppedLeading: func() {
 				log.Println("I'm not the leader anymore")
@@ -59,8 +71,18 @@ func TestRunElectionWithMemberIDCollision(t *testing.T) {
 
 	go func() {
 		defer wg.Done()
-		time.Sleep(time.Millisecond * 50) // make sure the first one becomes leader
-		g.Expect(etcd.RunElection(ctx, config)).Should(MatchError(ContainSubstring("creating lease")))
+		// Wait for the first member to observe a leader, which means the lease has been created
+		select {
+		case <-firstMemberObservedLeader:
+			// First member has created the lease, now try to create a conflicting one
+		case <-time.After(5 * time.Second):
+			t.Error("timeout waiting for first member to observe leader")
+			return
+		}
+		// Use a cancellable context to prevent hanging if this goroutine unexpectedly succeeds
+		member2Ctx, cancelMember2 := context.WithTimeout(ctx, 5*time.Second)
+		defer cancelMember2()
+		g.Expect(etcd.RunElection(member2Ctx, config)).Should(MatchError(ContainSubstring("creating lease")))
 	}()
 
 	wg.Wait()
@@ -69,7 +91,8 @@ func TestRunElectionWithMemberIDCollision(t *testing.T) {
 func TestRunElectionWithTwoMembersAndReelection(t *testing.T) {
 	t.Parallel()
 	g := NewWithT(t)
-	ctx := context.Background()
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
 	cli := client(g)
 	defer cli.Close()
 
@@ -85,24 +108,27 @@ func TestRunElectionWithTwoMembersAndReelection(t *testing.T) {
 		LeaseDurationSeconds: 1,
 	}
 
-	member1Ctx, _ := context.WithCancel(ctx)
+	member1Ctx, cancelMember1 := context.WithCancel(ctx)
 	member2Ctx, cancelMember2 := context.WithCancel(ctx)
 
 	config1 := configBase
 	config1.EtcdConfig.Client = cliMember1
-	config1.MemberID = "my-host"
-	uniqueID := rand.Int63()
+	config1.MemberID = randomElectionNameForTest("my-host")
+	uniqueID := rand.Uint64()
 	config1.MemberUniqueID = &uniqueID
 	config1.Callbacks = baseCallbacksForName(config1.MemberID)
+	syncMembers := make(chan (any))
 	config1.Callbacks.OnStartedLeading = func(_ context.Context) {
 		log.Println("I'm my-host, the new leader!!!!")
-		log.Println("Loosing the leadership on purpose by stopping renewing the lease")
+		close(syncMembers)
+		log.Println("Losing the leadership on purpose by stopping renewing the lease")
 		g.Expect(cliMember1.Lease.Close()).To(Succeed())
 		log.Println("Member1 leases closed")
+		cancelMember1()
 	}
 
 	config2 := configBase
-	config2.MemberID = "my-other-host"
+	config2.MemberID = randomElectionNameForTest("my-other-host")
 	config2.Callbacks = baseCallbacksForName(config2.MemberID)
 	config2.Callbacks.OnStartedLeading = func(_ context.Context) {
 		log.Println("I'm my-other-host, the new leader!!!!")
@@ -116,17 +142,18 @@ func TestRunElectionWithTwoMembersAndReelection(t *testing.T) {
 	go func() {
 		defer wg.Done()
 		g.Expect(etcd.RunElection(member1Ctx, &config1)).To(Succeed())
-		log.Println("Member1 routine done")
+		log.Printf("%s routine done\n", config1.MemberID)
 	}()
 
 	go func() {
 		defer wg.Done()
-		time.Sleep(time.Millisecond * 50) // Make sure member1 becomes leader
+		<-syncMembers
 		g.Expect(etcd.RunElection(member2Ctx, &config2)).To(Succeed())
-		log.Println("Member2 routine done")
+		log.Printf("%s routine done\n", config2.MemberID)
 	}()
 
 	wg.Wait()
+
 }
 
 func baseCallbacksForName(name string) etcd.LeaderCallbacks {

pkg/etcd/etcd_suite_test.go

@@ -15,7 +15,6 @@ import (
 	"time"
 
 	"github.com/pkg/errors"
-	"github.com/sirupsen/logrus"
 )
 
 const (
@@ -28,8 +27,8 @@ const (
 )
 
 func TestMain(m *testing.M) {
-	logrus.SetLevel(logrus.DebugLevel)
-	ctx := context.Background()
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
 	expectSuccess(startEtcd(ctx), "starting etcd")
 
 	os.Exit(runTestsWithCleanup(m, func() {

pkg/instance/instance.go

@@ -0,0 +1,748 @@
+package instance
+
+import (
+	"context"
+	"fmt"
+	"net"
+	"slices"
+	"strconv"
+	"strings"
+	"sync"
+
+	log "log/slog"
+
+	"github.com/vishvananda/netlink"
+	v1 "k8s.io/api/core/v1"
+
+	"github.com/kube-vip/kube-vip/pkg/arp"
+	"github.com/kube-vip/kube-vip/pkg/cluster"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/networkinterface"
+	"github.com/kube-vip/kube-vip/pkg/node"
+	"github.com/kube-vip/kube-vip/pkg/route"
+	"github.com/kube-vip/kube-vip/pkg/sysctl"
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	"github.com/kube-vip/kube-vip/pkg/vip"
+)
+
+// Instance defines an instance of everything needed to manage vips
+type Instance struct {
+	// Virtual IP / Load Balancer configuration
+	VIPConfigs []*kubevip.Config
+
+	// cluster instances
+	Clusters []*cluster.Cluster
+
+	// Service uses DHCP
+	IsDHCPv4            bool
+	IsDHCPv6            bool
+	DHCPInterface       string
+	DHCPInterfaceHwaddr string
+	DHCPInterfaceIP     string
+	DHCPInterfaceIPv4   string
+	DHCPInterfaceIPv6   string
+	DHCPHostname        string
+	DHCPv4Client        vip.DHCPClient
+	DHCPv6Client        vip.DHCPClient
+	macvlanName         string
+	dhcpBroadcast       bool
+
+	// Service use Vlan
+	IsVLAN        bool
+	VLANInterface string
+
+	// External Gateway IP the service is forwarded from
+	UPNPGatewayIPs []string
+
+	// Kubernetes service mapping
+	ServiceSnapshot *v1.Service
+
+	dnsAddresses []string
+
+	// AddCalled determined that ActionAdd was already performed for the instance
+	AddCalled bool
+
+	// LabelAdded determined that node was labeled with
+	// service-provided.kube-vip.io label
+	LabelAdded bool
+}
+
+type Port struct {
+	Port uint16
+	Type string
+}
+
+func NewInstance(ctx context.Context, svc *v1.Service, config *kubevip.Config,
+	intfMgr *networkinterface.Manager, arpMgr *arp.Manager, routeMgr *route.Manager,
+	nodeLabelMgr node.Labeler, wg *sync.WaitGroup) (*Instance, error) {
+	instanceAddresses, instanceHostnames := FetchServiceAddresses(svc)
+	log.Info("new instance", "namespace", svc.Namespace, "service", svc.Name, "addresses", instanceAddresses, "hostnames", instanceHostnames)
+
+	var newVips []*kubevip.Config
+	var link netlink.Link
+	var err error
+	var dnsAddresses []string
+
+	// Create new service
+	instance := &Instance{
+		ServiceSnapshot: svc,
+		dnsAddresses:    dnsAddresses,
+	}
+
+	for _, address := range instanceAddresses {
+		// Detect if we're using a specific interface for services
+		var svcInterface string
+
+		svcInterface = svc.Annotations[kubevip.ServiceVlan]
+		if svcInterface != "" {
+			parent, tag, err := utils.ParseVLANInterface(svcInterface)
+			if err != nil {
+				log.Error("failed to validate VLAN", "err", err)
+			}
+
+			if err := instance.addVLAN(parent, tag); err != nil {
+				log.Error("failed to create VLAN", "err", err)
+			}
+		} else {
+			// If no vlan defined use specific interface from annotation
+			svcInterface = svc.Annotations[kubevip.ServiceInterface]
+		}
+
+		if svcInterface == kubevip.Auto {
+			link, err = autoFindInterface(address)
+			if err != nil {
+				log.Error("automatically discover network interface for annotated IP", "address", address, "err", err)
+			} else {
+				if link == nil {
+					log.Error("automatically discover network interface for annotated IP address", "address", address)
+				}
+			}
+			if link == nil {
+				svcInterface = ""
+			} else {
+				svcInterface = getAutoInterfaceName(link, config.Interface)
+			}
+		}
+		// If it is still blank then use the
+		if svcInterface == "" {
+			switch config.ServicesInterface {
+			case kubevip.Auto:
+				link, err = autoFindInterface(address)
+				if err != nil {
+					log.Error("failed to automatically discover network interface for address", "ip", address, "err", err, "interface", config.Interface)
+				} else if link == nil {
+					log.Error("failed to automatically discover network interface for address", "ip", address, "defaulting to", config.Interface)
+				}
+				svcInterface = getAutoInterfaceName(link, config.Interface)
+			case "":
+				svcInterface = config.Interface
+			default:
+				svcInterface = config.ServicesInterface
+			}
+		}
+
+		if link == nil {
+			if link, err = netlink.LinkByName(svcInterface); err != nil {
+				return nil, fmt.Errorf("failed to get interface %s: %w", svcInterface, err)
+			}
+			if link == nil {
+				return nil, fmt.Errorf("failed to get interface %s", svcInterface)
+			}
+		}
+
+		cidrs := vip.Split(config.VIPSubnet)
+
+		ipv4AutoSubnet := false
+		ipv6AutoSubnet := false
+		if cidrs[0] == kubevip.Auto {
+			ipv4AutoSubnet = true
+		}
+
+		if len(cidrs) > 1 && cidrs[1] == kubevip.Auto {
+			ipv6AutoSubnet = true
+		}
+
+		if (config.Address != "" || config.VIP != "") && (ipv4AutoSubnet || ipv6AutoSubnet) {
+			return nil, fmt.Errorf("auto subnet discovery cannot be used if VIP address was provided")
+		}
+
+		subnet := ""
+		var err error
+		if utils.IsIPv4(address) {
+			if ipv4AutoSubnet {
+				subnet, err = autoFindSubnet(link, address)
+				if err != nil {
+					return nil, fmt.Errorf("failed to automatically find subnet for service %s/%s with IP address %s on interface %s: %w", svc.Namespace, svc.Name, address, svcInterface, err)
+				}
+			} else {
+				if cidrs[0] != "" && cidrs[0] != kubevip.Auto {
+					subnet = cidrs[0]
+				} else {
+					subnet = strconv.Itoa(vip.DefaultMaskIPv4)
+				}
+			}
+		} else {
+			if ipv6AutoSubnet {
+				subnet, err = autoFindSubnet(link, address)
+				if err != nil {
+					return nil, fmt.Errorf("failed to automatically find subnet for service %s/%s with IP address %s on interface %s: %w", svc.Namespace, svc.Name, address, svcInterface, err)
+				}
+			} else {
+				if len(cidrs) > 1 && cidrs[1] != "" && cidrs[1] != kubevip.Auto {
+					subnet = cidrs[1]
+				} else {
+					subnet = strconv.Itoa(vip.DefaultMaskIPv6)
+				}
+			}
+		}
+
+		// Generate new Virtual IP configuration
+		newVips = append(newVips, &kubevip.Config{
+			VIP:                         address,
+			Interface:                   svcInterface,
+			SingleNode:                  true,
+			EnableARP:                   config.EnableARP,
+			EnableBGP:                   config.EnableBGP,
+			VIPSubnet:                   subnet,
+			EnableRoutingTable:          config.EnableRoutingTable,
+			RoutingTableID:              config.RoutingTableID,
+			RoutingTableType:            config.RoutingTableType,
+			RoutingProtocol:             config.RoutingProtocol,
+			SkipDAD:                     config.SkipDAD,
+			ArpBroadcastRate:            config.ArpBroadcastRate,
+			EnableServiceSecurity:       config.EnableServiceSecurity,
+			DNSMode:                     config.DNSMode,
+			DHCPMode:                    config.DHCPMode,
+			DHCPBackoffAttempts:         config.DHCPBackoffAttempts,
+			DisableServiceUpdates:       config.DisableServiceUpdates,
+			EnableServicesElection:      config.EnableServicesElection,
+			PreserveVIPOnLeadershipLoss: config.PreserveVIPOnLeadershipLoss,
+			KubernetesLeaderElection: kubevip.KubernetesLeaderElection{
+				EnableLeaderElection: config.EnableLeaderElection,
+			},
+		})
+	}
+
+	for _, hostname := range instanceHostnames {
+		log.Info("hostname", "addr", hostname)
+		// Detect if we're using a specific interface for services
+		var svcInterface string
+
+		svcInterface = svc.Annotations[kubevip.ServiceVlan]
+		if svcInterface != "" {
+			parent, tag, err := utils.ParseVLANInterface(svcInterface)
+			if err != nil {
+				log.Error("failed to validate VLAN", "err", err)
+			}
+
+			if err := instance.addVLAN(parent, tag); err != nil {
+				log.Error("failed to create VLAN", "err", err)
+			}
+		} else {
+			// If no vlan defined use specific interface from annotation
+			svcInterface = svc.Annotations[kubevip.ServiceInterface]
+		}
+
+		// If it is still blank then use the
+		if svcInterface == "" {
+			switch config.ServicesInterface {
+			case "":
+				svcInterface = config.Interface
+			default:
+				svcInterface = config.ServicesInterface
+			}
+		}
+
+		if link == nil {
+			if link, err = netlink.LinkByName(svcInterface); err != nil {
+				return nil, fmt.Errorf("failed to get interface %s: %w", svcInterface, err)
+			}
+			if link == nil {
+				return nil, fmt.Errorf("failed to get interface %s", svcInterface)
+			}
+		}
+
+		// Generate new Virtual IP configuration
+		newVips = append(newVips, &kubevip.Config{
+			VIP:                    hostname,
+			Interface:              svcInterface,
+			SingleNode:             true,
+			EnableARP:              config.EnableARP,
+			EnableBGP:              config.EnableBGP,
+			VIPSubnet:              config.VIPSubnet,
+			EnableRoutingTable:     config.EnableRoutingTable,
+			RoutingTableID:         config.RoutingTableID,
+			RoutingTableType:       config.RoutingTableType,
+			RoutingProtocol:        config.RoutingProtocol,
+			SkipDAD:                config.SkipDAD,
+			ArpBroadcastRate:       config.ArpBroadcastRate,
+			EnableServiceSecurity:  config.EnableServiceSecurity,
+			DNSMode:                config.DNSMode,
+			DHCPMode:               config.DHCPMode,
+			DisableServiceUpdates:  config.DisableServiceUpdates,
+			EnableServicesElection: config.EnableServicesElection,
+			KubernetesLeaderElection: kubevip.KubernetesLeaderElection{
+				EnableLeaderElection: config.EnableLeaderElection,
+			},
+		})
+	}
+
+	if svc.Annotations != nil {
+		instance.DHCPInterfaceHwaddr = svc.Annotations[kubevip.HwAddrKey]
+		requestedIP := svc.Annotations[kubevip.RequestedIP]
+		if requestedIP != "" {
+			requestedIPs := strings.Split(requestedIP, ",")
+			if len(requestedIPs) > 2 {
+				return nil, fmt.Errorf("annotation %q cannot request more than one IPv4 and one Ipv6 address", kubevip.RequestedIP)
+			}
+			for _, ip := range requestedIPs {
+				netip := net.ParseIP(ip)
+				if netip.To4() != nil {
+					instance.DHCPInterfaceIPv4 = ip
+				} else {
+					instance.DHCPInterfaceIPv6 = ip
+				}
+			}
+		}
+		instance.DHCPHostname = svc.Annotations[kubevip.LoadbalancerHostname]
+		instance.macvlanName = svc.Annotations[kubevip.MacvlanName]
+		instance.dhcpBroadcast = svc.Annotations[kubevip.DHCPBroadcast] == "true"
+	}
+
+	configPorts := make([]kubevip.Port, 0)
+	for _, p := range svc.Spec.Ports {
+		configPorts = append(configPorts, kubevip.Port{
+			Type: string(p.Protocol),
+			Port: int(p.Port),
+		})
+	}
+	// Generate Load Balancer config
+	newLB := kubevip.LoadBalancer{
+		Name:      fmt.Sprintf("%s-load-balancer", svc.Name),
+		Ports:     configPorts,
+		BindToVip: true,
+	}
+	for _, vip := range newVips {
+		// Add Load Balancer Configuration
+		vip.LoadBalancers = append(vip.LoadBalancers, newLB)
+	}
+	// Create Add configuration to the new service
+	instance.VIPConfigs = newVips
+
+	// If this was purposely created with the address '0.0.0.0', or '::'
+	// we will create a macvlan on the main interface and a DHCP client
+	if len(instanceAddresses) > 2 && (slices.Contains(instanceAddresses, "0.0.0.0") || slices.Contains(instanceAddresses, "::")) {
+		return nil, fmt.Errorf("DHCP cannot be used if more than 2 addresses (one IPv4 and one IPv6) were specified")
+	}
+	for i := range instance.VIPConfigs {
+		if instance.VIPConfigs[i].VIP == "0.0.0.0" {
+			err := instance.startDHCP(ctx, i, config.DHCPBackoffAttempts, wg)
+			if err != nil {
+				return nil, err
+			}
+			select {
+			case err := <-instance.DHCPv4Client.ErrorChannel():
+				return nil, fmt.Errorf("error starting DHCPv4 for %s/%s: error: %s",
+					instance.ServiceSnapshot.Namespace, instance.ServiceSnapshot.Name, err)
+			case ip := <-instance.DHCPv4Client.IPChannel():
+				instance.VIPConfigs[i].Interface = instance.DHCPInterface
+				instance.VIPConfigs[i].VIP = ip
+				instance.DHCPInterfaceIPv4 = ip
+			}
+		}
+		if instance.VIPConfigs[i].VIP == "::" {
+			err := instance.startDHCP(ctx, i, config.DHCPBackoffAttempts, wg)
+			if err != nil {
+				return nil, err
+			}
+			select {
+			case err := <-instance.DHCPv6Client.ErrorChannel():
+				return nil, fmt.Errorf("error starting DHCPv6 for %s/%s: error: %s",
+					instance.ServiceSnapshot.Namespace, instance.ServiceSnapshot.Name, err)
+			case ip := <-instance.DHCPv6Client.IPChannel():
+				instance.VIPConfigs[i].Interface = instance.DHCPInterface
+				instance.VIPConfigs[i].VIP = ip
+				instance.DHCPInterfaceIPv6 = ip
+			}
+		}
+
+		ddnsAnnotation, exists := svc.Annotations[kubevip.ServiceDDNS]
+
+		if exists {
+			instance.VIPConfigs[i].DDNS, err = strconv.ParseBool(ddnsAnnotation)
+			if err != nil {
+				log.Error("Failed to add service", "err", err)
+				return nil, err
+			}
+		}
+
+		if len(svc.Spec.IPFamilies) > 0 {
+			if len(svc.Spec.IPFamilies) > 1 {
+				instance.VIPConfigs[i].DHCPMode = utils.DualFamily
+				instance.VIPConfigs[i].DNSMode = utils.DualFamily
+				switch *svc.Spec.IPFamilyPolicy {
+				case v1.IPFamilyPolicyRequireDualStack:
+					instance.VIPConfigs[i].IsDualStack = true
+					instance.VIPConfigs[i].RequireDualStack = true
+				case v1.IPFamilyPolicyPreferDualStack:
+					instance.VIPConfigs[i].IsDualStack = true
+					instance.VIPConfigs[i].RequireDualStack = false
+				default:
+					instance.VIPConfigs[i].IsDualStack = false
+					instance.VIPConfigs[i].RequireDualStack = false
+				}
+			} else {
+				if strings.EqualFold(string(svc.Spec.IPFamilies[0]), utils.IPv4Family) {
+					instance.VIPConfigs[i].DHCPMode = strings.ToLower(utils.IPv4Family)
+					instance.VIPConfigs[i].DNSMode = strings.ToLower(utils.IPv4Family)
+				} else {
+					instance.VIPConfigs[i].DHCPMode = strings.ToLower(utils.IPv6Family)
+					instance.VIPConfigs[i].DNSMode = strings.ToLower(utils.IPv6Family)
+				}
+			}
+		}
+
+		instance.VIPConfigs[i].EgressWithNftables = config.EgressWithNftables
+
+		c, err := cluster.InitCluster(instance.VIPConfigs[i], false, intfMgr, arpMgr, routeMgr, nodeLabelMgr)
+		if err != nil {
+			log.Error("failed to add service", "err", err)
+			return nil, err
+		}
+
+		for i := range c.Network {
+			c.Network[i].SetServicePorts(svc)
+		}
+
+		instance.Clusters = append(instance.Clusters, c)
+		log.Info("(svcs) adding VIP", "ip", instance.VIPConfigs[i].VIP, "interface", instance.VIPConfigs[i].Interface, "namespace", svc.Namespace, "name", svc.Name)
+	}
+
+	return instance, nil
+}
+
+func autoFindInterface(ip string) (netlink.Link, error) {
+	links, err := netlink.LinkList()
+	if err != nil {
+		return nil, fmt.Errorf("failed to list network interfaces: %w", err)
+	}
+
+	address := net.ParseIP(ip)
+
+	family := netlink.FAMILY_V4
+
+	if address.To4() == nil {
+		family = netlink.FAMILY_V6
+	}
+
+	for _, link := range links {
+		addr, err := netlink.AddrList(link, family)
+		if err != nil {
+			return nil, fmt.Errorf("failed to get IP addresses for interface %s: %w", link.Attrs().Name, err)
+		}
+		for _, a := range addr {
+			if a.IPNet.Contains(address) {
+				return link, nil
+			}
+		}
+	}
+
+	return nil, nil
+}
+
+func autoFindSubnet(link netlink.Link, ip string) (string, error) {
+	address := net.ParseIP(ip)
+
+	family := netlink.FAMILY_V4
+	if address.To4() == nil {
+		family = netlink.FAMILY_V6
+	}
+
+	addr, err := netlink.AddrList(link, family)
+	if err != nil {
+		return "", fmt.Errorf("failed to get IP addresses for interface %s: %w", link.Attrs().Name, err)
+	}
+	for _, a := range addr {
+		if a.IPNet.Contains(address) {
+			m, _ := a.IPNet.Mask.Size()
+			return strconv.Itoa(m), nil
+		}
+	}
+	return "", fmt.Errorf("failed to find suitable subnet for address %s", ip)
+}
+
+func getAutoInterfaceName(link netlink.Link, defaultInterface string) string {
+	if link == nil {
+		return defaultInterface
+	}
+	return link.Attrs().Name
+}
+
+func (i *Instance) addVLAN(parentInterface string, tag int) error {
+	var parent netlink.Link
+
+	interfaceName := fmt.Sprintf("%s.%d", parentInterface, tag)
+	iface, err := netlink.LinkByName(interfaceName)
+	if err != nil {
+		// check if parent interface doesnt exist
+		parent, err = netlink.LinkByName(parentInterface)
+		if err != nil {
+			return fmt.Errorf("error finding VLAN parent interface %s:  %v", parentInterface, err)
+		}
+
+		log.Info("Creating new VLAN interface", "interface", interfaceName)
+
+		vlan := &netlink.Vlan{
+			LinkAttrs: netlink.LinkAttrs{
+				Name:        interfaceName,
+				ParentIndex: parent.Attrs().Index,
+			},
+			VlanId:       tag,
+			VlanProtocol: netlink.VLAN_PROTOCOL_8021Q,
+		}
+
+		err = netlink.LinkAdd(vlan)
+		if err != nil {
+			return fmt.Errorf("could not add VLAN %s: %v", interfaceName, err)
+		}
+
+		err = netlink.LinkSetUp(vlan)
+		if err != nil {
+			return fmt.Errorf("could not bring up VLAN interface [%s] : %v", interfaceName, err)
+		}
+
+		_, err = net.InterfaceByName(interfaceName)
+		if err != nil {
+			return fmt.Errorf("error finding new VLAN interface by name [%v]", err)
+		}
+	} else {
+		log.Info("Using existing VLAN interface", "interface", interfaceName)
+
+		if err := utils.ValidateVLANInterface(iface, parent, tag); err != nil {
+			return err
+		}
+	}
+
+	i.VLANInterface = interfaceName
+	i.IsVLAN = true
+
+	return nil
+}
+
+func (i *Instance) startDHCP(ctx context.Context, index int, backoffAttempts uint, wg *sync.WaitGroup) error {
+	if len(i.VIPConfigs) > 2 {
+		return fmt.Errorf("DHCP can be used with 2 VIP config maximally, got: %v", len(i.VIPConfigs))
+	}
+	parent, err := netlink.LinkByName(i.VIPConfigs[index].Interface)
+	if err != nil {
+		return fmt.Errorf("error finding VIP Interface, for building DHCP Link : %v", err)
+	}
+
+	interfaceName := i.macvlanName
+
+	if interfaceName == "" {
+		// Generate name from UID
+		interfaceName = fmt.Sprintf("vip-%s", i.ServiceSnapshot.UID[0:8])
+	}
+
+	// Check if the interface doesn't exist first
+	iface, err := net.InterfaceByName(interfaceName)
+	if err != nil {
+		log.Info("creating new macvlan interface for DHCP", "interface", interfaceName)
+
+		hwaddr, err := net.ParseMAC(i.DHCPInterfaceHwaddr)
+		if i.DHCPInterfaceHwaddr != "" && err != nil {
+			return err
+		} else if hwaddr == nil {
+			hwaddr, err = net.ParseMAC(vip.GenerateMac())
+			if err != nil {
+				return err
+			}
+		}
+
+		log.Info("new macvlan interface", "interface", interfaceName, "hardware address", hwaddr)
+		mac := &netlink.Macvlan{
+			LinkAttrs: netlink.LinkAttrs{
+				Name:         interfaceName,
+				ParentIndex:  parent.Attrs().Index,
+				HardwareAddr: hwaddr,
+			},
+			Mode: netlink.MACVLAN_MODE_DEFAULT,
+		}
+
+		err = netlink.LinkAdd(mac)
+		if err != nil {
+			return fmt.Errorf("could not add %s: %v", interfaceName, err)
+		}
+
+		err = netlink.LinkSetUp(mac)
+		if err != nil {
+			return fmt.Errorf("could not bring up interface [%s] : %v", interfaceName, err)
+		}
+
+		iface, err = net.InterfaceByName(interfaceName)
+		if err != nil {
+			return fmt.Errorf("error finding new DHCP interface by name [%v]", err)
+		}
+	} else {
+		log.Info("Using existing macvlan interface for DHCP", "interface", interfaceName)
+	}
+
+	var initRebootFlag bool
+	ip := net.ParseIP(i.VIPConfigs[index].VIP)
+
+	var client vip.DHCPClient
+	if ip.To4() != nil {
+		// Default rp_filter setting (https://github.com/kube-vip/kube-vip/issues/1170)
+		rpfilterSetting := "0"
+
+		// Check if we need to set an override rp_filter value for the interface
+		if i.ServiceSnapshot.Annotations[kubevip.RPFilter] != "" {
+			// Check the rp_filter value
+			rpFilter, err := strconv.Atoi(i.ServiceSnapshot.Annotations[kubevip.RPFilter])
+			if err != nil {
+				log.Error("[DHCP] unable to process rp_filter", "value", rpFilter)
+			} else {
+				if rpFilter >= 0 && rpFilter < 3 { // Ensure the value is 0,1,2
+					rpfilterSetting = i.ServiceSnapshot.Annotations[kubevip.RPFilter]
+				} else {
+					log.Error("[DHCP] rp_filter value not within range 0-2", "value", rpFilter)
+				}
+			}
+		}
+
+		err = sysctl.WriteProcSys("/proc/sys/net/ipv4/conf/"+interfaceName+"/rp_filter", rpfilterSetting)
+		if err != nil {
+			log.Error("[DHCP] unable to write rp_filter", "value", rpfilterSetting, "err", err)
+		}
+
+		if i.DHCPInterfaceIPv4 != "" {
+			initRebootFlag = true
+		}
+
+		client = vip.NewDHCPv4Client(iface, initRebootFlag, i.DHCPInterfaceIPv4, backoffAttempts, i.dhcpBroadcast)
+
+		// Add the client so that we can call it to stop function
+		i.DHCPv4Client = client
+
+		// Set that DHCPv4 is enabled
+		i.IsDHCPv4 = true
+	} else {
+		if i.DHCPInterfaceIPv6 != "" {
+			initRebootFlag = true
+		}
+
+		client, err = vip.NewDHCPv6Client(iface, parent, initRebootFlag, i.DHCPInterfaceIPv6, backoffAttempts)
+		if err != nil {
+			return fmt.Errorf("unable to create client: %w", err)
+		}
+
+		// Add the client so that we can call it to stop function
+		i.DHCPv6Client = client
+
+		// Set that DHCPv6 is enabled
+		i.IsDHCPv6 = true
+	}
+
+	// Add hostname to dhcp client if annotated
+	if i.DHCPHostname != "" {
+		log.Info("Hostname specified for dhcp lease", "interface", interfaceName, "hostname", i.DHCPHostname)
+		client.WithHostName(i.DHCPHostname)
+	}
+
+	wg.Go(func() {
+		if err := client.Start(ctx); err != nil {
+			log.Error("[instance] DHCP client error: %w")
+		}
+	})
+
+	// Set the name of the interface so that it can be removed on Service deletion
+	i.DHCPInterface = interfaceName
+	i.DHCPInterfaceHwaddr = iface.HardwareAddr.String()
+
+	return nil
+}
+
+// FetchLoadBalancerIngressAddresses tries to get the addresses from status.loadBalancerIP
+func FetchLoadBalancerIngress(s *v1.Service) ([]string, []string) {
+	// If the service has no status, return empty
+	lbStatusAddresses := []string{}
+	lbStatusHostnames := []string{}
+	if len(s.Status.LoadBalancer.Ingress) == 0 {
+		return lbStatusAddresses, lbStatusHostnames
+	}
+
+	for _, ingress := range s.Status.LoadBalancer.Ingress {
+		if ingress.IP != "" {
+			lbStatusAddresses = append(lbStatusAddresses, ingress.IP)
+		}
+		if ingress.Hostname != "" {
+			lbStatusHostnames = append(lbStatusHostnames, ingress.Hostname)
+		}
+	}
+	return lbStatusAddresses, lbStatusHostnames
+}
+
+// FetchServiceAddresses tries to get the addresses from annotations
+// kube-vip.io/loadbalancerIPs, then from spec.loadbalancerIP
+func FetchServiceAddresses(s *v1.Service) ([]string, []string) {
+	annotationAvailable := false
+	if s.Annotations != nil {
+
+		if v, annotationAvailable := s.Annotations[kubevip.LoadbalancerIPAnnotation]; annotationAvailable {
+			ips := strings.Split(v, ",")
+			var trimmedIPs []string
+			var trimmedHostnames []string
+			for _, a := range ips {
+				a = strings.TrimSpace(a)
+				ip := net.ParseIP(a)
+				if ip == nil {
+					// this is probably a DNS name
+					trimmedHostnames = append(trimmedHostnames, a)
+				} else {
+					trimmedIPs = append(trimmedIPs, ip.String())
+				}
+			}
+			return trimmedIPs, trimmedHostnames
+		}
+	}
+
+	lbStatusAddresses := []string{}
+	lbStatusHostnames := []string{}
+	if !annotationAvailable {
+		lbStatusAddresses, lbStatusHostnames = FetchLoadBalancerIngress(s)
+	}
+
+	// Spec.LoadBalancerIP legacy handling
+	// if the loadBalancerIP is different from Status.LoadBalancer.Ingress IPs
+	// return the legacy LB as spec wins over status.
+	if lbIP := net.ParseIP(s.Spec.LoadBalancerIP); lbIP != nil && len(lbStatusAddresses) > 0 {
+		isLbIPv4 := utils.IsIPv4(s.Spec.LoadBalancerIP)
+		for _, a := range lbStatusAddresses {
+			if lbStatusIP := net.ParseIP(a); lbStatusIP != nil && utils.IsIPv4(a) == isLbIPv4 && !lbIP.Equal(lbStatusIP) {
+				return []string{s.Spec.LoadBalancerIP}, []string{}
+			}
+		}
+	}
+
+	if len(lbStatusAddresses) > 0 || len(lbStatusHostnames) > 0 {
+		return lbStatusAddresses, lbStatusHostnames
+	}
+
+	if s.Spec.LoadBalancerIP != "" {
+		return []string{s.Spec.LoadBalancerIP}, []string{}
+	}
+
+	return []string{}, []string{}
+}
+
+func FindServiceInstance(svc *v1.Service, instances []*Instance) *Instance {
+	log.Debug("finding service", "namespace", svc.Namespace, "name", svc.Name, "UID", svc.UID)
+	for i := range instances {
+		if instances[i].ServiceSnapshot.UID == svc.UID {
+			return instances[i]
+		}
+	}
+	log.Debug("instance not found", "namespace", svc.Namespace, "name", svc.Name, "UID", svc.UID)
+	return nil
+}

pkg/k8s/client.go

@@ -6,7 +6,8 @@ import (
 	"net"
 	"time"
 
-	log "github.com/sirupsen/logrus"
+	log "log/slog"
+
 	"k8s.io/client-go/kubernetes"
 	"k8s.io/client-go/rest"
 	"k8s.io/client-go/tools/clientcmd"
@@ -99,21 +100,21 @@ func FindWorkingKubernetesAddress(configPath string, inCluster bool) (*rest.Conf
 		return nil, err
 	}
 	for x := range ips {
-		log.Debugf("[k8s client] checking with IP address [%s]", ips[x].String())
+		log.Debug("[k8s client] testing", "address", ips[x].String())
 		c, err := NewRestConfig(configPath, inCluster, net.JoinHostPort(ips[x].String(), "6443"))
 		if err != nil {
-			log.Errorf("failed to create k8s REST config: %v", err)
+			log.Error("failed to create k8s REST config", "err", err)
 		}
 
 		c.Timeout = 2 * time.Second
 		k, err := NewClientset(c)
 		if err != nil {
-			log.Errorf("failed to create k8s clientset: %v", err)
+			log.Error("failed to create k8s clientset", "err", err)
 		}
 
 		_, err = k.DiscoveryClient.ServerVersion()
 		if err == nil {
-			log.Infof("[k8s client] working with IP address [%s]", ips[x].String())
+			log.Info("[k8s client] working", "address", ips[x].String())
 			c.Timeout = defaultTimeout
 			return c, nil
 		}

pkg/kubevip/annotations.go

@@ -0,0 +1,94 @@
+package kubevip
+
+const (
+	// Hardware address of the host that has the VIP
+	HwAddrKey = "kube-vip.io/hwaddr"
+
+	// The IP address that is requested
+	RequestedIP = "kube-vip.io/requestedIP"
+
+	// The host that has the VIP
+	VipHost = "kube-vip.io/vipHost"
+
+	// Enable Egress on a service
+	Egress = "kube-vip.io/egress"
+
+	// Enable internal Egress
+	EgressInternal = "kube-vip.io/egress-internal"
+
+	// Egress should be IPv6
+	EgressIPv6 = "kube-vip.io/egress-ipv6"
+
+	// Ports that traffic is allowed to access from the egress VIP
+	EgressDestinationPorts = "kube-vip.io/egress-destination-ports"
+
+	// Allowed incoming ports to the VIP
+	EgressSourcePorts = "kube-vip.io/egress-source-ports"
+
+	// Allowed networks for the Egress to be enabled for
+	EgressAllowedNetworks = "kube-vip.io/egress-allowed-networks"
+
+	// Networks that we wont Egress for
+	EgressDeniedNetworks = "kube-vip.io/egress-denied-networks"
+
+	// EgressNoInternalTraffic, when enabled will prevent any internal traffic from being SNATed to the egress VIP, even if the internal SNAT rule is enabled
+	EgressNoInternalTraffic = "kube-vip.io/egress-no-internal-traffic"
+
+	// EgressDetectAPIServer, this will attempt to detect the API server and add it to the allowed networks for egress
+	EgressDetectAPIServer = "kube-vip.io/egress-detect-api-server"
+
+	// The current active endpoint(pod) for the Egress VIP
+	ActiveEndpoint = "kube-vip.io/active-endpoint"
+
+	// The current active endpoint(pod) for the Egress VIP (v6)
+	ActiveEndpointIPv6 = "kube-vip.io/active-endpoint-ipv6"
+
+	// The nftables egress table base name that owns this Service's SNAT chain
+	EgressNftablesTable = "kube-vip.io/egress-nftables-table"
+
+	// Flush the conntrack rules (remove existing sessions) once Egress is configured
+	FlushContrack = "kube-vip.io/flush-conntrack"
+
+	// Configure LoadBalancer IPs instead of relying on a controller
+	LoadbalancerIPAnnotation = "kube-vip.io/loadbalancerIPs"
+
+	// Ignore the LoadBalancer Service
+	LoadbalancerIgnore = "kube-vip.io/ignore"
+
+	// Used to configure DHCP with a Hostname
+	LoadbalancerHostname = "kube-vip.io/loadbalancerHostname"
+
+	// Define an interface name to bind the address of the LoadBalancer to
+	ServiceInterface = "kube-vip.io/serviceInterface"
+
+	// Specify VLAN subinterface for service (e.g. eth0.200)
+	ServiceVlan = "kube-vip.io/serviceVLAN"
+
+	ServiceSecurityIgnore = "kube-vip.io/ignore-service-security"
+
+	// Enable UPNP on a Service
+	UpnpEnabled = "kube-vip.io/forwardUPNP"
+
+	// Set the UPNP lease duration for a specific service using duration format (e.g., "30s", "1h")
+	UpnpLeaseDuration = "kube-vip.io/upnp-lease-duration"
+
+	RPFilter = "kube-vip.io/rp_filter" // Set the return path filter for a specific service interface
+
+	// Name of the service lease object
+	ServiceLease = "kube-vip.io/leaseName"
+
+	// Forces kube-vip to use per service election for this particular service
+	ForcePerServiceElection = "kube-vip.io/forcePerServiceElection"
+
+	// Allow service reconciliation even when no endpoints are present (Cluster policy only)
+	AllowReconcileWithoutEndpoints = "kube-vip.io/allow-reconcile-without-endpoints"
+
+	// Enable DDNS for the service
+	ServiceDDNS = "kube-vip.io/ddns"
+
+	// Forces kube-vip to use the specified veth interface when DHCP is being used for a service
+	MacvlanName = "kube-vip.io/macvlanName"
+
+	// Set the BROADCAST flag in DHCP DISCOVER/REQUEST packets
+	DHCPBroadcast = "kube-vip.io/dhcp-broadcast"
+)

pkg/kubevip/config_bgp.go

@@ -0,0 +1,335 @@
+package kubevip
+
+import (
+	"fmt"
+	"net"
+	"strconv"
+	"strings"
+
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	api "github.com/osrg/gobgp/v4/api"
+
+	"github.com/vishvananda/netlink"
+)
+
+// Peer defines a BGP Peer
+type BGPPeer struct {
+	Address      string
+	Port         uint16
+	Interface    string
+	AS           uint32
+	Password     string
+	MultiHop     bool
+	MpbgpNexthop string
+	MpbgpIPv4    string
+	MpbgpIPv6    string
+
+	// BFD Configuration
+	BFDEnabled          bool
+	BFDReceiveInterval  uint32
+	BFDTransmitInterval uint32
+	BFDDetectMultiplier uint32
+}
+
+// Config defines the BGP server configuration
+type BGPConfig struct {
+	AS           uint32
+	RouterID     string
+	SourceIP     string
+	SourceIF     string
+	MpbgpNexthop string
+	MpbgpIPv4    string
+	MpbgpIPv6    string
+
+	HoldTime          uint64
+	KeepaliveInterval uint64
+
+	Peers []BGPPeer
+
+	Zebra ZebraConfig
+}
+
+// Defines Zebra connection configuration. More on the topic - https://github.com/osrg/gobgp/blob/master/docs/sources/zebra.md#configuration
+type ZebraConfig struct {
+	Enabled      bool
+	URL          string
+	Version      uint32
+	SoftwareName string
+}
+
+// BGP Peer layout is as follows:
+// <address>:<AS>:<password>:<multihop>:<port>:<optional mpbgp options>:<BFD options>
+
+// <address> - IP address of the peer. For IPv6 addresses, the address should be enclosed in square brackets (e.g. [fd00:100:64::2]). For unnumbered peers, the address should be prefixed with "unnumbered:" followed by the interface name (e.g. unnumbered:eth0).
+// <AS> - Autonomous System number of the peer (e.g. 65000)
+// <password> - Optional password for BGP authentication (e.g. secret)
+// <multihop> - Optional flag to indicate if this is a multihop peer (true/false, default: false)
+// <port> - Optional BGP port number (default: 179)
+// <optional mpbgp options> - Optional MP-BGP parameters in the format of key=value pairs separated by ';' (e.g. mpbgp_nexthop=auto_sourceif;mpbgp_ipv4=)
+// <BFD options> - Optional BFD parameters (if any) in the format of semicolon-separated values (enable, receive_interval, transmit_interval, detect_multiplier) (e.g. true;300;300;3)
+
+// ParseBGPPeerConfig - take a string and parses it into an array of peers
+func ParseBGPPeerConfig(config string) (bgpPeers []BGPPeer, err error) {
+	peers := strings.Split(config, ",")
+	if len(peers) == 0 || config == "" {
+		return nil, fmt.Errorf("no BGP Peer configurations found")
+	}
+
+	for x := range peers {
+		peerStr := peers[x]
+		if peerStr == "" {
+			continue
+		}
+
+		// Look at address peer
+		isV6Peer := peerStr[0] == '['
+		isUnnumberedPeer := strings.HasPrefix(peerStr, "unnumbered:")
+
+		address := ""
+		if isV6Peer {
+			addressEndPos := strings.IndexByte(peerStr, ']')
+			if addressEndPos == -1 {
+				return nil, fmt.Errorf("no matching ] found for IPv6 BGP Peer")
+			}
+			address = peerStr[1:addressEndPos]
+			peerStr = peerStr[addressEndPos+1:]
+		} else if isUnnumberedPeer {
+			unnumberedEndPos := strings.IndexByte(peerStr, ':')
+			peerStr = peerStr[unnumberedEndPos+1:]
+		}
+
+		peer := strings.Split(peerStr, ":")
+		if len(peer) < 2 && !isUnnumberedPeer {
+			return nil, fmt.Errorf("mandatory peering params <host>:<AS> incomplete")
+		}
+
+		iface := ""
+		if isUnnumberedPeer {
+			iface = peer[0]
+		} else if !isV6Peer {
+			address = peer[0]
+		}
+
+		// Look at peer[1] for AS number
+		var ASNumber uint64
+		if len(peer) >= 2 {
+			ASNumber, err = strconv.ParseUint(peer[1], 10, 32)
+			if err != nil {
+				return nil, fmt.Errorf("BGP Peer AS format error [%s]", peer[1])
+			}
+		}
+
+		// Look at peer[2] for password
+		password := ""
+		if len(peer) >= 3 {
+			password = peer[2]
+		}
+
+		// Look at peer[3] for multihop
+		multiHop := false
+		if len(peer) >= 4 && peer[3] != "" {
+			multiHop, err = strconv.ParseBool(peer[3])
+			if err != nil {
+				return nil, fmt.Errorf("BGP MultiHop format error (true/false) [%s]", peer[3])
+			}
+		}
+
+		// Look at peer[4] for BGP port
+		var port uint64
+		if len(peer) >= 5 {
+			if peer[4] == "" {
+				port = 179
+			} else {
+				port, err = strconv.ParseUint(peer[4], 10, 16)
+				if err != nil {
+					return nil, fmt.Errorf("BGP Peer Port format error [%s]", peer[4])
+				}
+			}
+		} else if !isUnnumberedPeer {
+			port = 179
+		}
+
+		// Look at peer[5] for optional MP-BGP parameters
+		var mpbgpNexthop, mpbgpIPv4, mpbgpIPv6 string
+
+		if len(peer) >= 6 && peer[5] != "" {
+			configData := strings.Split(peer[5], ";")
+			for _, cfg := range configData {
+				c := strings.Split(cfg, "=")
+				if len(c) < 2 {
+					return nil, fmt.Errorf("peer configuration parameter '%s' is missing a value (expected key=value)", c[0])
+				}
+				switch c[0] {
+				case "mpbgp_nexthop":
+					mpbgpNexthop = c[1]
+				case "mpbgp_ipv4":
+					mpbgpIPv4 = c[1]
+				case "mpbgp_ipv6":
+					mpbgpIPv6 = c[1]
+				default:
+					return nil, fmt.Errorf("peer configuration parameter '%s' is not supported", c[0])
+				}
+			}
+
+		}
+
+		// Look at peer[6] for optional BFD parameters (if any)
+		bfdEnabled := false
+		bfdReceiveInterval := uint64(300)
+		bfdTransmitInterval := uint64(300)
+		bfdDetectMultiplier := uint64(3)
+
+		if len(peer) >= 7 && peer[6] != "" {
+			c := strings.Split(peer[6], ";")
+			if len(c) < 4 {
+				return nil, fmt.Errorf("BFD configuration error: at least 4 parameters are required (enable, receive_interval, transmit_interval, detect_multiplier) [%s]", peer[6])
+			}
+			bfdEnabled, err = strconv.ParseBool(c[0])
+			if err != nil {
+				return nil, fmt.Errorf("BFD configuration error: invalid value for bfd_enabled (true/false) [%s]", c[0])
+			}
+
+			if c[1] != "" {
+				bfdReceiveInterval, err = strconv.ParseUint(c[1], 10, 32)
+				if err != nil {
+					return nil, fmt.Errorf("BFD configuration error: invalid value for bfd_receive_interval [%s]", c[1])
+				}
+			}
+
+			if c[2] != "" {
+				bfdTransmitInterval, err = strconv.ParseUint(c[2], 10, 32)
+				if err != nil {
+					return nil, fmt.Errorf("BFD configuration error: invalid value for bfd_transmit_interval [%s]", c[2])
+				}
+			}
+			if c[3] != "" {
+				bfdDetectMultiplier, err = strconv.ParseUint(c[3], 10, 32)
+				if err != nil {
+					return nil, fmt.Errorf("BFD configuration error: invalid value for bfd_detect_multiplier [%s]", c[3])
+				}
+			}
+		}
+
+		peerConfig := BGPPeer{
+			Address: address,
+			//nolint:gosec // previously parsed into uint32
+			AS:                  uint32(ASNumber),
+			Port:                uint16(port),
+			Interface:           iface,
+			Password:            password,
+			MultiHop:            multiHop,
+			MpbgpNexthop:        mpbgpNexthop,
+			MpbgpIPv4:           mpbgpIPv4,
+			MpbgpIPv6:           mpbgpIPv6,
+			BFDEnabled:          bfdEnabled,
+			BFDReceiveInterval:  uint32(bfdReceiveInterval),
+			BFDTransmitInterval: uint32(bfdTransmitInterval),
+			BFDDetectMultiplier: uint32(bfdDetectMultiplier),
+		}
+
+		bgpPeers = append(bgpPeers, peerConfig)
+	}
+	return
+}
+
+func (p *BGPPeer) FindMpbgpAddresses(ap *api.Peer, server *BGPConfig) (string, string, error) {
+	var ipv4Address, ipv6Address string
+
+	mode := server.MpbgpNexthop
+	if p.MpbgpNexthop != "" {
+		mode = p.MpbgpNexthop
+	}
+
+	switch mode {
+	case "fixed":
+		ap.Transport.LocalAddress = server.SourceIP
+
+		ipv4 := server.MpbgpIPv4
+		if p.MpbgpIPv4 != "" {
+			ipv4 = p.MpbgpIPv4
+		}
+
+		ipv6 := server.MpbgpIPv6
+		if p.MpbgpIPv6 != "" {
+			ipv6 = p.MpbgpIPv6
+		}
+
+		if ipv4 == "" && ipv6 == "" {
+			return "", "", fmt.Errorf("to use MP-BGP with fixed address at least one IPv4 or IPv6 address has to be provided [current - IPv4: %s, IPv6: %s]",
+				ipv4, ipv6)
+		}
+
+		if ipv4 != "" {
+			if net.ParseIP(ipv4) == nil {
+				return "", "", fmt.Errorf("provided address '%s' is not a valid IPv4 address", ipv4)
+			}
+		}
+		if ipv6 != "" {
+			if net.ParseIP(ipv6) == nil {
+				return "", "", fmt.Errorf("provided address '%s' is not a valid IPv6 address", ipv6)
+			}
+		}
+
+		ipv4Address = ipv4
+		ipv6Address = ipv6
+	case "auto_sourceip":
+		ap.Transport.LocalAddress = server.SourceIP
+
+		// Resolve the local interface by SourceIP
+		iface, err := utils.GetInterfaceByIP(server.SourceIP)
+		if err != nil {
+			return "", "", fmt.Errorf("failed to get interface by IP: %v", err)
+		}
+
+		if utils.IsIPv4(server.SourceIP) {
+			// Get the non link-local IPv6 address on that interface
+			ipv6Address, err = utils.GetNonLinkLocalIP(iface, netlink.FAMILY_V6)
+			if err != nil {
+				return "", "", fmt.Errorf("failed to get non link-local IPv6 address: %v", err)
+			}
+		} else {
+			// Get the non link-local IPv4 address on that interface
+			ipv4Address, err = utils.GetNonLinkLocalIP(iface, netlink.FAMILY_V4)
+			if err != nil {
+				return "", "", fmt.Errorf("failed to get non link-local IPv4 address: %v", err)
+			}
+		}
+	case "auto_sourceif":
+		ap.Transport.BindInterface = server.SourceIF
+
+		iface, err := netlink.LinkByName(server.SourceIF)
+		if err != nil {
+			return "", "", fmt.Errorf("failed to get interface by name: %v", err)
+		}
+
+		// Get the non link-local IPv4 address on that interface
+		ipv4Address, err = utils.GetNonLinkLocalIP(&iface, netlink.FAMILY_V4)
+		if err != nil {
+			return "", "", fmt.Errorf("failed to get non link-local IPv4 address: %v", err)
+		}
+
+		// Get the non link-local IPv6 address on that interface
+		ipv6Address, err = utils.GetNonLinkLocalIP(&iface, netlink.FAMILY_V6)
+		if err != nil {
+			return "", "", fmt.Errorf("failed to get non link-local IPv6 address: %v", err)
+		}
+	default:
+		return "", "", fmt.Errorf("option %q for MP-BPG nexthop is not supported", mode)
+	}
+
+	return ipv4Address, ipv6Address, nil
+}
+
+func (p *BGPPeer) SetMpbgpOptions(server *BGPConfig) {
+	if p.MpbgpNexthop == "" {
+		p.MpbgpNexthop = server.MpbgpNexthop
+	}
+
+	if p.MpbgpIPv4 == "" {
+		p.MpbgpIPv4 = server.MpbgpIPv4
+	}
+
+	if p.MpbgpIPv6 == "" {
+		p.MpbgpIPv6 = server.MpbgpIPv6
+	}
+}

pkg/kubevip/config_bgp_test.go

@@ -0,0 +1,116 @@
+package kubevip
+
+import (
+	"reflect"
+	"testing"
+)
+
+func TestParseBGPPeerConfig(t *testing.T) {
+	type args struct {
+		config string
+	}
+	tests := []struct {
+		name         string
+		args         args
+		wantBgpPeers []BGPPeer
+		wantErr      bool
+	}{
+
+		{
+			name: "IPv4, default port",
+			args: args{config: "192.168.0.10:65000::false,192.168.0.11:65000::false"},
+			wantBgpPeers: []BGPPeer{
+				{Address: "192.168.0.10", Port: 179, AS: 65000, MultiHop: false, BFDEnabled: false, BFDReceiveInterval: 300, BFDTransmitInterval: 300, BFDDetectMultiplier: 3},
+				{Address: "192.168.0.11", Port: 179, AS: 65000, MultiHop: false, BFDEnabled: false, BFDReceiveInterval: 300, BFDTransmitInterval: 300, BFDDetectMultiplier: 3},
+			},
+		},
+		{
+			name: "IPv4, different port",
+			args: args{config: "192.168.0.10:65000::false:180,192.168.0.11:65000::false:190"},
+			wantBgpPeers: []BGPPeer{
+				{Address: "192.168.0.10", Port: 180, AS: 65000, MultiHop: false, BFDEnabled: false, BFDReceiveInterval: 300, BFDTransmitInterval: 300, BFDDetectMultiplier: 3},
+				{Address: "192.168.0.11", Port: 190, AS: 65000, MultiHop: false, BFDEnabled: false, BFDReceiveInterval: 300, BFDTransmitInterval: 300, BFDDetectMultiplier: 3},
+			},
+		},
+		{
+			name: "IPv6, multi-protocol",
+			args: args{config: "[fd00:1111:2222:3333:c7d9:7235:6bf7:5d52]:65501::false::mpbgp_nexthop=auto_sourceif"},
+			wantBgpPeers: []BGPPeer{
+				{Address: "fd00:1111:2222:3333:c7d9:7235:6bf7:5d52", Port: 179, AS: 65501, MultiHop: false, MpbgpNexthop: "auto_sourceif", BFDEnabled: false, BFDReceiveInterval: 300, BFDTransmitInterval: 300, BFDDetectMultiplier: 3},
+			},
+		},
+		{
+			name: "IPv6, multi-protocol, BFD, no multi-protocol options",
+			args: args{config: "[fd00:1111:2222:3333:c7d9:7235:6bf7:5d52]:65501::false:::true;300;300;3"},
+			wantBgpPeers: []BGPPeer{
+				{Address: "fd00:1111:2222:3333:c7d9:7235:6bf7:5d52", Port: 179, AS: 65501, MultiHop: false, BFDEnabled: true, BFDReceiveInterval: 300, BFDTransmitInterval: 300, BFDDetectMultiplier: 3},
+			},
+		},
+		{
+			name: "IPv6, multi-protocol, BFD",
+			args: args{config: "[fd00:1111:2222:3333:c7d9:7235:6bf7:5d52]:65501::false::mpbgp_nexthop=auto_sourceif:true;300;300;3"},
+			wantBgpPeers: []BGPPeer{
+				{Address: "fd00:1111:2222:3333:c7d9:7235:6bf7:5d52", Port: 179, AS: 65501, MultiHop: false, MpbgpNexthop: "auto_sourceif", BFDEnabled: true, BFDReceiveInterval: 300, BFDTransmitInterval: 300, BFDDetectMultiplier: 3},
+			},
+		},
+		{
+			name: "IPv6 bracketed, with password and multihop",
+			args: args{config: "[fd00:100:64::2]:65000:secret:true"},
+			wantBgpPeers: []BGPPeer{
+				{Address: "fd00:100:64::2", Port: 179, AS: 65000, Password: "secret", MultiHop: true, BFDEnabled: false, BFDReceiveInterval: 300, BFDTransmitInterval: 300, BFDDetectMultiplier: 3},
+			},
+		},
+		{
+			name: "IPv6 bracketed, empty fields",
+			args: args{config: "[fd00:100:64::2]:65000::false"},
+			wantBgpPeers: []BGPPeer{
+				{Address: "fd00:100:64::2", Port: 179, AS: 65000, MultiHop: false, BFDEnabled: false, BFDReceiveInterval: 300, BFDTransmitInterval: 300, BFDDetectMultiplier: 3},
+			},
+		},
+		{
+			name: "Unnumbered",
+			args: args{config: "unnumbered:eth0,unnumbered:eth1:65000::true::mpbgp_nexthop=auto_sourceif"},
+			wantBgpPeers: []BGPPeer{
+				{Interface: "eth0", MultiHop: false, BFDEnabled: false, BFDReceiveInterval: 300, BFDTransmitInterval: 300, BFDDetectMultiplier: 3},
+				{Interface: "eth1", Port: 179, AS: 65000, MultiHop: true, MpbgpNexthop: "auto_sourceif", BFDEnabled: false, BFDReceiveInterval: 300, BFDTransmitInterval: 300, BFDDetectMultiplier: 3},
+			},
+		},
+		{
+			name:    "Completely empty config",
+			args:    args{config: ""},
+			wantErr: true,
+		},
+		{
+			name:    "Completely empty config (but with the seperators)",
+			args:    args{config: ":::::::"},
+			wantErr: true,
+		},
+		{
+			name:    "Malformed parameter (no value)",
+			args:    args{config: "1.2.3.4:65000/mpbgp_nexthop"},
+			wantErr: true,
+		},
+		{
+			name:    "Unsupported parameter",
+			args:    args{config: "1.2.3.4:65000;unknown=value"},
+			wantErr: true,
+		},
+		{
+			name:    "Malformed IPv6 (no matching bracket)",
+			args:    args{config: "[fd00:100:64::2:65000"},
+			wantErr: true,
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			gotBgpPeers, err := ParseBGPPeerConfig(tt.args.config)
+			if (err != nil) != tt.wantErr {
+				t.Errorf("ParseBGPPeerConfig() error = \n%v, wantErr \n%v %v", err, tt.wantErr, gotBgpPeers)
+				return
+			}
+			if !reflect.DeepEqual(gotBgpPeers, tt.wantBgpPeers) {
+				t.Errorf("ParseBGPPeerConfig() = \n%v, want \n%v", gotBgpPeers, tt.wantBgpPeers)
+			}
+		})
+	}
+}

pkg/kubevip/config_environment.go

@@ -6,10 +6,14 @@ import (
 	"math"
 	"math/bits"
 	"os"
+	"path/filepath"
 	"strconv"
+	"strings"
 
-	"github.com/kube-vip/kube-vip/pkg/bgp"
+	"github.com/kube-vip/kube-vip/pkg/debouncer"
 	"github.com/kube-vip/kube-vip/pkg/detector"
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	"sigs.k8s.io/yaml"
 )
 
 // ParseEnvironment - will popultate the configuration from environment variables
@@ -19,17 +23,20 @@ func ParseEnvironment(c *Config) error {
 	}
 	// Ensure that logging is set through the environment variables
 	env := os.Getenv(vipLogLevel)
-	// Set default value
-	if env == "" {
-		env = "4"
-	}
 
 	if env != "" {
-		logLevel, err := strconv.ParseUint(env, 10, 32)
+		logLevel, err := strconv.ParseInt(env, 10, 32)
 		if err != nil {
-			panic("Unable to parse environment variable [vip_loglevel], should be int")
+			return fmt.Errorf("unable to parse environment variable [vip_loglevel], should be int: %w", err)
 		}
-		c.Logging = int(logLevel)
+		c.Logging = int32(logLevel)
+	}
+
+	if env = os.Getenv(instanceName); env == "" {
+		env = os.Getenv(strings.ToUpper(instanceName))
+	}
+	if env != "" {
+		c.InstanceName = env
 	}
 
 	// Find interface
@@ -47,16 +54,37 @@ func ParseEnvironment(c *Config) error {
 		c.LoInterfaceGlobalScope = b
 	}
 
+	env = os.Getenv(vipLoseLeadership)
+	if env != "" {
+		b, err := strconv.ParseBool(env)
+		if err != nil {
+			return err
+		}
+		c.LoseLeadership = b
+	}
+
+	env = os.Getenv(vipLoseLeadershipTimeoutSeconds)
+	if env != "" {
+		i, err := strconv.ParseInt(env, 10, 32)
+		if err != nil {
+			return fmt.Errorf("parsing env var %s (value: %s): %w", vipLoseLeadershipTimeoutSeconds, env, err)
+		}
+		c.LoseLeadershipTimeoutSeconds = int(i)
+	}
 	// Find (services) interface
 	env = os.Getenv(vipServicesInterface)
 	if env != "" {
 		c.ServicesInterface = env
 	}
 
-	// Find provider configuration
-	env = os.Getenv(providerConfig)
+	// Tolerate a down interface
+	env = os.Getenv(vipAllowInterfaceNotUp)
 	if env != "" {
-		c.ProviderConfig = env
+		b, err := strconv.ParseBool(env)
+		if err != nil {
+			return err
+		}
+		c.AllowInterfaceNotUp = b
 	}
 
 	// Find Kubernetes Leader Election configuration
@@ -136,11 +164,11 @@ func ParseEnvironment(c *Config) error {
 	// Find vip port
 	env = os.Getenv(port)
 	if env != "" {
-		i, err := strconv.ParseInt(env, 10, 32)
+		i, err := strconv.ParseUint(env, 10, 16)
 		if err != nil {
 			return err
 		}
-		c.Port = int(i)
+		c.Port = uint16(i)
 	}
 
 	// Find vipDdns
@@ -242,12 +270,6 @@ func ParseEnvironment(c *Config) error {
 		}
 	}
 
-	// Find vip address cidr range
-	env = os.Getenv(vipCidr)
-	if env != "" {
-		c.VIPCIDR = env
-	}
-
 	// Find vip address subnet
 	env = os.Getenv(vipSubnet)
 	if env != "" {
@@ -305,6 +327,18 @@ func ParseEnvironment(c *Config) error {
 		c.ArpBroadcastRate = 3000
 	}
 
+	// Determine if VIP should be preserved on leadership loss
+	// true: VIP addresses remain on interface, only ARP/NDP broadcasting stops
+	// false (default): VIP addresses are deleted on leadership loss (legacy behavior)
+	env = os.Getenv(vipPreserveOnLeadershipLoss)
+	if env != "" {
+		b, err := strconv.ParseBool(env)
+		if err != nil {
+			return err
+		}
+		c.PreserveVIPOnLeadershipLoss = b
+	}
+
 	// Wireguard Mode
 	env = os.Getenv(vipWireguard)
 	if env != "" {
@@ -372,12 +406,46 @@ func ParseEnvironment(c *Config) error {
 		c.CleanRoutingTable = b
 	}
 
+	// Skip Duplicate Address Detection when adding the VIP address
+	env = os.Getenv(vipSkipDAD)
+	if env != "" {
+		b, err := strconv.ParseBool(env)
+		if err != nil {
+			return err
+		}
+		c.SkipDAD = b
+	}
+
 	// DNS mode
 	env = os.Getenv(dnsMode)
 	if env != "" {
 		c.DNSMode = env
 	}
 
+	// DHCP mode
+	env = os.Getenv(dhcpMode)
+	if env != "" {
+		c.DHCPMode = env
+	} else {
+		if c.DNSMode != "first" {
+			c.DHCPMode = c.DNSMode
+		} else {
+			c.DHCPMode = strings.ToLower(utils.IPv4Family)
+		}
+	}
+
+	// DHCP backoff attempts
+	env = os.Getenv(dhcpBackoffAttempts)
+	if env != "" {
+		tmp, err := strconv.ParseInt(env, 10, 32)
+		if err != nil {
+			return err
+		}
+		if tmp >= 0 {
+			c.DHCPBackoffAttempts = uint(tmp)
+		}
+	}
+
 	// Disable updates for services (status.LoadBalancer.Ingress will not be updated)
 	env = os.Getenv(disableServiceUpdates)
 	if env != "" {
@@ -398,6 +466,15 @@ func ParseEnvironment(c *Config) error {
 		c.EnableBGP = b
 	}
 
+	env = os.Getenv(bgpAttachIPToInterface)
+	if env != "" {
+		b, err := strconv.ParseBool(env)
+		if err != nil {
+			return err
+		}
+		c.BGPAttachIPToInterface = b
+	}
+
 	// BGP Router interface determines an interface that we can use to find an address for
 	env = os.Getenv(bgpRouterInterface)
 	if env != "" {
@@ -437,13 +514,31 @@ func ParseEnvironment(c *Config) error {
 	// Peer AS
 	env = os.Getenv(bgpPeers)
 	if env != "" {
-		peers, err := bgp.ParseBGPPeerConfig(env)
+		peers, err := ParseBGPPeerConfig(env)
 		if err != nil {
 			return err
 		}
 		c.BGPConfig.Peers = peers
 	}
 
+	// MPBGP mode
+	env = os.Getenv(mpbgpNexthop)
+	if env != "" {
+		c.BGPConfig.MpbgpNexthop = env
+	}
+
+	// MPBGP fixed IPv4
+	env = os.Getenv(mpbgpIPv4)
+	if env != "" {
+		c.BGPConfig.MpbgpIPv4 = env
+	}
+
+	// MPBGP fixed IPv6
+	env = os.Getenv(mpbgpIPv6)
+	if env != "" {
+		c.BGPConfig.MpbgpIPv6 = env
+	}
+
 	// BGP Peer mutlihop
 	env = os.Getenv(bgpMultiHop)
 	if env != "" {
@@ -498,28 +593,66 @@ func ParseEnvironment(c *Config) error {
 		c.BGPConfig.KeepaliveInterval = u64
 	}
 
-	// Enable the Equinix Metal API calls
-	env = os.Getenv(vipPacket)
+	// BGP health check options
+	env = os.Getenv(controlPlaneHealthCheckAddress)
 	if env != "" {
-		b, err := strconv.ParseBool(env)
+		c.ControlPlaneHealthCheck.Address = env
+	}
+	env = os.Getenv(controlPlaneHealthCheckPeriodSeconds)
+	if env != "" {
+		i, err := strconv.ParseInt(env, 10, 32)
+		if err != nil {
+			return fmt.Errorf("parsing env var %s (value: %s): %w", controlPlaneHealthCheckPeriodSeconds, env, err)
+		}
+		c.ControlPlaneHealthCheck.PeriodSeconds = int(i)
+	}
+	env = os.Getenv(controlPlaneHealthCheckTimeoutSeconds)
+	if env != "" {
+		i, err := strconv.ParseInt(env, 10, 32)
+		if err != nil {
+			return fmt.Errorf("parsing env var %s (value: %s): %w", controlPlaneHealthCheckTimeoutSeconds, env, err)
+		}
+		c.ControlPlaneHealthCheck.TimeoutSeconds = int(i)
+	}
+	env = os.Getenv(controlPlaneHealthCheckFailureThreshold)
+	if env != "" {
+		i, err := strconv.ParseInt(env, 10, 32)
+		if err != nil {
+			return fmt.Errorf("parsing env var %s (value: %s): %w", controlPlaneHealthCheckFailureThreshold, env, err)
+		}
+		c.ControlPlaneHealthCheck.FailureThreshold = int(i)
+	}
+	env = os.Getenv(controlPlaneHealthCheckCAPath)
+	if env != "" {
+		c.ControlPlaneHealthCheck.CAPath = env
+	}
+
+	env = os.Getenv(zebraEnable)
+	if env != "" {
+		result, err := strconv.ParseBool(env)
 		if err != nil {
 			return err
 		}
-		c.EnableMetal = b
+		c.BGPConfig.Zebra.Enabled = result
 	}
 
-	// Find the Equinix Metal project name
-	env = os.Getenv(vipPacketProject)
+	env = os.Getenv(zebraURL)
 	if env != "" {
-		// TODO - parse address net.Host()
-		c.MetalProject = env
+		c.BGPConfig.Zebra.URL = env
 	}
 
-	// Find the Equinix Metal project ID
-	env = os.Getenv(vipPacketProjectID)
+	env = os.Getenv(zebraVersion)
 	if env != "" {
-		// TODO - parse address net.Host()
-		c.MetalProjectID = env
+		u64, err := strconv.ParseUint(env, 10, 32)
+		if err != nil {
+			return err
+		}
+		c.BGPConfig.Zebra.Version = uint32(u64)
+	}
+
+	env = os.Getenv(zebraSoftwareName)
+	if env != "" {
+		c.BGPConfig.Zebra.SoftwareName = env
 	}
 
 	// Enable the load-balancer
@@ -535,11 +668,11 @@ func ParseEnvironment(c *Config) error {
 	// Find loadbalancer port
 	env = os.Getenv(lbPort)
 	if env != "" {
-		i, err := strconv.ParseInt(env, 10, 32)
+		i, err := strconv.ParseUint(env, 10, 16)
 		if err != nil {
 			return err
 		}
-		c.LoadBalancerPort = int(i)
+		c.LoadBalancerPort = uint16(i)
 	}
 
 	// Find loadbalancer forwarding method
@@ -594,19 +727,38 @@ func ParseEnvironment(c *Config) error {
 		c.EgressWithNftables = b
 	}
 
+	env = os.Getenv(perServiceElectionOnDemand)
+	if env != "" {
+		b, err := strconv.ParseBool(env)
+		if err != nil {
+			return err
+		}
+		c.PerServiceElectionOnDemand = b
+	}
+
+	// if this is set then we're enabling the internal SNAT rule that kube-vip adds to the egress chain
+	env = os.Getenv(egressEnableInternalSNAT)
+	if env != "" {
+		b, err := strconv.ParseBool(env)
+		if err != nil {
+			return err
+		}
+		c.EnableInternalSNAT = b
+	}
+
 	// check to see if we're using a specific path to the Kubernetes config file
 	env = os.Getenv(k8sConfigFile)
 	if env != "" {
 		c.K8sConfigFile = env
 	}
 
-	env = os.Getenv(enableEndpointSlices)
+	env = os.Getenv(enableEndpoints)
 	if env != "" {
 		b, err := strconv.ParseBool(env)
 		if err != nil {
 			return err
 		}
-		c.EnableEndpointSlices = b
+		c.EnableEndpoints = b
 	}
 
 	env = os.Getenv(mirrorDestInterface)
@@ -628,5 +780,364 @@ func ParseEnvironment(c *Config) error {
 		c.BackendHealthCheckInterval = int(i)
 	}
 
+	env = os.Getenv(healthCheckPort)
+	if env != "" {
+		i, err := strconv.ParseInt(env, 10, 32)
+		if err != nil {
+			return err
+		}
+		if i < 1024 {
+			return fmt.Errorf("health check port should be > 1024")
+		}
+		c.HealthCheckPort = int(i)
+	}
+
+	env = os.Getenv(enableUPNP)
+	if env != "" {
+		b, err := strconv.ParseBool(env)
+		if err != nil {
+			return err
+		}
+		c.EnableUPNP = b
+	}
+
+	if env = os.Getenv(egressClean); env == "" {
+		env = os.Getenv(strings.ToUpper(egressClean))
+	}
+	if env != "" {
+		b, err := strconv.ParseBool(env)
+		if err != nil {
+			return err
+		}
+		c.EgressClean = b
+	}
+
+	// check for configuration file path
+	env = os.Getenv(configFile)
+	if env != "" {
+		c.ConfigFile = env
+	}
+
 	return nil
 }
+
+// LoadConfigFromFile loads configuration from a JSON or YAML file
+func LoadConfigFromFile(configFilePath string) (*Config, error) {
+	if configFilePath == "" {
+		return nil, fmt.Errorf("config file path is empty")
+	}
+
+	// Check if file exists
+	if _, err := os.Stat(configFilePath); os.IsNotExist(err) {
+		return nil, fmt.Errorf("config file does not exist: %s", configFilePath)
+	}
+
+	// Read file content
+	data, err := os.ReadFile(configFilePath)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read config file %s: %v", configFilePath, err)
+	}
+
+	var config Config
+	ext := strings.ToLower(filepath.Ext(configFilePath))
+
+	switch ext {
+	case ".json":
+		err = json.Unmarshal(data, &config)
+		if err != nil {
+			return nil, fmt.Errorf("failed to parse JSON config file %s: %v", configFilePath, err)
+		}
+	case ".yaml", ".yml":
+		err = yaml.Unmarshal(data, &config)
+		if err != nil {
+			return nil, fmt.Errorf("failed to parse YAML config file %s: %v", configFilePath, err)
+		}
+	default:
+		return nil, fmt.Errorf("unsupported config file format %s. Supported formats: .json, .yaml, .yml", ext)
+	}
+
+	return &config, nil
+}
+
+// MergeConfigFromFile merges configuration loaded from file with existing config
+// Priority: command line flags > environment variables > config file
+func MergeConfigFromFile(c *Config, configFilePath string) error {
+	if configFilePath == "" {
+		return nil // No config file specified, nothing to merge
+	}
+
+	fileConfig, err := LoadConfigFromFile(configFilePath)
+	if err != nil {
+		return err
+	}
+
+	// Merge file config with existing config
+	// Only set values from file if they haven't been set by flags or env vars
+	mergeConfigValues(c, fileConfig)
+
+	return nil
+}
+
+// mergeConfigValues merges values from fileConfig into baseConfig
+// Only overwrites zero values in baseConfig
+func mergeConfigValues(baseConfig, fileConfig *Config) {
+	// Basic configuration
+	if baseConfig.Logging == 0 && fileConfig.Logging != 0 {
+		baseConfig.Logging = fileConfig.Logging
+	}
+
+	// Network configuration
+	if baseConfig.Interface == "" && fileConfig.Interface != "" {
+		baseConfig.Interface = fileConfig.Interface
+	}
+	if baseConfig.ServicesInterface == "" && fileConfig.ServicesInterface != "" {
+		baseConfig.ServicesInterface = fileConfig.ServicesInterface
+	}
+	if baseConfig.VIP == "" && fileConfig.VIP != "" {
+		baseConfig.VIP = fileConfig.VIP
+	}
+	if baseConfig.VIPSubnet == "" && fileConfig.VIPSubnet != "" {
+		baseConfig.VIPSubnet = fileConfig.VIPSubnet
+	}
+	if baseConfig.Address == "" && fileConfig.Address != "" {
+		baseConfig.Address = fileConfig.Address
+	}
+	if baseConfig.Port == 0 && fileConfig.Port != 0 {
+		baseConfig.Port = fileConfig.Port
+	}
+	if baseConfig.NodeName == "" && fileConfig.NodeName != "" {
+		baseConfig.NodeName = fileConfig.NodeName
+	}
+
+	// Boolean flags - only merge if not explicitly set
+	if !baseConfig.EnableARP && fileConfig.EnableARP {
+		baseConfig.EnableARP = fileConfig.EnableARP
+	}
+	if !baseConfig.EnableBGP && fileConfig.EnableBGP {
+		baseConfig.EnableBGP = fileConfig.EnableBGP
+	}
+	if !baseConfig.BGPAttachIPToInterface && fileConfig.BGPAttachIPToInterface {
+		baseConfig.BGPAttachIPToInterface = fileConfig.BGPAttachIPToInterface
+	}
+	if !baseConfig.EnableWireguard && fileConfig.EnableWireguard {
+		baseConfig.EnableWireguard = fileConfig.EnableWireguard
+	}
+	if !baseConfig.EnableRoutingTable && fileConfig.EnableRoutingTable {
+		baseConfig.EnableRoutingTable = fileConfig.EnableRoutingTable
+	}
+	if !baseConfig.EnableControlPlane && fileConfig.EnableControlPlane {
+		baseConfig.EnableControlPlane = fileConfig.EnableControlPlane
+	}
+	if !baseConfig.DetectControlPlane && fileConfig.DetectControlPlane {
+		baseConfig.DetectControlPlane = fileConfig.DetectControlPlane
+	}
+	if !baseConfig.EnableServices && fileConfig.EnableServices {
+		baseConfig.EnableServices = fileConfig.EnableServices
+	}
+	if !baseConfig.EnableServicesElection && fileConfig.EnableServicesElection {
+		baseConfig.EnableServicesElection = fileConfig.EnableServicesElection
+	}
+	if !baseConfig.EnableNodeLabeling && fileConfig.EnableNodeLabeling {
+		baseConfig.EnableNodeLabeling = fileConfig.EnableNodeLabeling
+	}
+	if !baseConfig.EnableLoadBalancer && fileConfig.EnableLoadBalancer {
+		baseConfig.EnableLoadBalancer = fileConfig.EnableLoadBalancer
+	}
+	if !baseConfig.DDNS && fileConfig.DDNS {
+		baseConfig.DDNS = fileConfig.DDNS
+	}
+	if !baseConfig.SingleNode && fileConfig.SingleNode {
+		baseConfig.SingleNode = fileConfig.SingleNode
+	}
+	if !baseConfig.StartAsLeader && fileConfig.StartAsLeader {
+		baseConfig.StartAsLeader = fileConfig.StartAsLeader
+	}
+	if !baseConfig.PreserveVIPOnLeadershipLoss && fileConfig.PreserveVIPOnLeadershipLoss {
+		baseConfig.PreserveVIPOnLeadershipLoss = fileConfig.PreserveVIPOnLeadershipLoss
+	}
+
+	// Service configuration
+	if baseConfig.Namespace == "" && fileConfig.Namespace != "" {
+		baseConfig.Namespace = fileConfig.Namespace
+	}
+	if baseConfig.ServiceNamespace == "" && fileConfig.ServiceNamespace != "" {
+		baseConfig.ServiceNamespace = fileConfig.ServiceNamespace
+	}
+	if baseConfig.ServicesLeaseName == "" && fileConfig.ServicesLeaseName != "" {
+		baseConfig.ServicesLeaseName = fileConfig.ServicesLeaseName
+	}
+	// LoadBalancer configuration
+	if baseConfig.LoadBalancerPort == 0 && fileConfig.LoadBalancerPort != 0 {
+		baseConfig.LoadBalancerPort = fileConfig.LoadBalancerPort
+	}
+	if baseConfig.LoadBalancerForwardingMethod == "" && fileConfig.LoadBalancerForwardingMethod != "" {
+		baseConfig.LoadBalancerForwardingMethod = fileConfig.LoadBalancerForwardingMethod
+	}
+	if baseConfig.LoadBalancerClassName == "" && fileConfig.LoadBalancerClassName != "" {
+		baseConfig.LoadBalancerClassName = fileConfig.LoadBalancerClassName
+	}
+
+	// Routing Table configuration
+	if baseConfig.RoutingTableID == 0 && fileConfig.RoutingTableID != 0 {
+		baseConfig.RoutingTableID = fileConfig.RoutingTableID
+	}
+	if baseConfig.RoutingTableType == 0 && fileConfig.RoutingTableType != 0 {
+		baseConfig.RoutingTableType = fileConfig.RoutingTableType
+	}
+	if baseConfig.RoutingProtocol == 0 && fileConfig.RoutingProtocol != 0 {
+		baseConfig.RoutingProtocol = fileConfig.RoutingProtocol
+	}
+
+	// BGP configuration
+	mergeBGPConfig(&baseConfig.BGPConfig, &fileConfig.BGPConfig)
+
+	// Kubernetes configuration
+	if baseConfig.K8sConfigFile == "" && fileConfig.K8sConfigFile != "" {
+		baseConfig.K8sConfigFile = fileConfig.K8sConfigFile
+	}
+
+	// Leader Election configuration
+	mergeLeaderElectionConfig(&baseConfig.KubernetesLeaderElection, &fileConfig.KubernetesLeaderElection)
+
+	// BGP health check configuration
+	mergeHealthCheck(&baseConfig.ControlPlaneHealthCheck, &fileConfig.ControlPlaneHealthCheck)
+
+	// Prometheus configuration
+	if baseConfig.PrometheusHTTPServer == "" && fileConfig.PrometheusHTTPServer != "" {
+		baseConfig.PrometheusHTTPServer = fileConfig.PrometheusHTTPServer
+	}
+
+	// DNS configuration
+	if baseConfig.DNSMode == "" && fileConfig.DNSMode != "" {
+		baseConfig.DNSMode = fileConfig.DNSMode
+	}
+
+	// DHCP configuration - mode
+	if baseConfig.DHCPMode == "" && fileConfig.DHCPMode != "" {
+		baseConfig.DHCPMode = fileConfig.DHCPMode
+	}
+
+	// DHCP configuration - backoff attempts
+	if baseConfig.DHCPBackoffAttempts == DefaultDHCPBackoffAttempts && fileConfig.DHCPBackoffAttempts != DefaultDHCPBackoffAttempts {
+		baseConfig.DHCPBackoffAttempts = fileConfig.DHCPBackoffAttempts
+	}
+
+	// Health check configuration (HTTP listener for kube-vip readiness)
+	if baseConfig.HealthCheckPort == 0 && fileConfig.HealthCheckPort != 0 {
+		baseConfig.HealthCheckPort = fileConfig.HealthCheckPort
+	}
+
+	// Instance configuration
+	if baseConfig.InstanceName == "" && fileConfig.InstanceName != "" {
+		baseConfig.InstanceName = fileConfig.InstanceName
+	}
+
+	// Egress configuration
+	if baseConfig.EgressPodCidr == "" && fileConfig.EgressPodCidr != "" {
+		baseConfig.EgressPodCidr = fileConfig.EgressPodCidr
+	}
+	if baseConfig.EgressServiceCidr == "" && fileConfig.EgressServiceCidr != "" {
+		baseConfig.EgressServiceCidr = fileConfig.EgressServiceCidr
+	}
+	// Mirror configuration
+	if baseConfig.MirrorDestInterface == "" && fileConfig.MirrorDestInterface != "" {
+		baseConfig.MirrorDestInterface = fileConfig.MirrorDestInterface
+	}
+
+	// Iptables configuration
+	if baseConfig.IptablesBackend == "" && fileConfig.IptablesBackend != "" {
+		baseConfig.IptablesBackend = fileConfig.IptablesBackend
+	}
+
+	// Backend health check interval
+	if baseConfig.BackendHealthCheckInterval == 0 && fileConfig.BackendHealthCheckInterval != 0 {
+		baseConfig.BackendHealthCheckInterval = fileConfig.BackendHealthCheckInterval
+	}
+
+	// ARP broadcast rate
+	if baseConfig.ArpBroadcastRate == 0 && fileConfig.ArpBroadcastRate != 0 {
+		baseConfig.ArpBroadcastRate = fileConfig.ArpBroadcastRate
+	}
+
+	// Annotations
+	if baseConfig.Annotations == "" && fileConfig.Annotations != "" {
+		baseConfig.Annotations = fileConfig.Annotations
+	}
+
+	// Load balancers slice
+	if len(baseConfig.LoadBalancers) == 0 && len(fileConfig.LoadBalancers) > 0 {
+		baseConfig.LoadBalancers = fileConfig.LoadBalancers
+	}
+
+	// Debounce time for watch events
+	if baseConfig.DebounceTime == debouncer.DefaultTime && fileConfig.DebounceTime != debouncer.DefaultTime {
+		baseConfig.DebounceTime = fileConfig.DebounceTime
+	}
+
+	if baseConfig.LoseLeadershipTimeoutSeconds == 0 && fileConfig.LoseLeadershipTimeoutSeconds != 0 {
+		baseConfig.LoseLeadershipTimeoutSeconds = fileConfig.LoseLeadershipTimeoutSeconds
+	}
+}
+
+// mergeBGPConfig merges BGP configuration
+func mergeBGPConfig(base, file *BGPConfig) {
+	if base.RouterID == "" && file.RouterID != "" {
+		base.RouterID = file.RouterID
+	}
+	if base.AS == 0 && file.AS != 0 {
+		base.AS = file.AS
+	}
+	if base.SourceIF == "" && file.SourceIF != "" {
+		base.SourceIF = file.SourceIF
+	}
+	if base.SourceIP == "" && file.SourceIP != "" {
+		base.SourceIP = file.SourceIP
+	}
+	if base.HoldTime == 0 && file.HoldTime != 0 {
+		base.HoldTime = file.HoldTime
+	}
+	if base.KeepaliveInterval == 0 && file.KeepaliveInterval != 0 {
+		base.KeepaliveInterval = file.KeepaliveInterval
+	}
+	if len(base.Peers) == 0 && len(file.Peers) > 0 {
+		base.Peers = file.Peers
+	}
+}
+
+// mergeLeaderElectionConfig merges leader election configuration
+func mergeLeaderElectionConfig(base, file *KubernetesLeaderElection) {
+	if base.LeaseName == "" && file.LeaseName != "" {
+		base.LeaseName = file.LeaseName
+	}
+	if base.LeaseDuration == 0 && file.LeaseDuration != 0 {
+		base.LeaseDuration = file.LeaseDuration
+	}
+	if base.RenewDeadline == 0 && file.RenewDeadline != 0 {
+		base.RenewDeadline = file.RenewDeadline
+	}
+	if base.RetryPeriod == 0 && file.RetryPeriod != 0 {
+		base.RetryPeriod = file.RetryPeriod
+	}
+	if len(base.LeaseAnnotations) == 0 && len(file.LeaseAnnotations) > 0 {
+		base.LeaseAnnotations = file.LeaseAnnotations
+	}
+}
+
+// mergeHealthCheck merges HTTP health check configuration for BGP route advertisement.
+func mergeHealthCheck(base, file *HealthCheck) {
+	if base.Address == "" && file.Address != "" {
+		base.Address = file.Address
+	}
+	if base.PeriodSeconds == 0 && file.PeriodSeconds != 0 {
+		base.PeriodSeconds = file.PeriodSeconds
+	}
+	if base.TimeoutSeconds == 0 && file.TimeoutSeconds != 0 {
+		base.TimeoutSeconds = file.TimeoutSeconds
+	}
+	if base.FailureThreshold == 0 && file.FailureThreshold != 0 {
+		base.FailureThreshold = file.FailureThreshold
+	}
+	if base.CAPath == "" && file.CAPath != "" {
+		base.CAPath = file.CAPath
+	}
+}

pkg/kubevip/config_environment_test.go

@@ -0,0 +1,41 @@
+package kubevip
+
+import (
+	"testing"
+)
+
+func TestParseEnvironmentSkipDAD(t *testing.T) {
+	cases := []struct {
+		name    string
+		value   string
+		want    bool
+		wantErr bool
+	}{
+		{name: "unset keeps default false", value: "", want: false},
+		{name: "true enables", value: "true", want: true},
+		{name: "false disables", value: "false", want: false},
+		{name: "garbage errors", value: "not-a-bool", wantErr: true},
+	}
+
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			if tc.value != "" {
+				t.Setenv(vipSkipDAD, tc.value)
+			}
+			c := &Config{}
+			err := ParseEnvironment(c)
+			if tc.wantErr {
+				if err == nil {
+					t.Fatal("expected an error, got nil")
+				}
+				return
+			}
+			if err != nil {
+				t.Fatalf("unexpected error: %v", err)
+			}
+			if c.SkipDAD != tc.want {
+				t.Fatalf("SkipDAD = %v, want %v", c.SkipDAD, tc.want)
+			}
+		})
+	}
+}

pkg/kubevip/config_envvar.go

@@ -9,38 +9,50 @@ const (
 	// vip_arpRate - defines the rate of gARP broadcasts
 	vipArpRate = "vip_arpRate"
 
+	// vipPreserveOnLeadershipLoss - if true, VIP addresses will remain on interface when leadership is lost
+	vipPreserveOnLeadershipLoss = "vip_preserve_on_leadership_loss"
+
 	// vipLeaderElection - defines if the kubernetes algorithm should be used
 	vipLeaderElection = "vip_leaderelection"
 
 	// vipLeaseName - defines the name of the lease lock
 	vipLeaseName = "vip_leasename"
 
-	// vipLeaderElection - defines if the kubernetes algorithm should be used
+	// vipLeaseDuration - defines how long the current leader is considered valid
 	vipLeaseDuration = "vip_leaseduration"
 
-	// vipLeaderElection - defines if the kubernetes algorithm should be used
+	// vipRenewDeadline - defines how long the leader has to renew the lease before losing leadership
 	vipRenewDeadline = "vip_renewdeadline"
 
-	// vipLeaderElection - defines if the kubernetes algorithm should be used
+	// vipRetryPeriod - defines the time between attempts to acquire/renew the lease
 	vipRetryPeriod = "vip_retryperiod"
 
-	// vipLeaderElection - defines the annotations given to the lease lock
+	// vipLeaseAnnotations - defines the annotations given to the lease lock
 	vipLeaseAnnotations = "vip_leaseannotations"
 
 	// vipLogLevel - defines the level of logging to produce (5 being the most verbose)
 	vipLogLevel = "vip_loglevel"
 
+	// instanceName identifies this kube-vip deployment when naming host-global resources
+	instanceName = "instance_name"
+
 	// vipInterface - defines the interface that the vip should bind too
 	vipInterface = "vip_interface"
 
+	// vipLoseLeadership - defines if leader should lose leadership if network interface is down
+	vipLoseLeadership = "vip_loseleadership"
+
+	// vipLoseLeadershipTimeout - defines the timeout for lose leadership
+	vipLoseLeadershipTimeoutSeconds = "vip_loseleadership_timeout_seconds"
+
 	// vipInterfaceLoGlobal - defines if the lo interface (if used) should have a global scope
 	vipInterfaceLoGlobal = "vip_interfaceloglobal"
 
 	// vipServicesInterface - defines the interface that the service vips should bind too
 	vipServicesInterface = "vip_servicesinterface"
 
-	// vipCidr - defines the cidr that the vip will use (for BGP)
-	vipCidr = "vip_cidr"
+	// vipAllowInterfaceNotUp - defines if kube-vip should tolerate a down interface
+	vipAllowInterfaceNotUp = "vip_allow_interface_not_up"
 
 	// vipSubnet - defines the subnet that the vip will use
 	vipSubnet = "vip_subnet"
@@ -54,6 +66,11 @@ const (
 	// egressWithNftables - enables using nftables over iptables
 	egressWithNftables = "egress_withnftables"
 
+	// perServiceElectionOnDemand - enables kube-vip to use per-service election for annotated services
+	perServiceElectionOnDemand = "per_service_election_on_demand"
+
+	// egressEnableInternalSNAT - enables the internal SNAT rule that kube-vip adds to the egress chain
+	egressEnableInternalSNAT = "egress_enableinternalsnat"
 	/////////////////////////////////////
 	// TO DO:
 	// Determine how to tidy this mess up
@@ -86,20 +103,10 @@ const (
 	// vipStartLeader - will start this instance as the leader of the cluster
 	vipStartLeader = "vip_startleader"
 
-	// vipPacket defines that the packet API will be used for EIP
-	vipPacket = "vip_packet"
-
-	// vipPacketProject defines which project within Packet to use
-	vipPacketProject = "vip_packetproject"
-
-	// vipPacketProjectID defines which projectID within Packet to use
-	vipPacketProjectID = "vip_packetprojectid"
-
-	// providerConfig defines a path to a configuration that should be parsed
-	providerConfig = "provider_config"
-
 	// bgpEnable defines if BGP should be enabled
 	bgpEnable = "bgp_enable"
+	// bgpAttachIPToInterface defines if BGP service VIPs should be assigned to the configured interface
+	bgpAttachIPToInterface = "bgp_attach_ip_to_interface"
 	// bgpRouterID defines the routerID for the BGP server
 	bgpRouterID = "bgp_routerid"
 	// bgpRouterInterface defines the interface that we can find the address for
@@ -124,6 +131,32 @@ const (
 	bgpHoldTime = "bgp_hold_time"
 	// bgpKeepaliveInterval defines bgp timers keepalive interval
 	bgpKeepaliveInterval = "bgp_keepalive_interval"
+	// controlPlaneHealthCheckAddress defines the URL for control-plane health checks (BGP route withdrawal)
+	controlPlaneHealthCheckAddress = "control_plane_health_check_address"
+	// controlPlaneHealthCheckPeriodSeconds defines the period between control-plane health checks
+	controlPlaneHealthCheckPeriodSeconds = "control_plane_health_check_period_seconds"
+	// controlPlaneHealthCheckTimeoutSeconds defines the timeout for each control-plane health check request
+	controlPlaneHealthCheckTimeoutSeconds = "control_plane_health_check_timeout_seconds"
+	// controlPlaneHealthCheckFailureThreshold defines consecutive failures before BGP route withdrawal
+	controlPlaneHealthCheckFailureThreshold = "control_plane_health_check_failure_threshold"
+	// controlPlaneHealthCheckCAPath defines the path to a CA certificate for control-plane health check TLS verification
+	controlPlaneHealthCheckCAPath = "control_plane_health_check_ca_path"
+
+	// zebraEnable defines if Zebra integraton should be enabled
+	zebraEnable = "zebra_enable"
+	// zebraUrl specifies path to the unix domain socket for connecting to Zebra daemon
+	zebraURL = "zebra_url"
+	// zebraVersion specifies Zebra API Version
+	zebraVersion = "zebra_version"
+	// zebraSoftwareName specifies Software Name for Zebra
+	zebraSoftwareName = "zebra_software_name"
+
+	// mpbgpNexthop defines MPBGP mode
+	mpbgpNexthop = "mpbgp_nexthop"
+	// mpbgpIPv4 defines fixed IPv4 to be used with MPBGP
+	mpbgpIPv4 = "mpbgp_ipv4"
+	// mpbgpIPv6 defines fixed IPv6 to be used with MPBGP
+	mpbgpIPv6 = "mpbgp_ipv6"
 
 	// vipWireguard - defines if wireguard will be used for vips
 	vipWireguard = "vip_wireguard" //nolint
@@ -148,6 +181,9 @@ const (
 	// vipCleanRoutingTable - defines if routing table will be cleaned of redundant routes on kube-vip's start
 	vipCleanRoutingTable = "vip_cleanroutingtable" //nolint
 
+	// vipSkipDAD - defines if Duplicate Address Detection is skipped when adding the VIP address (IFA_F_NODAD)
+	vipSkipDAD = "vip_skipdad" //nolint
+
 	// cpNamespace defines the namespace the control plane pods will run in
 	cpNamespace = "cp_namespace"
 
@@ -208,11 +244,17 @@ const (
 	// dnsMode defines mode that DNS lookup will be performed with (first, ipv4, ipv6, dual)
 	dnsMode = "dns_mode"
 
+	// dhcpMode defines mode that DHCP lookup will be performed with (ipv4, ipv6, dual)
+	dhcpMode = "dhcp_mode"
+
+	// dhcpBackoffAttempts defines how many times DHCP client will try to obtain an IP address
+	dhcpBackoffAttempts = "dhcp_backoff_attempts"
+
 	// disableServiceUpdates disables service updating
 	disableServiceUpdates = "disable_service_updates"
 
-	// enableEndpointSlices enables use of EndpointSlices instead of Endpoints
-	enableEndpointSlices = "enable_endpointslices"
+	// enableEndpoints enables use of Endpoints instead of EndpointSlices
+	enableEndpoints = "enable_endpoints"
 
 	// mirrorDestInterface is the network interface where all traffics that go through service interface
 	// will be mirrored to. The source interface is ServicesInterface by default, fall back to Interface if not set.
@@ -224,4 +266,19 @@ const (
 
 	// backendHealthCheckInterval Interval in seconds for checking backend health.
 	backendHealthCheckInterval = "backend_health_check_interval"
+
+	// healthCheckPort, if set to non-zero will be the port the health check will listen on
+	healthCheckPort = "health_check_port"
+
+	// enableUPNP enables UPNP functions
+	enableUPNP = "enable_upnp"
+
+	// egressClean enables egress cleaning on kube-vip's start
+	egressClean = "egress_clean"
+
+	// configFile defines the path to a JSON/YAML configuration file
+	configFile = "config_file"
+
+	// debounceTime defines what time should the event debouncer wait for events
+	debounceTime = "debounce_time"
 )

pkg/kubevip/config_file_test.go

@@ -0,0 +1,569 @@
+package kubevip
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+)
+
+func TestLoadConfigFromFile(t *testing.T) {
+	// Create temporary directory for test files
+	tmpDir, err := os.MkdirTemp("", "kube-vip-config-test")
+	if err != nil {
+		t.Fatalf("Failed to create temp dir: %v", err)
+	}
+	defer os.RemoveAll(tmpDir)
+
+	tests := []struct {
+		name           string
+		filename       string
+		content        string
+		expectedConfig *Config
+		wantErr        bool
+		errContains    string
+	}{
+		{
+			name:     "Valid YAML config",
+			filename: "config.yaml",
+			content: `
+logging: 2
+enableARP: true
+enableControlPlane: true
+enableServices: true
+address: "192.168.1.100"
+port: 6443
+interface: "eth0"
+namespace: "kube-system"
+instanceName: "release_a"
+vipSubnet: "192.168.1.0/24"
+leaseName: "test-lease"
+leaseDuration: 15
+renewDeadline: 10
+retryPeriod: 2
+prometheusHTTPServer: ":2112"
+`,
+			expectedConfig: &Config{
+				Logging:              2,
+				EnableARP:            true,
+				EnableControlPlane:   true,
+				EnableServices:       true,
+				Address:              "192.168.1.100",
+				Port:                 6443,
+				Interface:            "eth0",
+				Namespace:            "kube-system",
+				InstanceName:         "release_a",
+				VIPSubnet:            "192.168.1.0/24",
+				PrometheusHTTPServer: ":2112",
+				KubernetesLeaderElection: KubernetesLeaderElection{
+					LeaseName:     "test-lease",
+					LeaseDuration: 15,
+					RenewDeadline: 10,
+					RetryPeriod:   2,
+				},
+			},
+			wantErr: false,
+		},
+		{
+			name:     "Valid JSON config",
+			filename: "config.json",
+			content: `{
+  "logging": 3,
+  "enableBGP": true,
+  "enableServices": true,
+  "address": "10.0.0.100",
+  "port": 8443,
+  "interface": "ens192",
+  "namespace": "kube-system",
+  "loadBalancers": [
+    {
+      "name": "control-plane",
+      "ports": [
+        {
+          "type": "TCP",
+          "port": 6443
+        }
+      ],
+      "bindToVip": true,
+      "forwardingMethod": "local"
+    }
+  ]
+}`,
+			expectedConfig: &Config{
+				Logging:        3,
+				EnableBGP:      true,
+				EnableServices: true,
+				Address:        "10.0.0.100",
+				Port:           8443,
+				Interface:      "ens192",
+				Namespace:      "kube-system",
+				LoadBalancers: []LoadBalancer{
+					{
+						Name: "control-plane",
+						Ports: []Port{
+							{
+								Type: "TCP",
+								Port: 6443,
+							},
+						},
+						BindToVip:        true,
+						ForwardingMethod: "local",
+					},
+				},
+			},
+			wantErr: false,
+		},
+		{
+			name:     "Complex BGP config",
+			filename: "bgp-config.yaml",
+			content: `
+enableBGP: true
+bgpConfig:
+  routerID: "192.168.1.1"
+  as: 65000
+  sourceIF: "eth0"
+  holdTime: 60
+  keepaliveInterval: 20
+  peers:
+    - address: "192.168.1.2"
+      as: 65001
+      port: 179
+      multiHop: false
+    - address: "192.168.1.3"
+      as: 65002
+      port: 179
+      multiHop: true
+`,
+			expectedConfig: &Config{
+				EnableBGP: true,
+				BGPConfig: BGPConfig{
+					RouterID:          "192.168.1.1",
+					AS:                65000,
+					SourceIF:          "eth0",
+					HoldTime:          60,
+					KeepaliveInterval: 20,
+					Peers: []BGPPeer{
+						{
+							Address:  "192.168.1.2",
+							AS:       65001,
+							Port:     179,
+							MultiHop: false,
+						},
+						{
+							Address:  "192.168.1.3",
+							AS:       65002,
+							Port:     179,
+							MultiHop: true,
+						},
+					},
+				},
+			},
+			wantErr: false,
+		},
+		{
+			name:        "Invalid JSON",
+			filename:    "invalid.json",
+			content:     `{"logging": 2, "invalid": }`,
+			wantErr:     true,
+			errContains: "failed to parse JSON config file",
+		},
+		{
+			name:        "Invalid YAML",
+			filename:    "invalid.yaml",
+			content:     "logging: 2\ninvalid: [unclosed",
+			wantErr:     true,
+			errContains: "failed to parse YAML config file",
+		},
+		{
+			name:        "Unsupported format",
+			filename:    "config.txt",
+			content:     "logging=2",
+			wantErr:     true,
+			errContains: "unsupported config file format",
+		},
+		{
+			name:        "Empty path",
+			filename:    "",
+			content:     "",
+			wantErr:     true,
+			errContains: "config file path is empty",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			var filePath string
+			if tt.filename != "" {
+				filePath = filepath.Join(tmpDir, tt.filename)
+				if err := os.WriteFile(filePath, []byte(tt.content), 0600); err != nil {
+					t.Fatalf("Failed to write test file: %v", err)
+				}
+			}
+
+			config, err := LoadConfigFromFile(filePath)
+
+			if tt.wantErr {
+				if err == nil {
+					t.Errorf("LoadConfigFromFile() expected error, got nil")
+					return
+				}
+				if tt.errContains != "" && !containsString(err.Error(), tt.errContains) {
+					t.Errorf("LoadConfigFromFile() error = %v, expected to contain %v", err, tt.errContains)
+				}
+				return
+			}
+
+			if err != nil {
+				t.Errorf("LoadConfigFromFile() unexpected error = %v", err)
+				return
+			}
+
+			if config == nil {
+				t.Errorf("LoadConfigFromFile() returned nil config")
+				return
+			}
+
+			// Compare key fields
+			if config.Logging != tt.expectedConfig.Logging {
+				t.Errorf("Logging = %v, expected %v", config.Logging, tt.expectedConfig.Logging)
+			}
+			if config.EnableARP != tt.expectedConfig.EnableARP {
+				t.Errorf("EnableARP = %v, expected %v", config.EnableARP, tt.expectedConfig.EnableARP)
+			}
+			if config.EnableBGP != tt.expectedConfig.EnableBGP {
+				t.Errorf("EnableBGP = %v, expected %v", config.EnableBGP, tt.expectedConfig.EnableBGP)
+			}
+			if config.Address != tt.expectedConfig.Address {
+				t.Errorf("Address = %v, expected %v", config.Address, tt.expectedConfig.Address)
+			}
+			if config.Port != tt.expectedConfig.Port {
+				t.Errorf("Port = %v, expected %v", config.Port, tt.expectedConfig.Port)
+			}
+			if config.Interface != tt.expectedConfig.Interface {
+				t.Errorf("Interface = %v, expected %v", config.Interface, tt.expectedConfig.Interface)
+			}
+			if config.InstanceName != tt.expectedConfig.InstanceName {
+				t.Errorf("InstanceName = %v, expected %v", config.InstanceName, tt.expectedConfig.InstanceName)
+			}
+
+			// Test BGP config if present
+			if tt.expectedConfig.EnableBGP {
+				if config.BGPConfig.RouterID != tt.expectedConfig.BGPConfig.RouterID {
+					t.Errorf("BGPConfig.RouterID = %v, expected %v", config.BGPConfig.RouterID, tt.expectedConfig.BGPConfig.RouterID)
+				}
+				if config.BGPConfig.AS != tt.expectedConfig.BGPConfig.AS {
+					t.Errorf("BGPConfig.AS = %v, expected %v", config.BGPConfig.AS, tt.expectedConfig.BGPConfig.AS)
+				}
+				if len(config.BGPConfig.Peers) != len(tt.expectedConfig.BGPConfig.Peers) {
+					t.Errorf("BGPConfig.Peers length = %v, expected %v", len(config.BGPConfig.Peers), len(tt.expectedConfig.BGPConfig.Peers))
+				}
+			}
+
+			// Test LoadBalancers if present
+			if len(tt.expectedConfig.LoadBalancers) > 0 {
+				if len(config.LoadBalancers) != len(tt.expectedConfig.LoadBalancers) {
+					t.Errorf("LoadBalancers length = %v, expected %v", len(config.LoadBalancers), len(tt.expectedConfig.LoadBalancers))
+				}
+			}
+		})
+	}
+}
+
+func TestMergeConfigFromFile(t *testing.T) {
+	// Create temporary directory for test files
+	tmpDir, err := os.MkdirTemp("", "kube-vip-merge-test")
+	if err != nil {
+		t.Fatalf("Failed to create temp dir: %v", err)
+	}
+	defer os.RemoveAll(tmpDir)
+
+	// Create test config file
+	configFile := filepath.Join(tmpDir, "test-config.yaml")
+	configContent := `
+logging: 3
+enableARP: true
+enableServices: true
+address: "192.168.1.200"
+port: 6443
+interface: "eth1"
+namespace: "test-namespace"
+leaseName: "file-lease"
+leaseDuration: 20
+prometheusHTTPServer: ":3000"
+`
+	if err := os.WriteFile(configFile, []byte(configContent), 0600); err != nil {
+		t.Fatalf("Failed to write test config file: %v", err)
+	}
+
+	tests := []struct {
+		name           string
+		baseConfig     *Config
+		configFilePath string
+		expected       *Config
+		wantErr        bool
+	}{
+		{
+			name:           "Merge with empty base config",
+			baseConfig:     &Config{},
+			configFilePath: configFile,
+			expected: &Config{
+				Logging:              3,
+				EnableARP:            true,
+				EnableServices:       true,
+				Address:              "192.168.1.200",
+				Port:                 6443,
+				Interface:            "eth1",
+				Namespace:            "test-namespace",
+				PrometheusHTTPServer: ":3000",
+				KubernetesLeaderElection: KubernetesLeaderElection{
+					LeaseName:     "file-lease",
+					LeaseDuration: 20,
+				},
+			},
+			wantErr: false,
+		},
+		{
+			name: "Merge respects existing values (priority test)",
+			baseConfig: &Config{
+				Logging:   5,      // Should not be overridden
+				Port:      8443,   // Should not be overridden
+				Interface: "eth0", // Should not be overridden
+			},
+			configFilePath: configFile,
+			expected: &Config{
+				Logging:              5,                // From base (higher priority)
+				EnableARP:            true,             // From file
+				EnableServices:       true,             // From file
+				Address:              "192.168.1.200",  // From file
+				Port:                 8443,             // From base (higher priority)
+				Interface:            "eth0",           // From base (higher priority)
+				Namespace:            "test-namespace", // From file
+				PrometheusHTTPServer: ":3000",          // From file
+				KubernetesLeaderElection: KubernetesLeaderElection{
+					LeaseName:     "file-lease", // From file
+					LeaseDuration: 20,           // From file
+				},
+			},
+			wantErr: false,
+		},
+		{
+			name: "Empty config file path",
+			baseConfig: &Config{
+				Logging: 1,
+			},
+			configFilePath: "",
+			expected: &Config{
+				Logging: 1, // Unchanged
+			},
+			wantErr: false,
+		},
+		{
+			name: "Non-existent config file",
+			baseConfig: &Config{
+				Logging: 1,
+			},
+			configFilePath: "/non/existent/file.yaml",
+			expected: &Config{
+				Logging: 1,
+			},
+			wantErr: true,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			err := MergeConfigFromFile(tt.baseConfig, tt.configFilePath)
+
+			if tt.wantErr {
+				if err == nil {
+					t.Errorf("MergeConfigFromFile() expected error, got nil")
+				}
+				return
+			}
+
+			if err != nil {
+				t.Errorf("MergeConfigFromFile() unexpected error = %v", err)
+				return
+			}
+
+			// Compare key fields
+			if tt.baseConfig.Logging != tt.expected.Logging {
+				t.Errorf("Logging = %v, expected %v", tt.baseConfig.Logging, tt.expected.Logging)
+			}
+			if tt.baseConfig.EnableARP != tt.expected.EnableARP {
+				t.Errorf("EnableARP = %v, expected %v", tt.baseConfig.EnableARP, tt.expected.EnableARP)
+			}
+			if tt.baseConfig.Address != tt.expected.Address {
+				t.Errorf("Address = %v, expected %v", tt.baseConfig.Address, tt.expected.Address)
+			}
+			if tt.baseConfig.Port != tt.expected.Port {
+				t.Errorf("Port = %v, expected %v", tt.baseConfig.Port, tt.expected.Port)
+			}
+			if tt.baseConfig.Interface != tt.expected.Interface {
+				t.Errorf("Interface = %v, expected %v", tt.baseConfig.Interface, tt.expected.Interface)
+			}
+			if tt.baseConfig.Namespace != tt.expected.Namespace {
+				t.Errorf("Namespace = %v, expected %v", tt.baseConfig.Namespace, tt.expected.Namespace)
+			}
+			if tt.baseConfig.PrometheusHTTPServer != tt.expected.PrometheusHTTPServer {
+				t.Errorf("PrometheusHTTPServer = %v, expected %v", tt.baseConfig.PrometheusHTTPServer, tt.expected.PrometheusHTTPServer)
+			}
+		})
+	}
+}
+
+func TestMergeConfigValues(t *testing.T) {
+	tests := []struct {
+		name         string
+		baseConfig   *Config
+		fileConfig   *Config
+		expectedBase *Config
+	}{
+		{
+			name: "Merge basic configuration",
+			baseConfig: &Config{
+				Logging: 5, // Should not be overridden
+				Port:    0, // Should be overridden
+			},
+			fileConfig: &Config{
+				Logging:      2,
+				Port:         6443,
+				Interface:    "eth0",
+				Address:      "192.168.1.100",
+				InstanceName: "release_a",
+			},
+			expectedBase: &Config{
+				Logging:      5,               // From base (non-zero)
+				Port:         6443,            // From file (base was zero)
+				Interface:    "eth0",          // From file (base was empty)
+				Address:      "192.168.1.100", // From file (base was empty)
+				InstanceName: "release_a",     // From file (base was empty)
+			},
+		},
+		{
+			name: "Merge boolean flags",
+			baseConfig: &Config{
+				EnableARP: true, // Should not be overridden
+			},
+			fileConfig: &Config{
+				EnableARP:       false, // Should not override true
+				EnableBGP:       true,  // Should be set
+				EnableServices:  true,  // Should be set
+				EnableWireguard: false, // Should not be set (false doesn't override false)
+			},
+			expectedBase: &Config{
+				EnableARP:       true,  // From base (true has priority)
+				EnableBGP:       true,  // From file
+				EnableServices:  true,  // From file
+				EnableWireguard: false, // Remains false
+			},
+		},
+		{
+			name: "Merge BGP configuration",
+			baseConfig: &Config{
+				BGPConfig: BGPConfig{
+					RouterID: "1.1.1.1", // Should not be overridden
+				},
+			},
+			fileConfig: &Config{
+				BGPConfig: BGPConfig{
+					RouterID:          "2.2.2.2", // Should not override
+					AS:                65000,     // Should be set
+					SourceIF:          "eth0",    // Should be set
+					HoldTime:          30,        // Should be set
+					KeepaliveInterval: 10,        // Should be set
+				},
+			},
+			expectedBase: &Config{
+				BGPConfig: BGPConfig{
+					RouterID:          "1.1.1.1", // From base (non-empty)
+					AS:                65000,     // From file (base was zero)
+					SourceIF:          "eth0",    // From file (base was empty)
+					HoldTime:          30,        // From file (base was zero)
+					KeepaliveInterval: 10,        // From file (base was zero)
+				},
+			},
+		},
+		{
+			name: "Merge leader election configuration",
+			baseConfig: &Config{
+				KubernetesLeaderElection: KubernetesLeaderElection{
+					LeaseName: "base-lease", // Should not be overridden
+				},
+			},
+			fileConfig: &Config{
+				KubernetesLeaderElection: KubernetesLeaderElection{
+					LeaseName:     "file-lease", // Should not override
+					LeaseDuration: 15,           // Should be set
+					RenewDeadline: 10,           // Should be set
+					RetryPeriod:   2,            // Should be set
+				},
+			},
+			expectedBase: &Config{
+				KubernetesLeaderElection: KubernetesLeaderElection{
+					LeaseName:     "base-lease", // From base (non-empty)
+					LeaseDuration: 15,           // From file (base was zero)
+					RenewDeadline: 10,           // From file (base was zero)
+					RetryPeriod:   2,            // From file (base was zero)
+				},
+			},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			mergeConfigValues(tt.baseConfig, tt.fileConfig)
+
+			// Compare results
+			if tt.baseConfig.Logging != tt.expectedBase.Logging {
+				t.Errorf("Logging = %v, expected %v", tt.baseConfig.Logging, tt.expectedBase.Logging)
+			}
+			if tt.baseConfig.Port != tt.expectedBase.Port {
+				t.Errorf("Port = %v, expected %v", tt.baseConfig.Port, tt.expectedBase.Port)
+			}
+			if tt.baseConfig.Interface != tt.expectedBase.Interface {
+				t.Errorf("Interface = %v, expected %v", tt.baseConfig.Interface, tt.expectedBase.Interface)
+			}
+			if tt.baseConfig.InstanceName != tt.expectedBase.InstanceName {
+				t.Errorf("InstanceName = %v, expected %v", tt.baseConfig.InstanceName, tt.expectedBase.InstanceName)
+			}
+			if tt.baseConfig.EnableARP != tt.expectedBase.EnableARP {
+				t.Errorf("EnableARP = %v, expected %v", tt.baseConfig.EnableARP, tt.expectedBase.EnableARP)
+			}
+			if tt.baseConfig.EnableBGP != tt.expectedBase.EnableBGP {
+				t.Errorf("EnableBGP = %v, expected %v", tt.baseConfig.EnableBGP, tt.expectedBase.EnableBGP)
+			}
+			if tt.baseConfig.BGPConfig.RouterID != tt.expectedBase.BGPConfig.RouterID {
+				t.Errorf("BGPConfig.RouterID = %v, expected %v", tt.baseConfig.BGPConfig.RouterID, tt.expectedBase.BGPConfig.RouterID)
+			}
+			if tt.baseConfig.BGPConfig.AS != tt.expectedBase.BGPConfig.AS {
+				t.Errorf("BGPConfig.AS = %v, expected %v", tt.baseConfig.BGPConfig.AS, tt.expectedBase.BGPConfig.AS)
+			}
+			if tt.baseConfig.KubernetesLeaderElection.LeaseName != tt.expectedBase.KubernetesLeaderElection.LeaseName {
+				t.Errorf("KubernetesLeaderElection.LeaseName = %v, expected %v", tt.baseConfig.KubernetesLeaderElection.LeaseName, tt.expectedBase.KubernetesLeaderElection.LeaseName)
+			}
+		})
+	}
+}
+
+func TestLoadConfigFromFile_FileNotExists(t *testing.T) {
+	_, err := LoadConfigFromFile("/non/existent/path/config.yaml")
+	if err == nil {
+		t.Error("LoadConfigFromFile() expected error for non-existent file, got nil")
+	}
+	if !containsString(err.Error(), "config file does not exist") {
+		t.Errorf("LoadConfigFromFile() error = %v, expected to contain 'config file does not exist'", err)
+	}
+}
+
+// Helper function to check if a string contains a substring
+func containsString(str, substr string) bool {
+	return len(str) >= len(substr) && (str == substr || len(substr) == 0 ||
+		(len(substr) > 0 && func() bool {
+			for i := 0; i <= len(str)-len(substr); i++ {
+				if str[i:i+len(substr)] == substr {
+					return true
+				}
+			}
+			return false
+		}()))
+}

pkg/kubevip/config_generator.go

@@ -2,23 +2,59 @@ package kubevip
 
 import (
 	"fmt"
+	"log"
 	"strconv"
 
+	"github.com/google/go-containerregistry/pkg/name"
+	"github.com/kube-vip/kube-vip/pkg/debouncer"
 	appv1 "k8s.io/api/apps/v1"
 	corev1 "k8s.io/api/core/v1"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/apis/meta/v1/unstructured"
+	"k8s.io/apimachinery/pkg/runtime"
 	applyCoreV1 "k8s.io/client-go/applyconfigurations/core/v1"
 	applyMetaV1 "k8s.io/client-go/applyconfigurations/meta/v1"
 	applyRbacV1 "k8s.io/client-go/applyconfigurations/rbac/v1"
-
 	"sigs.k8s.io/yaml"
 )
 
+// TransformApplyObjectToManifest transforms an apply object into a normal Kubernetes manifest
+func TransformApplyObjectToManifest(applyObject interface{}) string {
+	// Convert the apply object to an unstructured object
+	unstructuredObj := &unstructured.Unstructured{}
+	err := runtime.DefaultUnstructuredConverter.FromUnstructured(applyConfigToMap(applyObject), unstructuredObj)
+	if err != nil {
+		log.Fatalf("Error converting apply object to unstructured: %v", err)
+	}
+
+	// Marshal the unstructured object into YAML
+	yamlData, err := yaml.Marshal(unstructuredObj.Object)
+	if err != nil {
+		log.Fatalf("Error marshaling unstructured object to YAML: %v", err)
+	}
+
+	return string(yamlData)
+}
+
+// Helper function to convert apply configuration to a map
+func applyConfigToMap(applyConfig interface{}) map[string]interface{} {
+	data, err := runtime.DefaultUnstructuredConverter.ToUnstructured(applyConfig)
+	if err != nil {
+		log.Fatalf("Error converting apply configuration to map: %v", err)
+	}
+	return data
+}
+
 // GenerateSA will create the service account for kube-vip
-func GenerateSA() *applyCoreV1.ServiceAccountApplyConfiguration {
+func GenerateSA(c *Config) *applyCoreV1.ServiceAccountApplyConfiguration {
 	kind := "ServiceAccount"
 	name := "kube-vip"
-	namespace := "kube-system"
+	var namespace string
+	if c.ServiceNamespace != "" {
+		namespace = c.ServiceNamespace
+	} else {
+		namespace = metav1.NamespaceSystem
+	}
 	newManifest := &applyCoreV1.ServiceAccountApplyConfiguration{
 		TypeMetaApplyConfiguration: applyMetaV1.TypeMetaApplyConfiguration{APIVersion: &corev1.SchemeGroupVersion.Version, Kind: &kind},
 		ObjectMetaApplyConfiguration: &applyMetaV1.ObjectMetaApplyConfiguration{
@@ -30,15 +66,31 @@ func GenerateSA() *applyCoreV1.ServiceAccountApplyConfiguration {
 }
 
 // GenerateCR will generate the Cluster role for kube-vip
-func GenerateCR() *applyRbacV1.ClusterRoleApplyConfiguration {
-	name := "system:kube-vip-role"
-	roleRefKind := "ClusterRole"
-	apiVersion := "rbac.authorization.k8s.io/v1"
+func GenerateRole(c *Config, role bool) *applyRbacV1.RoleApplyConfiguration {
+	var kind, name string
+	var namespace *string
+	if role {
+		kind = "Role"
+		name = "kube-vip"
+		if c.ServiceNamespace != "" {
+			namespace = &c.ServiceNamespace
+		} else {
+			// If the namespace is empty then we need to set it to the system namespace
+			copiedNamespace := metav1.NamespaceSystem
+			namespace = &copiedNamespace
+		}
 
-	newManifest := &applyRbacV1.ClusterRoleApplyConfiguration{
-		TypeMetaApplyConfiguration: applyMetaV1.TypeMetaApplyConfiguration{APIVersion: &apiVersion, Kind: &roleRefKind},
+	} else {
+		kind = "ClusterRole"
+		name = "system:kube-vip-role"
+
+	}
+	apiVersion := "rbac.authorization.k8s.io/v1"
+	newManifest := &applyRbacV1.RoleApplyConfiguration{
+		TypeMetaApplyConfiguration: applyMetaV1.TypeMetaApplyConfiguration{APIVersion: &apiVersion, Kind: &kind},
 		ObjectMetaApplyConfiguration: &applyMetaV1.ObjectMetaApplyConfiguration{
-			Name: &name,
+			Name:      &name,
+			Namespace: namespace,
 		},
 		Rules: []applyRbacV1.PolicyRuleApplyConfiguration{
 			{
@@ -49,7 +101,7 @@ func GenerateCR() *applyRbacV1.ClusterRoleApplyConfiguration {
 			{
 				APIGroups: []string{""},
 				Resources: []string{"services", "endpoints"},
-				Verbs:     []string{"list", "get", "watch", "endoints"},
+				Verbs:     []string{"list", "get", "watch", "update"},
 			},
 			{
 				APIGroups: []string{""},
@@ -61,38 +113,61 @@ func GenerateCR() *applyRbacV1.ClusterRoleApplyConfiguration {
 				Resources: []string{"leases"},
 				Verbs:     []string{"list", "get", "watch", "update", "create"},
 			},
+			{
+				APIGroups: []string{"discovery.k8s.io"},
+				Resources: []string{"endpointslices"},
+				Verbs:     []string{"list", "get", "watch", "update"},
+			},
+			{
+				APIGroups: []string{""},
+				Resources: []string{"pods"},
+				Verbs:     []string{"list"},
+			},
 		},
 	}
+	if !role {
+		newManifest.Rules = append(newManifest.Rules, applyRbacV1.PolicyRuleApplyConfiguration{
+			APIGroups: []string{"networking.k8s.io"},
+			Resources: []string{"servicecidrs"},
+			Verbs:     []string{"list", "get", "watch"},
+		})
+	}
 	return newManifest
 }
 
-// GenerateCRB will generate the clusterRoleBinding
-func GenerateCRB() *applyRbacV1.ClusterRoleBindingApplyConfiguration {
-	kind := "ClusterRoleBinding"
+// GenerateCRB will generate the clusterRoleBinding or rolebinding
+func GenerateRoleBinding(rolebinding bool, saCfg *applyCoreV1.ServiceAccountApplyConfiguration, crCfg *applyRbacV1.RoleApplyConfiguration) *applyRbacV1.RoleBindingApplyConfiguration {
 	apiVersion := "rbac.authorization.k8s.io/v1"
-	subjectKind := "ServiceAccount"
 	apiGroup := "rbac.authorization.k8s.io"
-	roleRefKind := "ClusterRole"
-	roleRefName := "system:kube-vip-role"
-	name := "kube-vip"
-	bindName := "system:kube-vip-role-binding"
-	namespace := "kube-system"
-
-	newManifest := &applyRbacV1.ClusterRoleBindingApplyConfiguration{
+	var kind, bindName string
+	var namespace, objectNamespace *string
+	if rolebinding {
+		kind = "RoleBinding"
+		bindName = "kube-vip"
+		namespace = nil
+		objectNamespace = saCfg.Namespace
+	} else {
+		kind = "ClusterRoleBinding"
+		bindName = "system:kube-vip-binding"
+		namespace = saCfg.Namespace
+		objectNamespace = nil
+	}
+	newManifest := &applyRbacV1.RoleBindingApplyConfiguration{
 		TypeMetaApplyConfiguration: applyMetaV1.TypeMetaApplyConfiguration{APIVersion: &apiVersion, Kind: &kind},
 		ObjectMetaApplyConfiguration: &applyMetaV1.ObjectMetaApplyConfiguration{
-			Name: &bindName,
+			Name:      &bindName,
+			Namespace: objectNamespace,
 		},
 		RoleRef: &applyRbacV1.RoleRefApplyConfiguration{
 			APIGroup: &apiGroup,
-			Kind:     &roleRefKind,
-			Name:     &roleRefName,
+			Kind:     crCfg.Kind,
+			Name:     crCfg.Name,
 		},
 		Subjects: []applyRbacV1.SubjectApplyConfiguration{
 			{
-				Kind:      &subjectKind,
-				Name:      &name,
-				Namespace: &namespace,
+				Kind:      saCfg.Kind,
+				Name:      saCfg.Name,
+				Namespace: namespace,
 			},
 		},
 	}
@@ -100,7 +175,11 @@ func GenerateCRB() *applyRbacV1.ClusterRoleBindingApplyConfiguration {
 }
 
 // generatePodSpec will take a kube-vip config and generate a Pod spec
-func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod {
+func generatePodSpec(c *Config, image, imageVersion string, inCluster bool) (*corev1.Pod, error) {
+	imageRef, err := name.NewTag(image, name.WeakValidation, name.WithDefaultTag(imageVersion))
+	if err != nil {
+		return nil, fmt.Errorf("cannot parse %q: %w", image, err)
+	}
 	command := "manager"
 
 	// Determine where the pods should be living (for multi-tenancy)
@@ -130,6 +209,12 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 			},
 		},
 	}
+	if c.InstanceName != "" {
+		newEnvironment = append(newEnvironment, corev1.EnvVar{
+			Name:  instanceName,
+			Value: c.InstanceName,
+		})
+	}
 
 	// If we're specifically saying which interface to use then add it to the manifest
 	if c.Interface != "" {
@@ -161,16 +246,15 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 		newEnvironment = append(newEnvironment, svcInterface...)
 	}
 
-	// If a CIDR is used add it to the manifest
-	if c.VIPCIDR != "" {
-		// build environment variables
-		cidr := []corev1.EnvVar{
+	// Tolerate a down interface
+	if c.AllowInterfaceNotUp {
+		allowIface := []corev1.EnvVar{
 			{
-				Name:  vipCidr,
-				Value: c.VIPCIDR,
+				Name:  vipAllowInterfaceNotUp,
+				Value: strconv.FormatBool(c.AllowInterfaceNotUp),
 			},
 		}
-		newEnvironment = append(newEnvironment, cidr...)
+		newEnvironment = append(newEnvironment, allowIface...)
 	}
 
 	// If a subnet is required for the VIP
@@ -196,6 +280,28 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 		newEnvironment = append(newEnvironment, dnsModeSelector...)
 	}
 
+	if c.DHCPMode != "" {
+		// build environment variables
+		dhcpModeSelector := []corev1.EnvVar{
+			{
+				Name:  dhcpMode,
+				Value: c.DHCPMode,
+			},
+		}
+		newEnvironment = append(newEnvironment, dhcpModeSelector...)
+	}
+
+	if c.DHCPBackoffAttempts != DefaultDHCPBackoffAttempts {
+		// build environment variables
+		dhcpBackoff := []corev1.EnvVar{
+			{
+				Name:  dhcpBackoffAttempts,
+				Value: strconv.FormatUint(uint64(c.DHCPBackoffAttempts), 10),
+			},
+		}
+		newEnvironment = append(newEnvironment, dhcpBackoff...)
+	}
+
 	// If we're doing the hybrid mode
 	if c.EnableControlPlane {
 		cp := []corev1.EnvVar{
@@ -308,6 +414,23 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 		newEnvironment = append(newEnvironment, leaderElection...)
 	}
 
+	if c.LoseLeadership {
+		loseLeadership := []corev1.EnvVar{
+			{
+				Name:  vipLoseLeadership,
+				Value: strconv.FormatBool(c.LoseLeadership),
+			},
+		}
+
+		if c.LoseLeadershipTimeoutSeconds > 0 {
+			loseLeadership = append(loseLeadership, corev1.EnvVar{
+				Name:  vipLoseLeadership,
+				Value: fmt.Sprintf("%d", c.LoseLeadershipTimeoutSeconds),
+			})
+		}
+		newEnvironment = append(newEnvironment, loseLeadership...)
+	}
+
 	// If we're enabling node labeling on leader election
 	if c.EnableNodeLabeling {
 		EnableNodeLabeling := []corev1.EnvVar{
@@ -331,40 +454,6 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 
 	}
 
-	// If we're specifying a configuration
-	if c.ProviderConfig != "" {
-		provider := []corev1.EnvVar{
-			{
-				Name:  providerConfig,
-				Value: c.ProviderConfig,
-			},
-		}
-		newEnvironment = append(newEnvironment, provider...)
-	}
-
-	// If Equinix Metal is enabled then add it to the manifest
-	if c.EnableMetal {
-		packet := []corev1.EnvVar{
-			{
-				Name:  vipPacket,
-				Value: strconv.FormatBool(c.EnableMetal),
-			},
-			{
-				Name:  vipPacketProject,
-				Value: c.MetalProject,
-			},
-			{
-				Name:  vipPacketProjectID,
-				Value: c.MetalProjectID,
-			},
-			{
-				Name:  "PACKET_AUTH_TOKEN",
-				Value: c.MetalAPIKey,
-			},
-		}
-		newEnvironment = append(newEnvironment, packet...)
-	}
-
 	// Detect and enable wireguard mode
 	if c.EnableWireguard {
 		wireguard := []corev1.EnvVar{
@@ -386,8 +475,7 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 		}
 		newEnvironment = append(newEnvironment, routingtable...)
 	}
-
-	// If BGP, but we're not using Equinix Metal
+	// If BGP
 	if c.EnableBGP {
 		bgp := []corev1.EnvVar{
 			{
@@ -395,10 +483,17 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 				Value: strconv.FormatBool(c.EnableBGP),
 			},
 		}
+		if c.BGPAttachIPToInterface {
+			bgp = append(bgp, corev1.EnvVar{
+				Name:  bgpAttachIPToInterface,
+				Value: strconv.FormatBool(c.BGPAttachIPToInterface),
+			})
+		}
 		newEnvironment = append(newEnvironment, bgp...)
 	}
-	// If BGP, but we're not using Equinix Metal
-	if c.EnableBGP && !c.EnableMetal {
+
+	// If BGP
+	if c.EnableBGP {
 		bgpConfig := []corev1.EnvVar{
 			{
 				Name:  bgpRouterID,
@@ -460,6 +555,40 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 
 	}
 
+	if c.ControlPlaneHealthCheck.Address != "" {
+		healthCheckVars := []corev1.EnvVar{
+			{
+				Name:  controlPlaneHealthCheckAddress,
+				Value: c.ControlPlaneHealthCheck.Address,
+			},
+		}
+		if c.ControlPlaneHealthCheck.PeriodSeconds > 0 {
+			healthCheckVars = append(healthCheckVars, corev1.EnvVar{
+				Name:  controlPlaneHealthCheckPeriodSeconds,
+				Value: fmt.Sprintf("%d", c.ControlPlaneHealthCheck.PeriodSeconds),
+			})
+		}
+		if c.ControlPlaneHealthCheck.TimeoutSeconds > 0 {
+			healthCheckVars = append(healthCheckVars, corev1.EnvVar{
+				Name:  controlPlaneHealthCheckTimeoutSeconds,
+				Value: fmt.Sprintf("%d", c.ControlPlaneHealthCheck.TimeoutSeconds),
+			})
+		}
+		if c.ControlPlaneHealthCheck.FailureThreshold > 0 {
+			healthCheckVars = append(healthCheckVars, corev1.EnvVar{
+				Name:  controlPlaneHealthCheckFailureThreshold,
+				Value: fmt.Sprintf("%d", c.ControlPlaneHealthCheck.FailureThreshold),
+			})
+		}
+		if c.ControlPlaneHealthCheck.CAPath != "" {
+			healthCheckVars = append(healthCheckVars, corev1.EnvVar{
+				Name:  controlPlaneHealthCheckCAPath,
+				Value: c.ControlPlaneHealthCheck.CAPath,
+			})
+		}
+		newEnvironment = append(newEnvironment, healthCheckVars...)
+	}
+
 	// If the load-balancer is enabled then add the configuration to the manifest
 	if c.EnableLoadBalancer {
 		lb := []corev1.EnvVar{
@@ -500,10 +629,10 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 	}
 	newEnvironment = append(newEnvironment, prometheus...)
 
-	if c.EnableEndpointSlices {
+	if c.EnableEndpoints {
 		newEnvironment = append(newEnvironment, corev1.EnvVar{
-			Name:  enableEndpointSlices,
-			Value: strconv.FormatBool(c.EnableEndpointSlices),
+			Name:  enableEndpoints,
+			Value: strconv.FormatBool(c.EnableEndpoints),
 		})
 	}
 
@@ -528,9 +657,19 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 		newEnvironment = append(newEnvironment, mdif...)
 	}
 
+	if c.HealthCheckPort != 0 {
+		healthPort := []corev1.EnvVar{
+			{
+				Name:  healthCheckPort,
+				Value: fmt.Sprintf("%d", c.HealthCheckPort),
+			},
+		}
+		newEnvironment = append(newEnvironment, healthPort...)
+	}
+
 	var securityContext *corev1.SecurityContext
 	if c.LoadBalancerForwardingMethod == "masquerade" {
-		var privileged = true
+		privileged := true
 		securityContext = &corev1.SecurityContext{
 			Privileged: &privileged,
 		}
@@ -541,10 +680,32 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 					"NET_ADMIN",
 					"NET_RAW",
 				},
+				Drop: []corev1.Capability{
+					"ALL",
+				},
 			},
 		}
 	}
 
+	if c.PreserveVIPOnLeadershipLoss {
+		preserveVIPOnLeadershipLoss := []corev1.EnvVar{
+			{
+				Name:  vipPreserveOnLeadershipLoss,
+				Value: strconv.FormatBool(c.PreserveVIPOnLeadershipLoss),
+			},
+		}
+		newEnvironment = append(newEnvironment, preserveVIPOnLeadershipLoss...)
+	}
+
+	if c.DebounceTime != debouncer.DefaultTime {
+		debTime := corev1.EnvVar{
+			Name:  debounceTime,
+			Value: c.DebounceTime,
+		}
+
+		newEnvironment = append(newEnvironment, debTime)
+	}
+
 	newManifest := &corev1.Pod{
 		TypeMeta: metav1.TypeMeta{
 			Kind:       "Pod",
@@ -558,7 +719,7 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 			Containers: []corev1.Container{
 				{
 					Name:            "kube-vip",
-					Image:           fmt.Sprintf("ghcr.io/kube-vip/kube-vip:%s", imageVersion),
+					Image:           imageRef.Name(),
 					ImagePullPolicy: corev1.PullIfNotPresent,
 					SecurityContext: securityContext,
 					Args: []string{
@@ -599,38 +760,24 @@ func generatePodSpec(c *Config, imageVersion string, inCluster bool) *corev1.Pod
 		newManifest.Spec.HostAliases = append(newManifest.Spec.HostAliases, hostAlias)
 	}
 
-	if c.ProviderConfig != "" {
-		providerConfigMount := corev1.VolumeMount{
-			Name:      "cloud-sa-volume",
-			MountPath: "/etc/cloud-sa",
-			ReadOnly:  true,
-		}
-		newManifest.Spec.Containers[0].VolumeMounts = append(newManifest.Spec.Containers[0].VolumeMounts, providerConfigMount)
-
-		providerConfigVolume := corev1.Volume{
-			Name: "cloud-sa-volume",
-			VolumeSource: corev1.VolumeSource{
-				Secret: &corev1.SecretVolumeSource{
-					SecretName: "metal-cloud-config",
-				},
-			},
-		}
-		newManifest.Spec.Volumes = append(newManifest.Spec.Volumes, providerConfigVolume)
-
-	}
-
-	return newManifest
+	return newManifest, nil
 }
 
 // GeneratePodManifestFromConfig will take a kube-vip config and generate a manifest
-func GeneratePodManifestFromConfig(c *Config, imageVersion string, inCluster bool) string {
-	newManifest := generatePodSpec(c, imageVersion, inCluster)
-	b, _ := yaml.Marshal(newManifest)
-	return string(b)
+func GeneratePodManifestFromConfig(c *Config, image, imageVersion string, inCluster bool) (string, error) {
+	newManifest, err := generatePodSpec(c, image, imageVersion, inCluster)
+	if err != nil {
+		return "", err
+	}
+	b, err := yaml.Marshal(newManifest)
+	if err != nil {
+		return "", fmt.Errorf("failed to marshal manifest: %w", err)
+	}
+	return string(b), nil
 }
 
 // GenerateDaemonsetManifestFromConfig will take a kube-vip config and generate a manifest
-func GenerateDaemonsetManifestFromConfig(c *Config, imageVersion string, inCluster, taint bool) string {
+func GenerateDaemonsetManifestFromConfig(c *Config, image, imageVersion string, inCluster, taint bool) (string, error) {
 	// Determine where the pod should be deployed
 	var namespace string
 	if c.ServiceNamespace != "" {
@@ -639,7 +786,11 @@ func GenerateDaemonsetManifestFromConfig(c *Config, imageVersion string, inClust
 		namespace = metav1.NamespaceSystem
 	}
 
-	podSpec := generatePodSpec(c, imageVersion, inCluster).Spec
+	pod, err := generatePodSpec(c, image, imageVersion, inCluster)
+	if err != nil {
+		return "", err
+	}
+
 	newManifest := &appv1.DaemonSet{
 		TypeMeta: metav1.TypeMeta{
 			Kind:       "DaemonSet",
@@ -666,7 +817,7 @@ func GenerateDaemonsetManifestFromConfig(c *Config, imageVersion string, inClust
 						"app.kubernetes.io/version": imageVersion,
 					},
 				},
-				Spec: podSpec,
+				Spec: pod.Spec,
 			},
 		},
 	}
@@ -716,5 +867,5 @@ func GenerateDaemonsetManifestFromConfig(c *Config, imageVersion string, inClust
 	delete(m, "status")
 
 	b, _ = yaml.Marshal(m)
-	return string(b)
+	return string(b), nil
 }

pkg/kubevip/config_generator_test.go

@@ -1,6 +1,38 @@
 package kubevip
 
-import "testing"
+import (
+	"os"
+	"slices"
+	"strings"
+	"testing"
+
+	applyRbacV1 "k8s.io/client-go/applyconfigurations/rbac/v1"
+)
+
+func TestGenerateRoleServiceCIDRAccess(t *testing.T) {
+	clusterRole := GenerateRole(&Config{}, false)
+	if !hasServiceCIDRRule(clusterRole) {
+		t.Fatal("generated ClusterRole is missing ServiceCIDR access")
+	}
+
+	role := GenerateRole(&Config{ServiceNamespace: "kube-vip"}, true)
+	if hasServiceCIDRRule(role) {
+		t.Fatal("generated namespaced Role contains ineffective ServiceCIDR access")
+	}
+}
+
+func hasServiceCIDRRule(role *applyRbacV1.RoleApplyConfiguration) bool {
+	for _, rule := range role.Rules {
+		if slices.Contains(rule.APIGroups, "networking.k8s.io") &&
+			slices.Contains(rule.Resources, "servicecidrs") &&
+			slices.Contains(rule.Verbs, "get") &&
+			slices.Contains(rule.Verbs, "list") &&
+			slices.Contains(rule.Verbs, "watch") {
+			return true
+		}
+	}
+	return false
+}
 
 func TestParseEnvironment(t *testing.T) {
 
@@ -9,8 +41,8 @@ func TestParseEnvironment(t *testing.T) {
 		c       *Config
 		wantErr bool
 	}{
-		{"", nil, false},
-		{"", &Config{Interface: "eth0", ServicesInterface: "eth1"}, false},
+		{"nil config", nil, false},
+		{"basic config", &Config{Interface: "eth0", ServicesInterface: "eth1"}, false},
 	}
 	for _, tt := range tests {
 		t.Logf("%v", tt.c)
@@ -21,3 +53,146 @@ func TestParseEnvironment(t *testing.T) {
 		})
 	}
 }
+
+func TestParseEnvironmentInstanceName(t *testing.T) {
+	tests := []struct {
+		name      string
+		lowercase string
+		uppercase string
+		want      string
+	}{
+		{name: "lowercase", lowercase: "release_a", want: "release_a"},
+		{name: "uppercase fallback", uppercase: "release_b", want: "release_b"},
+		{name: "lowercase takes precedence", lowercase: "release_a", uppercase: "release_b", want: "release_a"},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Setenv(instanceName, tt.lowercase)
+			t.Setenv(strings.ToUpper(instanceName), tt.uppercase)
+
+			config := &Config{}
+			if err := ParseEnvironment(config); err != nil {
+				t.Fatalf("ParseEnvironment() error = %v", err)
+			}
+			if config.InstanceName != tt.want {
+				t.Fatalf("InstanceName = %q, want %q", config.InstanceName, tt.want)
+			}
+		})
+	}
+}
+
+func TestParseEnvironmentBGPAttachIPToInterface(t *testing.T) {
+	t.Setenv(bgpAttachIPToInterface, "true")
+
+	config := &Config{}
+	if err := ParseEnvironment(config); err != nil {
+		t.Fatalf("ParseEnvironment() error = %v", err)
+	}
+	if !config.BGPAttachIPToInterface {
+		t.Fatal("BGPAttachIPToInterface = false, want true")
+	}
+}
+
+func TestGeneratePodSpecBGPAttachIPToInterface(t *testing.T) {
+	pod, err := generatePodSpec(&Config{
+		EnableBGP:              true,
+		BGPAttachIPToInterface: true,
+	}, "ghcr.io/kube-vip/kube-vip", "v0.0.0", true)
+	if err != nil {
+		t.Fatalf("generatePodSpec() error = %v", err)
+	}
+
+	for _, env := range pod.Spec.Containers[0].Env {
+		if env.Name == bgpAttachIPToInterface && env.Value == "true" {
+			return
+		}
+	}
+	t.Fatalf("%s=true is missing from generated pod environment", bgpAttachIPToInterface)
+}
+
+func TestGeneratePodSpecInstanceName(t *testing.T) {
+	tests := []struct {
+		name         string
+		instanceName string
+		wantPresent  bool
+	}{
+		{name: "configured", instanceName: "release_a", wantPresent: true},
+		{name: "empty", wantPresent: false},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			pod, err := generatePodSpec(&Config{InstanceName: tt.instanceName}, "ghcr.io/kube-vip/kube-vip", "v0.0.0", true)
+			if err != nil {
+				t.Fatalf("generatePodSpec() error = %v", err)
+			}
+
+			var value string
+			found := false
+			for _, env := range pod.Spec.Containers[0].Env {
+				if env.Name == instanceName {
+					found = true
+					value = env.Value
+					break
+				}
+			}
+			if found != tt.wantPresent {
+				t.Fatalf("instance_name present = %t, want %t", found, tt.wantPresent)
+			}
+			if found && value != tt.instanceName {
+				t.Fatalf("instance_name = %q, want %q", value, tt.instanceName)
+			}
+		})
+	}
+}
+
+func TestParseEnvironmentConfigFile(t *testing.T) {
+	// Save original environment
+	originalConfigFile := os.Getenv("config_file")
+	defer func() {
+		if originalConfigFile != "" {
+			os.Setenv("config_file", originalConfigFile)
+		} else {
+			os.Unsetenv("config_file")
+		}
+	}()
+
+	tests := []struct {
+		name           string
+		envValue       string
+		expectedConfig string
+	}{
+		{
+			name:           "config_file environment variable set",
+			envValue:       "/etc/kube-vip/config.yaml",
+			expectedConfig: "/etc/kube-vip/config.yaml",
+		},
+		{
+			name:           "config_file environment variable empty",
+			envValue:       "",
+			expectedConfig: "",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			// Set environment variable
+			if tt.envValue != "" {
+				os.Setenv("config_file", tt.envValue)
+			} else {
+				os.Unsetenv("config_file")
+			}
+
+			config := &Config{}
+			err := ParseEnvironment(config)
+			if err != nil {
+				t.Errorf("ParseEnvironment() unexpected error = %v", err)
+			}
+
+			if config.ConfigFile != tt.expectedConfig {
+				t.Errorf("ConfigFile = %v, expected %v", config.ConfigFile, tt.expectedConfig)
+			}
+		})
+	}
+}

pkg/kubevip/config_manager.go

@@ -1,9 +1,11 @@
 package kubevip
 
 import (
+	"errors"
 	"fmt"
 
-	log "github.com/sirupsen/logrus"
+	log "log/slog"
+
 	"github.com/vishvananda/netlink"
 )
 
@@ -11,16 +13,34 @@ const (
 	Auto = "auto"
 )
 
+var ErrInterfaceNotUp = errors.New("interface is not up")
+
+func (c *Config) CheckSubnetExists() error {
+	if c.VIPSubnet == "" && c.VIP != "" && c.Address == "" {
+		return fmt.Errorf("vip_subnet must be set if using vip_address instead of address environment variable")
+	}
+
+	return nil
+}
+
 func (c *Config) CheckInterface() error {
 	if c.Interface != "" {
 		if err := isValidInterface(c.Interface); err != nil {
-			return fmt.Errorf("%s is not valid interface, reason: %w", c.Interface, err)
+			if errors.Is(err, ErrInterfaceNotUp) && c.AllowInterfaceNotUp {
+				log.Warn("interface is not up, continuing as allowInterfaceNotUp is set", "interface", c.Interface)
+			} else {
+				return fmt.Errorf("%s is not valid interface, reason: %w", c.Interface, err)
+			}
 		}
 	}
 
 	if c.ServicesInterface != "" {
 		if err := isValidInterface(c.ServicesInterface); err != nil {
-			return fmt.Errorf("%s is not valid interface, reason: %w", c.ServicesInterface, err)
+			if errors.Is(err, ErrInterfaceNotUp) && c.AllowInterfaceNotUp {
+				log.Warn("interface is not up, continuing as allowInterfaceNotUp is set", "interface", c.ServicesInterface)
+			} else {
+				return fmt.Errorf("%s is not valid interface, reason: %w", c.ServicesInterface, err)
+			}
 		}
 	}
 
@@ -44,12 +64,12 @@ func isValidInterface(iface string) error {
 	// userspace has set operational state. Interface must be considered for user
 	// data as setting operational state has not been implemented in every driver."
 	if attrs.OperState == netlink.OperUnknown {
-		log.Warningf(
-			"the status of the interface %s is unknown. Ensure your interface is ready to accept traffic, if so you can safely ignore this message",
+		log.Warn(
+			"the status of the interface is unknown. Ensure your interface is ready to accept traffic, if so you can safely ignore this message", "interface",
 			iface,
 		)
 	} else if attrs.OperState != netlink.OperUp {
-		return fmt.Errorf("%s is not up", iface)
+		return fmt.Errorf("%s %w", iface, ErrInterfaceNotUp)
 	}
 
 	return nil

pkg/kubevip/config_manager_test.go

@@ -0,0 +1,56 @@
+package kubevip
+
+import "testing"
+
+func TestCheckSubnetExists(t *testing.T) {
+	tests := []struct {
+		name    string
+		config  Config
+		wantErr bool
+	}{
+		{
+			name: "vip only without subnet returns error",
+			config: Config{
+				VIP: "172.18.0.20",
+			},
+			wantErr: true,
+		},
+		{
+			name: "vip with subnet does not return error",
+			config: Config{
+				VIP:       "172.18.0.20",
+				VIPSubnet: "32",
+			},
+			wantErr: false,
+		},
+		{
+			name: "address only without subnet does not return error",
+			config: Config{
+				Address: "172.18.0.20",
+			},
+			wantErr: false,
+		},
+		{
+			name: "address overrides vip without subnet",
+			config: Config{
+				VIP:     "172.18.0.20",
+				Address: "172.18.0.30",
+			},
+			wantErr: false,
+		},
+		{
+			name:    "empty config does not return error",
+			config:  Config{},
+			wantErr: false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			err := tt.config.CheckSubnetExists()
+			if (err != nil) != tt.wantErr {
+				t.Errorf("CheckSubnetExists() error = %v, wantErr %v", err, tt.wantErr)
+			}
+		})
+	}
+}

pkg/kubevip/config_types.go

@@ -1,13 +1,9 @@
 package kubevip
 
-import (
-	"github.com/kube-vip/kube-vip/pkg/bgp"
-)
-
 // Config defines all of the settings for the Kube-Vip Pod
 type Config struct {
 	// Logging, settings
-	Logging int `yaml:"logging"`
+	Logging int32 `yaml:"logging"`
 
 	// EnableARP, will use ARP to advertise the VIP address
 	EnableARP bool `yaml:"enableARP"`
@@ -15,6 +11,9 @@ type Config struct {
 	// EnableBGP, will use BGP to advertise the VIP address
 	EnableBGP bool `yaml:"enableBGP"`
 
+	// BGPAttachIPToInterface assigns BGP-advertised service VIPs to the configured interface
+	BGPAttachIPToInterface bool `yaml:"bgpAttachIPToInterface"`
+
 	// EnableWireguard, will use wireguard to advertise the VIP address
 	EnableWireguard bool `yaml:"enableWireguard"`
 
@@ -54,6 +53,16 @@ type Config struct {
 	// ArpBroadcastRate, defines how often kube-vip will update the network about updates to the network
 	ArpBroadcastRate int64 `yaml:"arpBroadcastRate"`
 
+	// PreserveVIPOnLeadershipLoss, if true, VIP addresses will remain on interface when leadership is lost (only ARP/NDP broadcasting stops)
+	// If false, VIP addresses are deleted on leadership loss (legacy behavior)
+	PreserveVIPOnLeadershipLoss bool `yaml:"preserveVipOnLeadershipLoss"`
+
+	// LoseLeadership enables leadership loss if VIP interface(physical) is down
+	LoseLeadership bool `yaml:"loseLeadership"`
+
+	// LoseLeadershipTimeoutSeconds defines the timeout after which interface will be considered down. Default is 30s
+	LoseLeadershipTimeoutSeconds int `yaml:"loseLeadershipTimeoutSeconds"`
+
 	// Annotations will define if we're going to wait and lookup configuration from Kubernetes node annotations
 	Annotations string
 
@@ -76,18 +85,18 @@ type Config struct {
 	// VipSubnet is the Subnet that is applied to the VIP
 	VIPSubnet string `yaml:"vipSubnet"`
 
-	// VIPCIDR is cidr range for the VIP (primarily needed for BGP)
-	VIPCIDR string `yaml:"vipCidr"`
-
 	// Address is the IP or DNS Name to use as a VirtualIP
 	Address string `yaml:"address"`
 
 	// Listen port for the VirtualIP
-	Port int `yaml:"port"`
+	Port uint16 `yaml:"port"`
 
 	// Namespace will define which namespace the control plane pods will run in
 	Namespace string `yaml:"namespace"`
 
+	// InstanceName identifies this kube-vip deployment when naming host-global resources.
+	InstanceName string `yaml:"instanceName"`
+
 	// Namespace will define which namespace the control plane pods will run in
 	ServiceNamespace string `yaml:"serviceNamespace"`
 
@@ -109,11 +118,14 @@ type Config struct {
 	// ServicesInterface is the network interface to bind to for services (optional)
 	ServicesInterface string `yaml:"servicesInterface,omitempty"`
 
+	// AllowInterfaceNotUp allows kube-vip to start even when the interface is not up
+	AllowInterfaceNotUp bool `yaml:"allowInterfaceNotUp,omitempty"`
+
 	// EnableLoadBalancer, provides the flexibility to make the load-balancer optional
 	EnableLoadBalancer bool `yaml:"enableLoadBalancer"`
 
 	// Listen port for the IPVS Service
-	LoadBalancerPort int `yaml:"lbPort"`
+	LoadBalancerPort uint16 `yaml:"lbPort"`
 
 	// Forwarding method for the IPVS Service
 	LoadBalancerForwardingMethod string `yaml:"lbForwardingMethod"`
@@ -130,25 +142,17 @@ type Config struct {
 	// Clean routing table of redundant routes on start
 	CleanRoutingTable bool `yaml:"cleanRoutingTable"`
 
+	// Skip Duplicate Address Detection when adding the VIP address (IFA_F_NODAD)
+	SkipDAD bool `yaml:"skipDAD"`
+
 	// BGP Configuration
-	BGPConfig     bgp.Config
-	BGPPeerConfig bgp.Peer
+	BGPConfig     BGPConfig
+	BGPPeerConfig BGPPeer
 	BGPPeers      []string
 
-	// EnableMetal, will use the metal API to update the EIP <-> VIP (if BGP is enabled then BGP will be used)
-	EnableMetal bool `yaml:"enableMetal"`
-
-	// MetalAPIKey, is the API token used to authenticate to the API
-	MetalAPIKey string
-
-	// MetalProject, is the name of a particular defined project
-	MetalProject string
-
-	// MetalProjectID, is the name of a particular defined project
-	MetalProjectID string
-
-	// ProviderConfig, is the path to a provider configuration file
-	ProviderConfig string
+	// ControlPlaneHealthCheck configures HTTP polling of the control plane when using BGP without
+	// leader election. If the health check fails, the BGP route will be withdrawn.
+	ControlPlaneHealthCheck HealthCheck `yaml:"controlPlaneHealthCheck,omitempty"`
 
 	// LoadBalancers are the various services we can load balance over
 	LoadBalancers []LoadBalancer `yaml:"loadBalancers,omitempty"`
@@ -164,6 +168,9 @@ type Config struct {
 	// EgressServiceCidr, this contains the service cidr range to ignore
 	EgressServiceCidr string
 
+	// EnableInternalSNAT, this will enable the internal SNAT rule that kube-vip adds to the egress chain
+	EnableInternalSNAT bool
+
 	// EgressWithNftables, this will use the iptables-nftables OVER iptables
 	EgressWithNftables bool
 
@@ -176,11 +183,21 @@ type Config struct {
 	// DNSMode, this will set the mode DSN lookup will be performed (first, ipv4, ipv6, dual)
 	DNSMode string `yaml:"dnsDualStackMode"`
 
+	// IsDualStack reports if service is DualStack.
+	IsDualStack bool
+
+	// RequireDualStack defines if DualStack is required for the service. Based on service's Spec.ipFamilyPolicy field.
+	RequireDualStack bool
+
+	// DNSMode, this will set the mode DHCP lookup will be performed for DDNS (ipv4, ipv6, dual). By default will be the same as DNSMode.
+	// If DNSMode is 'first', IPv4 will be used.
+	DHCPMode string `yaml:"dhcpDualStackMode"`
+
 	// DisableServiceUpdates, if true, kube-vip will only advertise service, but it will not update service's Status.LoadBalancer.Ingress slice
 	DisableServiceUpdates bool `yaml:"disableServiceUpdates"`
 
-	// EnableEndpointSlices, if enabled, EndpointSlices will be used instead of Endpoints
-	EnableEndpointSlices bool `yaml:"enableEndpointSlices"`
+	// EnableEndpoints, if enabled, Endpoints will be used instead of EndpointSlices
+	EnableEndpoints bool `yaml:"enableEndpoints"`
 
 	// MirrorDestInterface is the network interface where all traffics that go through service interface
 	// will be mirrored to. If ServicesInterface is not set, fall back to Interface.
@@ -195,6 +212,27 @@ type Config struct {
 
 	// LoInterfaceGlobalScope, if true will set global scope when using the lo interface, otherwise a host scope will be used
 	LoInterfaceGlobalScope bool `yaml:"loInterfaceGlobalScope"`
+
+	// HealthCheckPort, if non-zero then will enable the healthcheck to return ok on this port
+	HealthCheckPort int `yaml:"healthCheckPort"`
+
+	// EnableUPNP, enables UPNP functions
+	EnableUPNP bool `yaml:"enableUPNP"`
+
+	// EgressClean, enables egress cleaning on Kube-vip's start
+	EgressClean bool `yaml:"egressClean"`
+
+	// ConfigFile defines the path to a JSON/YAML configuration file
+	ConfigFile string `yaml:"configFile"`
+
+	// DHCPBackoffAttempts defines how many times will DHCP client try to obtain address (unlimited when 0)
+	DHCPBackoffAttempts uint `yaml:"dhcpBackoffAttempts"`
+
+	// DebounceTime defines how long will event debouncer wait for the events to arrive
+	DebounceTime string `yaml:"debounceTime"`
+
+	// PerServiceElectionOnDemand will enable kube-vip to handle services with per-service election when annotation is used
+	PerServiceElectionOnDemand bool `yaml:"perServiceElectionOnDemand"`
 }
 
 // KubernetesLeaderElection defines all of the settings for Kubernetes KubernetesLeaderElection
@@ -226,12 +264,29 @@ type Etcd struct {
 	Endpoints      []string
 }
 
+// HealthCheck defines HTTP health-check settings for control-plane polling when using BGP
+// without leader election.
+type HealthCheck struct {
+	// Address is the URL to poll to check the health of the control-plane. If the health
+	// check fails, the BGP route will be withdrawn.
+	Address string `yaml:"address"`
+	// PeriodSeconds is the interval in seconds between health checks.
+	PeriodSeconds int `yaml:"periodSeconds"`
+	// TimeoutSeconds is the timeout per health check request. If a request takes longer
+	// than this timeout, the health check is considered failed.
+	TimeoutSeconds int `yaml:"timeoutSeconds"`
+	// FailureThreshold is the number of consecutive failures before route withdrawal.
+	FailureThreshold int `yaml:"failureThreshold"`
+	// CAPath is the CA certificate path used for TLS verification when Address is an HTTPS URL.
+	CAPath string `yaml:"caPath"`
+}
+
 // LoadBalancer contains the configuration of a load balancing instance
 type LoadBalancer struct {
 	// Name of a LoadBalancer
 	Name string `yaml:"name"`
 
-	//Ports exposed by a LoadBalancer
+	// Ports exposed by a LoadBalancer
 	Ports []Port
 
 	// BindToVip will bind the load balancer port to the VIP itself

pkg/kubevip/config_validation.go

@@ -0,0 +1,73 @@
+package kubevip
+
+import (
+	"fmt"
+	"net/url"
+	"strings"
+)
+
+const (
+	// nftables object names are limited to 255 bytes. Reserve space for the
+	// prefix and address-family suffix added to the instance name.
+	nftablesNameMaxLength     = 255
+	egressNftablesTablePrefix = "kube_vip_"
+	egressNftablesTableSuffix = "_v4"
+	instanceNameMaxLength     = nftablesNameMaxLength - len(egressNftablesTablePrefix) - len(egressNftablesTableSuffix)
+)
+
+// Validate runs configuration checks that are independent of host state.
+// This should be called after all config sources (flags, file, env vars) are merged.
+func (c *Config) Validate() error {
+	if err := validateHealthCheckAddress(c.ControlPlaneHealthCheck.Address); err != nil {
+		return err
+	}
+	if err := validateInstanceName(c.InstanceName); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+func validateInstanceName(name string) error {
+	if name == "" {
+		return nil
+	}
+	if len(name) > instanceNameMaxLength {
+		return fmt.Errorf("instance_name is %d bytes, must not exceed %d bytes so the %q prefix and %q or %q suffix fit within the nftables %d-byte name limit",
+			len(name), instanceNameMaxLength, egressNftablesTablePrefix, "_v4", "_v6", nftablesNameMaxLength)
+	}
+
+	for position, char := range name {
+		if isValidNftablesNameCharacter(char) {
+			continue
+		}
+		return fmt.Errorf("instance_name %q contains invalid character %q at byte %d; only ASCII letters, digits, '.', '-' and '_' are allowed",
+			name, char, position)
+	}
+	return nil
+}
+
+func isValidNftablesNameCharacter(char rune) bool {
+	return char >= 'a' && char <= 'z' ||
+		char >= 'A' && char <= 'Z' ||
+		char >= '0' && char <= '9' ||
+		char == '_' || char == '-' || char == '.'
+}
+
+func validateHealthCheckAddress(address string) error {
+	if address == "" {
+		return nil
+	}
+
+	parsedURL, err := url.ParseRequestURI(address)
+	if err != nil {
+		return fmt.Errorf("control_plane_health_check_address %q is not a valid URL: %w", address, err)
+	}
+
+	scheme := strings.ToLower(parsedURL.Scheme)
+	if scheme != "http" && scheme != "https" {
+		return fmt.Errorf("control_plane_health_check_address %q has unsupported scheme %q, expected http or https", address, parsedURL.Scheme)
+	}
+
+	return nil
+}

pkg/kubevip/config_validation_test.go

@@ -0,0 +1,70 @@
+package kubevip
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestValidate_HealthCheckAddress(t *testing.T) {
+	tests := []struct {
+		name    string
+		address string
+		wantErr bool
+	}{
+		{"empty address (disabled)", "", false},
+		{"valid http URL", "http://localhost:6443/livez", false},
+		{"valid https URL", "https://localhost:6443/livez", false},
+		{"https with path", "https://127.0.0.1:6443/readyz?verbose", false},
+		{"invalid URL", "not-a-url", true},
+		{"ftp scheme", "ftp://localhost/file", true},
+		{"tcp scheme", "tcp://localhost:6443", true},
+		{"missing scheme", "localhost:6443/livez", true},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			c := &Config{ControlPlaneHealthCheck: HealthCheck{Address: tt.address}}
+			err := c.Validate()
+			if (err != nil) != tt.wantErr {
+				t.Errorf("Validate() error = %v, wantErr %v", err, tt.wantErr)
+			}
+		})
+	}
+}
+
+func TestValidate_InstanceName(t *testing.T) {
+	tests := []struct {
+		name         string
+		instanceName string
+		wantErr      bool
+	}{
+		{name: "empty uses legacy default", instanceName: "", wantErr: false},
+		{name: "letters digits and separators", instanceName: "release_01.prod-a", wantErr: false},
+		{name: "exact maximum length", instanceName: strings.Repeat("a", instanceNameMaxLength), wantErr: false},
+		{name: "exceeds maximum length", instanceName: strings.Repeat("a", instanceNameMaxLength+1), wantErr: true},
+		{name: "space", instanceName: "release a", wantErr: true},
+		{name: "slash", instanceName: "namespace/release", wantErr: true},
+		{name: "dollar sign", instanceName: "release$a", wantErr: true},
+		{name: "at sign", instanceName: "release@a", wantErr: true},
+		{name: "newline", instanceName: "release\na", wantErr: true},
+		{name: "null byte", instanceName: "release\x00a", wantErr: true},
+		{name: "unicode", instanceName: "rilascio-à", wantErr: true},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			config := &Config{InstanceName: tt.instanceName}
+			err := config.Validate()
+			if (err != nil) != tt.wantErr {
+				t.Fatalf("Validate() error = %v, wantErr %t", err, tt.wantErr)
+			}
+		})
+	}
+}
+
+func TestInstanceNameLimitReservesNftablesPrefixAndFamilySuffix(t *testing.T) {
+	name := strings.Repeat("a", instanceNameMaxLength)
+	if got := len(egressNftablesTablePrefix + name + egressNftablesTableSuffix); got != nftablesNameMaxLength {
+		t.Fatalf("family-specific table name length = %d, want %d", got, nftablesNameMaxLength)
+	}
+}

pkg/kubevip/constants.go

@@ -0,0 +1,7 @@
+package kubevip
+
+const (
+	LBClassName = "kube-vip.io/kube-vip-class"
+
+	DefaultDHCPBackoffAttempts = 3
+)

pkg/kubevip/labels.go

@@ -0,0 +1,22 @@
+package kubevip
+
+import (
+	"slices"
+)
+
+const (
+	// ServiceProvided is the name of the label that will be added to the node
+	ServiceProvided = "service-provided.kube-vip.io"
+
+	// label used on nodes, which announce the LoadBalancer IP
+	HasIP = "kube-vip.io/has-ip"
+)
+
+var kubevipLabelKeys = []string{
+	ServiceProvided,
+	HasIP,
+}
+
+func GetKeysForCleanup() []string {
+	return slices.Clone(kubevipLabelKeys)
+}

pkg/lease/lease.go

@@ -0,0 +1,270 @@
+package lease
+
+import (
+	"context"
+	"fmt"
+	log "log/slog"
+	"os"
+	"strings"
+	"sync"
+	"sync/atomic"
+
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	v1 "k8s.io/api/core/v1"
+)
+
+// Manager is used to manage leases.
+type Manager struct {
+	leases map[string]*Lease
+	lock   sync.Mutex
+}
+
+// NewManager creates new lease manager.
+func NewManager() *Manager {
+	return &Manager{
+		leases: make(map[string]*Lease),
+	}
+}
+
+// Add adds lease to the manager.
+// It returns three values:
+// - lease for the object
+// - isNewObject, which reports if it is a new object that is being handled
+// - isSharedLease, which is true if object shares the lease with another object
+// If object is new but not shared, we should start leaderelection and sync it
+// If object is new and shared, we should only sync it as the leaderelection should be already handled
+// If object is not new we should do nothing
+func (m *Manager) Add(ctx context.Context, id ID) *Lease {
+	m.lock.Lock()
+	defer m.lock.Unlock()
+
+	// A lease whose context is already cancelled cannot be handed out again:
+	// anything derived from it would be cancelled straight away. Replace it.
+	if l, exists := m.leases[id.NamespacedName()]; !exists || l.Ctx.Err() != nil {
+		leaseCtx, leaseCancel := context.WithCancel(ctx)
+		m.leases[id.NamespacedName()] = newLease(leaseCtx, leaseCancel)
+	}
+
+	return m.leases[id.NamespacedName()]
+}
+
+// Delete removes the object from the lease it was added to and cancels that lease
+// once its last object is gone. With a common lease, the siblings that still use
+// it keep it alive.
+//
+// The lease the caller was given has to be passed in, because cleanup is usually
+// deferred to a goroutine that runs long after the object went away. By then the
+// lease of that name may already have been replaced, for instance because the
+// service was torn down and rebuilt, and cancelling the replacement would leave
+// the service unhandled. A stale caller is therefore ignored.
+//
+// Teardown paths have to call this synchronously rather than leaving it to the
+// deferred cleanup: until the lease is out of the map, Add hands the same
+// instance back, so a service that is rebuilt straight away gets parented to a
+// lease that the pending cleanup is about to cancel.
+func (m *Manager) Delete(id ID, objectName string, l *Lease) {
+	m.lock.Lock()
+	defer m.lock.Unlock()
+
+	current := m.currentFor(id, l)
+	if current == nil {
+		return
+	}
+
+	current.delete(objectName)
+	if current.cnt.Load() < 1 {
+		m.retire(id, current)
+	}
+}
+
+// currentFor returns the registered lease for id, or nil when the caller is
+// stale, meaning the lease it holds is no longer the registered one. Callers have
+// to hold m.lock.
+func (m *Manager) currentFor(id ID, l *Lease) *Lease {
+	current, exist := m.leases[id.NamespacedName()]
+	if !exist || (l != nil && current != l) {
+		return nil
+	}
+	return current
+}
+
+// retire cancels the lease and drops it from the manager. Callers have to hold
+// m.lock.
+func (m *Manager) retire(id ID, l *Lease) {
+	l.Cancel()
+	delete(m.leases, id.NamespacedName())
+}
+
+// Get returns lease for the service.
+func (m *Manager) Get(id ID) *Lease {
+	m.lock.Lock()
+	defer m.lock.Unlock()
+
+	if lease, exist := m.leases[id.NamespacedName()]; exist {
+		return lease
+	}
+	return nil
+}
+
+// Lease holds lease data.
+type Lease struct {
+	Ctx      context.Context
+	Cancel   context.CancelFunc
+	Started  chan any
+	services sync.Map
+	cnt      atomic.Int64
+	Elected  atomic.Bool
+	Mtx      sync.Mutex
+	locked   bool
+}
+
+func newLease(ctx context.Context, cancel context.CancelFunc) *Lease {
+	return &Lease{
+		Ctx:     ctx,
+		Cancel:  cancel,
+		Started: make(chan any),
+	}
+}
+
+// Add adds the object to the lease and increments counter
+// it will return true if object was added
+func (l *Lease) Add(name string) bool {
+	if _, exists := l.services.LoadOrStore(name, true); !exists {
+		l.cnt.Add(1)
+		return true
+	}
+	return false
+}
+
+// delete removes the service from the lease and decrements the counter
+func (l *Lease) delete(service string) {
+	if _, exists := l.services.Load(service); exists {
+		l.services.Delete(service)
+		l.cnt.Add(-1)
+	}
+}
+
+func (l *Lease) Lock() {
+	l.Mtx.Lock()
+	l.locked = true
+}
+
+func (l *Lease) Unlock() {
+	if l.locked {
+		l.locked = false
+		l.Mtx.Unlock()
+	}
+}
+
+// ServiceName gets lease name and id for the service.
+func ServiceName(service *v1.Service) (string, string) {
+	name, exists := service.Annotations[kubevip.ServiceLease]
+	if !exists || name == "" {
+		name = fmt.Sprintf("kubevip-%s", service.Name)
+	}
+
+	serviceLeaseParts := strings.Split(name, "/")
+	namespace := service.Namespace
+
+	if len(serviceLeaseParts) > 1 {
+		namespace = serviceLeaseParts[0]
+		name = serviceLeaseParts[1]
+	}
+
+	return namespace, name
+}
+
+func ServiceNamespacedName(service *v1.Service) string {
+	return fmt.Sprintf("%s/%s", service.Namespace, service.Name)
+}
+
+func ObjectName(id ID, suffix string) string {
+	return fmt.Sprintf("%s-%s", id.NamespacedName(), suffix)
+}
+
+func NamespaceName(lease string, c *kubevip.Config) (string, string) {
+	leaseName := lease
+	leasnameParts := strings.Split(lease, "/")
+	var ns string
+	var err error
+	if len(leasnameParts) > 1 {
+		ns = leasnameParts[0]
+		leaseName = leasnameParts[1]
+	} else {
+		ns, err = returnNamespace()
+		if err != nil {
+			log.Warn("unable to auto-detect namespace, dropping to config", "namespace", c.Namespace)
+			ns = c.Namespace
+		}
+	}
+	return ns, leaseName
+}
+
+func returnNamespace() (string, error) {
+	if data, err := os.ReadFile("/var/run/secrets/kubernetes.io/serviceaccount/namespace"); err == nil {
+		if ns := strings.TrimSpace(string(data)); len(ns) > 0 {
+			return ns, nil
+		}
+		return "", err
+	}
+	return "", fmt.Errorf("unable to find Namespace")
+}
+
+type ID interface {
+	Name() string
+	Namespace() string
+	NamespacedName() string
+}
+
+type CommonID struct {
+	namespace string
+	name      string
+}
+
+func NewID(leaseType, namespace, name string) ID {
+	if leaseType == "etcd" {
+		return newEtcdID(namespace, name)
+	}
+	return newKubernetesID(namespace, name)
+}
+
+func newKubernetesID(namespace, name string) ID {
+	return &KubernetesID{
+		CommonID: CommonID{
+			namespace: namespace,
+			name:      name,
+		},
+	}
+}
+func newEtcdID(namespace, name string) ID {
+	return &EtcdID{
+		CommonID: CommonID{
+			namespace: namespace,
+			name:      name,
+		},
+	}
+}
+
+func (c *CommonID) Name() string {
+	return c.name
+}
+
+func (c *CommonID) Namespace() string {
+	return c.namespace
+}
+
+type KubernetesID struct {
+	CommonID
+}
+
+func (k *KubernetesID) NamespacedName() string {
+	return fmt.Sprintf("%s/%s", k.namespace, k.name)
+}
+
+type EtcdID struct {
+	CommonID
+}
+
+func (e *EtcdID) NamespacedName() string {
+	return fmt.Sprintf("%s-%s", e.namespace, e.name)
+}

pkg/loadbalancer/ipvs.go

@@ -1,17 +1,24 @@
 package loadbalancer
 
 import (
+	"context"
+	"errors"
 	"fmt"
 	"net"
 	"net/netip"
+	"os"
 	"strings"
 	"sync"
 
+	log "log/slog"
+
 	"github.com/cloudflare/ipvs"
 	"github.com/cloudflare/ipvs/netmask"
 	"github.com/kube-vip/kube-vip/pkg/backend"
 	"github.com/kube-vip/kube-vip/pkg/sysctl"
-	log "github.com/sirupsen/logrus"
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	"github.com/kube-vip/kube-vip/pkg/vip"
+	"github.com/vishvananda/netlink"
 )
 
 /*
@@ -39,47 +46,66 @@ const (
 type IPVSLoadBalancer struct {
 	client              ipvs.Client
 	loadBalancerService ipvs.Service
-	Port                int
+	Port                uint16
 	forwardingMethod    ipvs.ForwardType
 	backendMap          backend.Map
 	interval            int
 	lock                sync.Mutex
-	stop                chan struct{}
+	killFunc            func()
+	address             string
+	family              ipvs.AddressFamily
 }
 
-func NewIPVSLB(address string, port int, forwardingMethod string, backendHealthCheckInterval int) (*IPVSLoadBalancer, error) {
+func NewIPVSLB(ctx context.Context, network vip.Network, port uint16, forwardingMethod string, backendHealthCheckInterval int,
+	nftables bool, killFunc func(), wg *sync.WaitGroup) (*IPVSLoadBalancer, error) {
+	log.Info("Starting IPVS LoadBalancer", "network", network)
+
+	address := network.IP()
+
 	// Create IPVS client
 	c, err := ipvs.New()
 	if err != nil {
-		log.Errorf("ensure IPVS kernel modules are loaded")
-		log.Fatalf("Error starting IPVS [%v]", err)
+		log.Error("ensure IPVS kernel modules are loaded")
+		log.Error("error starting IPVS", "err", err)
+		return nil, fmt.Errorf("starting IPVS: %w", err)
 	}
 	i, err := c.Info()
 	if err != nil {
-		log.Errorf("ensure IPVS kernel modules are loaded")
-		log.Fatalf("Error getting IPVS version [%v]", err)
-	}
-	log.Infof("IPVS Loadbalancer enabled for %d.%d.%d", i.Version[0], i.Version[1], i.Version[2])
-
-	if strings.ToLower(forwardingMethod) == "masquerade" {
-		err = sysctl.WriteProcSys("/proc/sys/net/ipv4/vs/conntrack", "1")
-		if err != nil {
-			log.Fatalf("Error ensuring net.ipv4.vs.conntrack enabled [%v]", err)
+		log.Error("ensure IPVS kernel modules are loaded")
+		log.Error("error retrieving IPVS info", "err", err)
+		if errors.Is(err, os.ErrPermission) {
+			log.Error("no permission to get IPVS info - please ensure that kube-vip is running with proper capabilities/privileged mode")
 		}
-		log.Infof("sysctl set net.ipv4.vs.conntrack to 1")
-
-		err = sysctl.WriteProcSys("/proc/sys/net/ipv4/ip_forward", "1")
-		if err != nil {
-			log.Fatalf("Error ensuring net.ipv4.ip_forward enabled [%v]", err)
-		}
-		log.Infof("sysctl set net.ipv4.ip_forward to 1")
+		return nil, fmt.Errorf("retrieving IPVS: %w", err)
 	}
+	log.Info("IPVS Loadbalancer enabled", "version", fmt.Sprintf("%d.%d.%d", i.Version[0], i.Version[1], i.Version[2]))
 
 	ip, family := ipAndFamily(address)
 
-	netMask := netmask.MaskFrom(31, 32) // For ipv4
+	if strings.ToLower(forwardingMethod) == "masquerade" {
+		if err := enableProcSys("/proc/sys/net/ipv4/vs/conntrack", "net.ipv4.vs.conntrack"); err != nil {
+			return nil, err
+		}
+		if family == ipvs.INET6 {
+			if err := enableProcSys("/proc/sys/net/ipv6/conf/all/forwarding", "net.ipv6.conf.all.forwarding"); err != nil {
+				return nil, err
+			}
+		} else {
+			if err := enableProcSys("/proc/sys/net/ipv4/ip_forward", "net.ipv4.ip_forward"); err != nil {
+				return nil, err
+			}
+		}
+	}
+
+	netMask := netmask.MaskFrom(31, vip.DefaultMaskIPv4) // For ipv4
 	if family == ipvs.INET6 {
-		netMask = netmask.MaskFrom(128, 128) // For ipv6
+		netMask = netmask.MaskFrom(128, vip.DefaultMaskIPv6) // For ipv6
+	}
+
+	var fwmark uint32
+
+	if nftables && forwardingMethod == "masquerade" && family == ipvs.INET {
+		fwmark = network.IPVSMark()
 	}
 
 	// Generate out API Server LoadBalancer instance
@@ -87,9 +113,10 @@ func NewIPVSLB(address string, port int, forwardingMethod string, backendHealthC
 		Netmask:   netMask,
 		Family:    family,
 		Protocol:  ipvs.TCP,
-		Port:      uint16(port),
+		Port:      port,
 		Address:   ip,
 		Scheduler: ROUNDROBIN,
+		FWMark:    fwmark,
 	}
 
 	var m ipvs.ForwardType
@@ -106,7 +133,7 @@ func NewIPVSLB(address string, port int, forwardingMethod string, backendHealthC
 		m = ipvs.Bypass
 	default:
 		m = ipvs.Local
-		log.Warnf("unknown forwarding method. Defaulting to Local")
+		log.Warn("unknown forwarding method. Defaulting to Local")
 	}
 
 	if backendHealthCheckInterval <= 0 {
@@ -120,18 +147,32 @@ func NewIPVSLB(address string, port int, forwardingMethod string, backendHealthC
 		forwardingMethod:    m,
 		interval:            backendHealthCheckInterval,
 		backendMap:          make(backend.Map),
+		killFunc:            killFunc,
+		address:             address,
+		family:              family,
 	}
 
-	if strings.ToLower(forwardingMethod) == "masquerade" {
-		go lb.healthCheck()
-	}
+	wg.Go(func() {
+		lb.healthCheck(ctx)
+	})
 
 	// Return our created load-balancer
 	return lb, nil
 }
 
+func enableProcSys(path, name string) error {
+	isSet, err := sysctl.EnableProcSys(path)
+	if err != nil {
+		return fmt.Errorf("ensuring %s enabled: %w", name, err)
+	}
+	if isSet {
+		log.Info(fmt.Sprintf("sysctl set %s to 1", name))
+	}
+	return nil
+}
+
 func (lb *IPVSLoadBalancer) RemoveIPVSLB() error {
-	close(lb.stop)
+	log.Info("Stopping IPVS LoadBalancer", "address", lb.address)
 	err := lb.client.RemoveService(lb.loadBalancerService)
 	if err != nil {
 		return fmt.Errorf("error removing existing IPVS service: %v", err)
@@ -139,15 +180,32 @@ func (lb *IPVSLoadBalancer) RemoveIPVSLB() error {
 	return nil
 }
 
-func (lb *IPVSLoadBalancer) AddBackend(address string, port int) error {
-	backend := backend.Entry{Addr: address, Port: port}
+func (lb *IPVSLoadBalancer) AddBackend(address string, port uint16) error {
+	isLocal := false
+	var err error
+
+	// Discard backend if it is of different IP family than LB address.
+	if _, family := ipAndFamily(address); family != lb.family {
+		return nil
+	}
+
+	if lb.forwardingMethod == ipvs.Local {
+		log.Info("checking if backend is local", "addr", address)
+		isLocal, err = lb.isLocal(address)
+		if err != nil {
+			log.Error("checking if backend is local", "err", err)
+		}
+		log.Info("checked if backend is local", "addr", address, "local", isLocal)
+	}
+
+	backend := backend.Entry{Addr: address, Port: port, IsLocal: isLocal}
 
 	lb.lock.Lock()
 	defer lb.lock.Unlock()
 	if _, ok := lb.backendMap[backend]; !ok {
 		isHealth := backend.Check()
 		if isHealth {
-			err := lb.addBackend(backend)
+			err := lb.addBackend(address, port)
 			if err != nil {
 				return err
 			}
@@ -157,18 +215,19 @@ func (lb *IPVSLoadBalancer) AddBackend(address string, port int) error {
 	return nil
 }
 
-func (lb *IPVSLoadBalancer) addBackend(backend backend.Entry) error {
+func (lb *IPVSLoadBalancer) addBackend(address string, port uint16) error {
+	backend := backend.Entry{Addr: address, Port: port}
 	// Check if this is the first backend
 	backends, err := lb.client.Destinations(lb.loadBalancerService)
 	if err != nil && strings.Contains(err.Error(), "file does not exist") {
-		log.Errorf("Error querying backends %s", err)
+		log.Error("querying backends", "err", err)
 	}
 	// If this is our first backend, then we can create the load-balancer service and add a backend
 	if len(backends) == 0 {
 		err = lb.client.CreateService(lb.loadBalancerService)
 		// If we've an error it could be that the IPVS lb instance has been left from a previous leadership
 		if err != nil && strings.Contains(err.Error(), "file exists") {
-			log.Warnf("load balancer for API server already exists, attempting to remove and re-create")
+			log.Warn("load balancer for API server already exists, attempting to remove and re-create")
 			err = lb.client.RemoveService(lb.loadBalancerService)
 			if err != nil {
 				return fmt.Errorf("error re-creating IPVS service: %v", err)
@@ -179,10 +238,12 @@ func (lb *IPVSLoadBalancer) addBackend(backend backend.Entry) error {
 			}
 		} else if err != nil {
 			// Fatal error at this point as IPVS is probably not working
-			log.Errorf("Unable to create an IPVS service, ensure IPVS kernel modules are loaded")
-			log.Fatalf("IPVS service error: %v", err)
+			log.Error("Unable to create an IPVS service, ensure IPVS kernel modules are loaded")
+			log.Error("IPVS service", "err", err)
+			return utils.WrapPanicError(err, "unable to create an IPVS service")
+
 		}
-		log.Infof("Created Load-Balancer services on [%s:%d]", lb.addrString(), lb.Port)
+		log.Info("load-Balancer services created", "address", lb.addrString(), "port", lb.Port)
 	}
 
 	ip, family := ipAndFamily(backend.Addr)
@@ -195,7 +256,7 @@ func (lb *IPVSLoadBalancer) addBackend(backend backend.Entry) error {
 
 	dst := ipvs.Destination{
 		Address:   ip,
-		Port:      uint16(backend.Port),
+		Port:      backend.Port,
 		Family:    family,
 		Weight:    1,
 		FwdMethod: lb.forwardingMethod,
@@ -211,12 +272,12 @@ func (lb *IPVSLoadBalancer) addBackend(backend backend.Entry) error {
 		// file exists is fine, we will just return at this point
 		return nil
 	}
-	log.Infof("Added backend for [%s:%d] on [%s:%d]", lb.addrString(), lb.Port, backend.Addr, backend.Port)
+	log.Info("backend added", "src addr", lb.addrString(), "src port", lb.Port, "dst addr", backend.Addr, "dst port", backend.Port)
 
 	return nil
 }
 
-func (lb *IPVSLoadBalancer) RemoveBackend(address string, port int) error {
+func (lb *IPVSLoadBalancer) RemoveBackend(address string, port uint16) error {
 	backend := backend.Entry{Addr: address, Port: port}
 
 	lb.lock.Lock()
@@ -233,7 +294,7 @@ func (lb *IPVSLoadBalancer) RemoveBackend(address string, port int) error {
 	return nil
 }
 
-func (lb *IPVSLoadBalancer) removeBackend(address string, port int) error {
+func (lb *IPVSLoadBalancer) removeBackend(address string, port uint16) error {
 	ip, family := ipAndFamily(address)
 	if family != lb.loadBalancerService.Family {
 		return nil
@@ -241,12 +302,12 @@ func (lb *IPVSLoadBalancer) removeBackend(address string, port int) error {
 
 	dst := ipvs.Destination{
 		Address: ip,
-		Port:    uint16(port),
+		Port:    port,
 		Family:  family,
 		Weight:  1,
 	}
 	err := lb.client.RemoveDestination(lb.loadBalancerService, dst)
-	if err != nil {
+	if err != nil && !errors.Is(err, os.ErrNotExist) {
 		return fmt.Errorf("error removing backend: %v", err)
 	}
 	return nil
@@ -264,32 +325,87 @@ func ipAndFamily(address string) (netip.Addr, ipvs.AddressFamily) {
 	return netip.AddrFrom4([4]byte(ipAddr.To4())), ipvs.INET
 }
 
-func (lb *IPVSLoadBalancer) healthCheck() {
-	backend.Watch(func() {
+func (lb *IPVSLoadBalancer) healthCheck(ctx context.Context) {
+	backend.Watch(ctx, lb.interval, func() {
 		lb.lock.Lock()
+		defer lb.lock.Unlock()
 		for backend, oldStatus := range lb.backendMap {
 			newStatus := backend.Check()
 			if newStatus {
 				// old status -> health
 				if !oldStatus {
-					err := lb.AddBackend(backend.Addr, backend.Port)
+					err := lb.addBackend(backend.Addr, backend.Port)
 					if err != nil {
-						log.Errorf("failed to add backend: %s", err)
+						log.Error("add backend", "err", err)
 					}
 					lb.backendMap[backend] = newStatus
 				}
 			} else {
 				// old status -> not health
 				if oldStatus {
-					log.Infof("healthCheck failed for backend %s:%d, attempting to remove from load balancer", backend.Addr, backend.Port)
+					log.Info("healthCheck failed - removing backend", "address", backend.Addr, "port", backend.Port)
 					err := lb.removeBackend(backend.Addr, backend.Port)
 					if err != nil {
-						log.Errorf("failed to remove backend %s:%d: %s", backend.Addr, backend.Port, err)
+						log.Error("failed to remove backend", "address", backend.Addr, "port", backend.Port, "err", err)
 					}
 					lb.backendMap[backend] = newStatus
 				}
+				if lb.forwardingMethod == ipvs.Local && !lb.localBackendExists() {
+					if lb.killFunc != nil {
+						log.Error("no local backends available, restarting kube-vip")
+						lb.killFunc()
+					}
+				}
 			}
 		}
-		lb.lock.Unlock()
-	}, lb.interval, lb.stop)
+	})
+}
+
+func (lb *IPVSLoadBalancer) isLocal(address string) (bool, error) {
+	target := net.ParseIP(address)
+	if target == nil {
+		return false, fmt.Errorf("unable to parse IP address %s", address)
+	}
+
+	links, err := netlink.LinkList()
+	if err != nil {
+		return false, fmt.Errorf("listing links: %w", err)
+	}
+
+	family := netlink.FAMILY_V4
+	if utils.IsIPv6(address) {
+		family = netlink.FAMILY_V6
+	}
+
+	for _, link := range links {
+		if link.Type() == "veth" {
+			continue
+		}
+
+		addrs, err := netlink.AddrList(link, family)
+		if err != nil {
+			log.Error("listing addresses", "link", link.Attrs().Name, "error", err.Error())
+			continue
+		}
+
+		for _, addr := range addrs {
+			if addr.Scope != int(netlink.SCOPE_UNIVERSE) {
+				continue
+			}
+			if addr.IP.Equal(target) {
+				return true, nil
+			}
+		}
+	}
+
+	return false, nil
+}
+
+func (lb *IPVSLoadBalancer) localBackendExists() bool {
+	for backend, isHealthy := range lb.backendMap {
+		if backend.IsLocal && isHealthy {
+			return true
+		}
+	}
+	return false
 }

pkg/loadbalancer/ipvs_test.go

@@ -6,6 +6,7 @@ import (
 	"testing"
 
 	"github.com/cloudflare/ipvs"
+	"github.com/kube-vip/kube-vip/pkg/utils"
 )
 
 func Test_ipAndFamily(t *testing.T) {
@@ -19,7 +20,7 @@ func Test_ipAndFamily(t *testing.T) {
 		want1 ipvs.AddressFamily
 	}{
 		{
-			name: "IPv4",
+			name: utils.IPv4Family,
 			args: args{
 				address: "192.168.0.20",
 			},
@@ -27,7 +28,7 @@ func Test_ipAndFamily(t *testing.T) {
 			want1: ipvs.INET,
 		},
 		{
-			name: "IPv6",
+			name: utils.IPv6Family,
 			args: args{
 				address: "ff02::3",
 			},

pkg/manager/cluster.go

@@ -1,30 +0,0 @@
-package manager
-
-import (
-	"github.com/pkg/errors"
-
-	"github.com/kube-vip/kube-vip/pkg/cluster"
-	"github.com/kube-vip/kube-vip/pkg/etcd"
-)
-
-func initClusterManager(sm *Manager) (*cluster.Manager, error) {
-	m := &cluster.Manager{
-		SignalChan: sm.signalChan,
-	}
-
-	switch sm.config.LeaderElectionType {
-	case "kubernetes", "":
-		m.KubernetesClient = sm.clientSet
-		m.RetryWatcherClient = sm.rwClientSet
-	case "etcd":
-		client, err := etcd.NewClient(sm.config)
-		if err != nil {
-			return nil, err
-		}
-		m.EtcdClient = client
-	default:
-		return nil, errors.Errorf("invalid LeaderElectionMode %s not supported", sm.config.LeaderElectionType)
-	}
-
-	return m, nil
-}

pkg/manager/instance.go

@@ -1,388 +0,0 @@
-package manager
-
-import (
-	"fmt"
-	"net"
-	"strconv"
-
-	log "github.com/sirupsen/logrus"
-	"github.com/vishvananda/netlink"
-	v1 "k8s.io/api/core/v1"
-
-	"github.com/kube-vip/kube-vip/pkg/cluster"
-	"github.com/kube-vip/kube-vip/pkg/kubevip"
-	"github.com/kube-vip/kube-vip/pkg/vip"
-)
-
-// Instance defines an instance of everything needed to manage vips
-type Instance struct {
-	// Virtual IP / Load Balancer configuration
-	vipConfigs []*kubevip.Config
-
-	// cluster instances
-	clusters []*cluster.Cluster
-
-	// Service uses DHCP
-	isDHCP              bool
-	dhcpInterface       string
-	dhcpInterfaceHwaddr string
-	dhcpInterfaceIP     string
-	dhcpHostname        string
-	dhcpClient          *vip.DHCPClient
-
-	// External Gateway IP the service is forwarded from
-	upnpGatewayIPs []string
-
-	// Kubernetes service mapping
-	VIPs          []string
-	UID           string
-	ExternalPorts []Port
-
-	serviceSnapshot *v1.Service
-}
-
-type Port struct {
-	Port int32
-	Type string
-}
-
-func NewInstance(svc *v1.Service, config *kubevip.Config) (*Instance, error) {
-	instanceAddresses := fetchServiceAddresses(svc)
-	instanceUID := string(svc.UID)
-
-	var newVips []*kubevip.Config
-	var link netlink.Link
-	var err error
-
-	for _, address := range instanceAddresses {
-		// Detect if we're using a specific interface for services
-		var svcInterface string
-		svcInterface = svc.Annotations[serviceInterface] // If the service has a specific interface defined, then use it
-		if svcInterface == kubevip.Auto {
-			link, err = autoFindInterface(address)
-			if err != nil {
-				log.Errorf("failed to automatically discover network interface for annotated IP address [%s] with error: %s", address, err.Error())
-			} else {
-				if link == nil {
-					log.Errorf("failed to automatically discover network interface for annotated IP address [%s]", address)
-				}
-			}
-			if link == nil {
-				svcInterface = ""
-			} else {
-				svcInterface = getAutoInterfaceName(link, config.Interface)
-			}
-		}
-		// If it is still blank then use the
-		if svcInterface == "" {
-			switch config.ServicesInterface {
-			case kubevip.Auto:
-				link, err = autoFindInterface(address)
-				if err != nil {
-					log.Errorf("failed to automatically discover network interface for IP address [%s] with error: %s - defaulting to: %s", address, err.Error(), config.Interface)
-				} else if link == nil {
-					log.Errorf("failed to automatically discover network interface for IP address [%s] - defaulting to: %s", address, config.Interface)
-				}
-				svcInterface = getAutoInterfaceName(link, config.Interface)
-			case "":
-				svcInterface = config.Interface
-			default:
-				svcInterface = config.ServicesInterface
-			}
-		}
-
-		if link == nil {
-			if link, err = netlink.LinkByName(svcInterface); err != nil {
-				return nil, fmt.Errorf("failed to get interface %s: %w", svcInterface, err)
-			}
-			if link == nil {
-				return nil, fmt.Errorf("failed to get interface %s", svcInterface)
-			}
-		}
-
-		cidrs := vip.Split(config.VIPCIDR)
-
-		ipv4AutoSubnet := false
-		ipv6AutoSubnet := false
-		if cidrs[0] == kubevip.Auto {
-			ipv4AutoSubnet = true
-		}
-
-		if len(cidrs) > 1 && cidrs[1] == kubevip.Auto {
-			ipv6AutoSubnet = true
-		}
-
-		if (config.Address != "" || config.VIP != "") && (ipv4AutoSubnet || ipv6AutoSubnet) {
-			return nil, fmt.Errorf("auto subnet discovery cannot be used if VIP address was provided")
-		}
-
-		subnet := ""
-		var err error
-		if vip.IsIPv4(address) {
-			if ipv4AutoSubnet {
-				subnet, err = autoFindSubnet(link, address)
-				if err != nil {
-					return nil, fmt.Errorf("failed to automatically find subnet for service %s/%s with IP address %s on interface %s: %w", svc.Namespace, svc.Name, address, svcInterface, err)
-				}
-			} else {
-				if cidrs[0] != "" {
-					subnet = cidrs[0]
-				} else {
-					subnet = "32"
-				}
-			}
-		} else {
-			if ipv6AutoSubnet {
-				subnet, err = autoFindSubnet(link, address)
-				if err != nil {
-					return nil, fmt.Errorf("failed to automatically find subnet for service %s/%s with IP address %s on interface %s: %w", svc.Namespace, svc.Name, address, svcInterface, err)
-				}
-			} else {
-				if len(cidrs) > 1 {
-					subnet = cidrs[1]
-				} else {
-					subnet = "128"
-				}
-			}
-		}
-
-		//log.Info("new instance", "svc", *svc, "interface", svcInterface)
-
-		// Generate new Virtual IP configuration
-		newVips = append(newVips, &kubevip.Config{
-			VIP:                    address,
-			Interface:              svcInterface,
-			SingleNode:             true,
-			EnableARP:              config.EnableARP,
-			EnableBGP:              config.EnableBGP,
-			VIPCIDR:                subnet,
-			VIPSubnet:              config.VIPSubnet,
-			EnableRoutingTable:     config.EnableRoutingTable,
-			RoutingTableID:         config.RoutingTableID,
-			RoutingTableType:       config.RoutingTableType,
-			RoutingProtocol:        config.RoutingProtocol,
-			ArpBroadcastRate:       config.ArpBroadcastRate,
-			EnableServiceSecurity:  config.EnableServiceSecurity,
-			DNSMode:                config.DNSMode,
-			DisableServiceUpdates:  config.DisableServiceUpdates,
-			EnableServicesElection: config.EnableServicesElection,
-			KubernetesLeaderElection: kubevip.KubernetesLeaderElection{
-				EnableLeaderElection: config.EnableLeaderElection,
-			},
-		})
-	}
-
-	// Create new service
-	instance := &Instance{
-		UID:             instanceUID,
-		VIPs:            instanceAddresses,
-		serviceSnapshot: svc,
-	}
-	for _, port := range svc.Spec.Ports {
-		instance.ExternalPorts = append(instance.ExternalPorts, Port{
-			Port: port.Port,
-			Type: string(port.Protocol),
-		})
-	}
-
-	if svc.Annotations != nil {
-		instance.dhcpInterfaceHwaddr = svc.Annotations[hwAddrKey]
-		instance.dhcpInterfaceIP = svc.Annotations[requestedIP]
-		instance.dhcpHostname = svc.Annotations[loadbalancerHostname]
-	}
-
-	configPorts := make([]kubevip.Port, 0)
-	for _, p := range instance.ExternalPorts {
-		configPorts = append(configPorts, kubevip.Port{
-			Type: p.Type,
-			Port: int(p.Port),
-		})
-	}
-	// Generate Load Balancer config
-	newLB := kubevip.LoadBalancer{
-		Name:      fmt.Sprintf("%s-load-balancer", svc.Name),
-		Ports:     configPorts,
-		BindToVip: true,
-	}
-	for _, vip := range newVips {
-		// Add Load Balancer Configuration
-		vip.LoadBalancers = append(vip.LoadBalancers, newLB)
-	}
-	// Create Add configuration to the new service
-	instance.vipConfigs = newVips
-
-	// If this was purposely created with the address 0.0.0.0,
-	// we will create a macvlan on the main interface and a DHCP client
-	// TODO: Consider how best to handle DHCP with multiple addresses
-	if len(instanceAddresses) == 1 && instanceAddresses[0] == "0.0.0.0" {
-		err := instance.startDHCP()
-		if err != nil {
-			return nil, err
-		}
-		select {
-		case err := <-instance.dhcpClient.ErrorChannel():
-			return nil, fmt.Errorf("error starting DHCP for %s/%s: error: %s",
-				instance.serviceSnapshot.Namespace, instance.serviceSnapshot.Name, err)
-		case ip := <-instance.dhcpClient.IPChannel():
-			instance.vipConfigs[0].Interface = instance.dhcpInterface
-			instance.vipConfigs[0].VIP = ip
-			instance.dhcpInterfaceIP = ip
-		}
-	}
-
-	for _, vipConfig := range instance.vipConfigs {
-		c, err := cluster.InitCluster(vipConfig, false)
-		if err != nil {
-			log.Errorf("Failed to add Service %s/%s", svc.Namespace, svc.Name)
-			return nil, err
-		}
-
-		for i := range c.Network {
-			c.Network[i].SetServicePorts(svc)
-		}
-
-		instance.clusters = append(instance.clusters, c)
-		log.Infof("(svcs) adding VIP [%s] via %s for [%s/%s]", vipConfig.VIP, vipConfig.Interface, svc.Namespace, svc.Name)
-
-	}
-
-	return instance, nil
-}
-
-func autoFindInterface(ip string) (netlink.Link, error) {
-	links, err := netlink.LinkList()
-	if err != nil {
-		return nil, fmt.Errorf("failed to list network interfaces: %w", err)
-	}
-
-	address := net.ParseIP(ip)
-
-	family := netlink.FAMILY_V4
-
-	if address.To4() == nil {
-		family = netlink.FAMILY_V6
-	}
-
-	for _, link := range links {
-		addr, err := netlink.AddrList(link, family)
-		if err != nil {
-			return nil, fmt.Errorf("failed to get IP addresses for interface %s: %w", link.Attrs().Name, err)
-		}
-		for _, a := range addr {
-			if a.IPNet.Contains(address) {
-				return link, nil
-			}
-		}
-	}
-
-	return nil, nil
-}
-
-func autoFindSubnet(link netlink.Link, ip string) (string, error) {
-	address := net.ParseIP(ip)
-
-	family := netlink.FAMILY_V4
-	if address.To4() == nil {
-		family = netlink.FAMILY_V6
-	}
-
-	addr, err := netlink.AddrList(link, family)
-	if err != nil {
-		return "", fmt.Errorf("failed to get IP addresses for interface %s: %w", link.Attrs().Name, err)
-	}
-	for _, a := range addr {
-		if a.IPNet.Contains(address) {
-			m, _ := a.IPNet.Mask.Size()
-			return strconv.Itoa(m), nil
-		}
-	}
-	return "", fmt.Errorf("failed to find suitable subnet for address %s", ip)
-}
-
-func getAutoInterfaceName(link netlink.Link, defaultInterface string) string {
-	if link == nil {
-		return defaultInterface
-	}
-	return link.Attrs().Name
-}
-
-func (i *Instance) startDHCP() error {
-	if len(i.vipConfigs) != 1 {
-		return fmt.Errorf("DHCP requires exactly 1 VIP config, got: %v", len(i.vipConfigs))
-	}
-	parent, err := netlink.LinkByName(i.vipConfigs[0].Interface)
-	if err != nil {
-		return fmt.Errorf("error finding VIP Interface, for building DHCP Link : %v", err)
-	}
-
-	// Generate name from UID
-	interfaceName := fmt.Sprintf("vip-%s", i.UID[0:8])
-
-	// Check if the interface doesn't exist first
-	iface, err := net.InterfaceByName(interfaceName)
-	if err != nil {
-		log.Infof("Creating new macvlan interface for DHCP [%s]", interfaceName)
-
-		hwaddr, err := net.ParseMAC(i.dhcpInterfaceHwaddr)
-		if i.dhcpInterfaceHwaddr != "" && err != nil {
-			return err
-		} else if hwaddr == nil {
-			hwaddr, err = net.ParseMAC(vip.GenerateMac())
-			if err != nil {
-				return err
-			}
-		}
-
-		log.Infof("New interface [%s] mac is %s", interfaceName, hwaddr)
-		mac := &netlink.Macvlan{
-			LinkAttrs: netlink.LinkAttrs{
-				Name:         interfaceName,
-				ParentIndex:  parent.Attrs().Index,
-				HardwareAddr: hwaddr,
-			},
-			Mode: netlink.MACVLAN_MODE_DEFAULT,
-		}
-
-		err = netlink.LinkAdd(mac)
-		if err != nil {
-			return fmt.Errorf("could not add %s: %v", interfaceName, err)
-		}
-
-		err = netlink.LinkSetUp(mac)
-		if err != nil {
-			return fmt.Errorf("could not bring up interface [%s] : %v", interfaceName, err)
-		}
-
-		iface, err = net.InterfaceByName(interfaceName)
-		if err != nil {
-			return fmt.Errorf("error finding new DHCP interface by name [%v]", err)
-		}
-	} else {
-		log.Infof("Using existing macvlan interface for DHCP [%s]", interfaceName)
-	}
-
-	var initRebootFlag bool
-	if i.dhcpInterfaceIP != "" {
-		initRebootFlag = true
-	}
-
-	client := vip.NewDHCPClient(iface, initRebootFlag, i.dhcpInterfaceIP)
-
-	// Add hostname to dhcp client if annotated
-	if i.dhcpHostname != "" {
-		log.Infof("Hostname specified for dhcp lease: [%s] - [%s]", interfaceName, i.dhcpHostname)
-		client.WithHostName(i.dhcpHostname)
-	}
-
-	go client.Start()
-
-	// Set that DHCP is enabled
-	i.isDHCP = true
-	// Set the name of the interface so that it can be removed on Service deletion
-	i.dhcpInterface = interfaceName
-	i.dhcpInterfaceHwaddr = iface.HardwareAddr.String()
-	// Add the client so that we can call it to stop function
-	i.dhcpClient = client
-
-	return nil
-}

pkg/manager/manager.go

@@ -3,30 +3,40 @@ package manager
 import (
 	"context"
 	"fmt"
+	"net/http"
 	"os"
 	"os/signal"
 	"path/filepath"
-	"strconv"
 	"strings"
 	"sync"
+	"sync/atomic"
 	"syscall"
 	"time"
 
+	log "log/slog"
+
+	"github.com/kube-vip/kube-vip/pkg/arp"
 	"github.com/kube-vip/kube-vip/pkg/bgp"
+	"github.com/kube-vip/kube-vip/pkg/cluster"
+	"github.com/kube-vip/kube-vip/pkg/election"
+	"github.com/kube-vip/kube-vip/pkg/iptables"
 	"github.com/kube-vip/kube-vip/pkg/k8s"
 	"github.com/kube-vip/kube-vip/pkg/kubevip"
-	"github.com/kube-vip/kube-vip/pkg/trafficmirror"
+	"github.com/kube-vip/kube-vip/pkg/lease"
+	"github.com/kube-vip/kube-vip/pkg/manager/worker"
+	"github.com/kube-vip/kube-vip/pkg/networkinterface"
+	"github.com/kube-vip/kube-vip/pkg/nftables"
+	"github.com/kube-vip/kube-vip/pkg/node"
+	"github.com/kube-vip/kube-vip/pkg/route"
+	"github.com/kube-vip/kube-vip/pkg/services"
 	"github.com/kube-vip/kube-vip/pkg/upnp"
 	"github.com/kube-vip/kube-vip/pkg/utils"
+	"github.com/kube-vip/kube-vip/pkg/vip"
 	"github.com/prometheus/client_golang/prometheus"
-	log "github.com/sirupsen/logrus"
-	v1 "k8s.io/api/core/v1"
 	"k8s.io/client-go/kubernetes"
 	"k8s.io/client-go/rest"
 )
 
-const plunderLock = "plndr-svcs-lock"
-
 // Manager degines the manager of the load-balancing services
 type Manager struct {
 	clientSet   *kubernetes.Clientset
@@ -37,34 +47,49 @@ type Manager struct {
 	// Manager services
 	// service bool
 
-	// Keeps track of all running instances
-	serviceInstances []*Instance
-
-	// UPNP functionality
-	upnp bool
 	// BGP Manager, this is a singleton that manages all BGP advertisements
 	bgpServer *bgp.Server
 
 	// This channel is used to catch an OS signal and trigger a shutdown
 	signalChan chan os.Signal
 
-	// This channel is used to signal a shutdown
-	shutdownChan chan struct{}
+	sigint sync.Once
+
+	svcProcessor *services.Processor
 
 	// This is a prometheus counter used to count the number of events received
 	// from the service watcher
 	countServiceWatchEvent *prometheus.CounterVec
 
-	// This is a prometheus gauge indicating the state of the sessions.
-	// 1 means "ESTABLISHED", 0 means "NOT ESTABLISHED"
-	bgpSessionInfoGauge *prometheus.GaugeVec
-
 	// This mutex is to protect calls from various goroutines
 	mutex sync.Mutex
+
+	// This tracks used network interfaces and guards them with mutex for concurrent changes.
+	intfMgr *networkinterface.Manager
+
+	// This tracks VIPs and performs ARP/NDP advertisement.
+	arpMgr *arp.Manager
+
+	// This tracks node labels and performs label management
+	// implementation will be decided in constructor
+	// based on config.EnableNodeLabeling
+	nodeLabelManager node.LabelManager
+
+	// This variable reports if manager is being closed
+	closing atomic.Bool
+
+	// Will be used for leaderelection when required
+	electionMgr *election.Manager
+
+	// Will handle leases
+	leaseMgr *lease.Manager
+
+	// Will handle routes
+	routeMgr *route.Manager
 }
 
 // New will create a new managing object
-func New(configMap string, config *kubevip.Config) (*Manager, error) {
+func New(ctx context.Context, configMap string, config *kubevip.Config) (*Manager, error) {
 
 	// Instance identity should be the same as k8s node name to ensure better compatibility.
 	// By default k8s sets node name to `hostname -s`,
@@ -72,14 +97,15 @@ func New(configMap string, config *kubevip.Config) (*Manager, error) {
 	// we set it to hostname as a fallback.
 	// This mimics legacy behavior and should work on old kube-vip installations.
 	if config.NodeName == "" {
-		log.Warning("Node name is missing from the config, fall back to hostname")
+		log.Warn("Node name is missing from the config, fall back to hostname")
 		hostname, err := os.Hostname()
 		if err != nil {
 			return nil, fmt.Errorf("could not get hostname: %v", err)
 		}
 		config.NodeName = hostname
 	}
-	log.Infof("Using node name [%v]", config.NodeName)
+	config.NodeName = normalizeNodeName(config.NodeName)
+	log.Info("using node name", "name", config.NodeName)
 
 	adminConfigPath := "/etc/kubernetes/admin.conf"
 	homeConfigPath := filepath.Join(os.Getenv("HOME"), ".kube", "config")
@@ -91,9 +117,24 @@ func New(configMap string, config *kubevip.Config) (*Manager, error) {
 	switch {
 	case config.LeaderElectionType == "etcd":
 		// Do nothing, we don't construct a k8s client for etcd leader election
+	case config.K8sConfigFile != "" && config.K8sConfigFile != adminConfigPath &&
+		config.K8sConfigFile != homeConfigPath && utils.FileExists(config.K8sConfigFile):
+		// An explicitly configured kubeconfig (k8s_config_file env or
+		// --k8sConfigPath) takes precedence over the well-known host paths.
+		// KubernetesAddr, when set, overrides the API endpoint - static pods
+		// on control plane hosts use it to reach their local API server
+		// instead of a VIP that may not be up yet.
+		clientConfig, err = k8s.NewRestConfig(config.K8sConfigFile, false, config.KubernetesAddr)
+		if err != nil {
+			return nil, fmt.Errorf("could not create k8s REST config from file %q: %w", config.K8sConfigFile, err)
+		}
+		if clientset, err = k8s.NewClientset(clientConfig); err != nil {
+			return nil, fmt.Errorf("could not create k8s clientset: %w", err)
+		}
+		log.Info("Using Kubernetes configuration from explicit file", "path", config.K8sConfigFile, "address", config.KubernetesAddr)
 	case utils.FileExists(adminConfigPath):
 		if config.KubernetesAddr != "" {
-			log.Infof("k8s address [%s]", config.KubernetesAddr)
+			log.Info("k8s address", "address", config.KubernetesAddr)
 			clientConfig, err = k8s.NewRestConfig(adminConfigPath, false, config.KubernetesAddr)
 		} else if config.EnableControlPlane {
 			// If this is a control plane host it will likely have started as a static pod or won't have the
@@ -114,7 +155,7 @@ func New(configMap string, config *kubevip.Config) (*Manager, error) {
 		if clientset, err = k8s.NewClientset(clientConfig); err != nil {
 			return nil, fmt.Errorf("could not create k8s clientset: %w", err)
 		}
-		log.Debugf("Using external Kubernetes configuration from file: %q", adminConfigPath)
+		log.Debug("Using external Kubernetes configuration from file", "path", adminConfigPath)
 	case utils.FileExists(homeConfigPath):
 		clientConfig, err = k8s.NewRestConfig(homeConfigPath, false, "")
 		if err != nil {
@@ -124,7 +165,7 @@ func New(configMap string, config *kubevip.Config) (*Manager, error) {
 		if err != nil {
 			return nil, fmt.Errorf("could not create k8s clientset from external file: %q: %w", homeConfigPath, err)
 		}
-		log.Debugf("Using external Kubernetes configuration from file [%s]", homeConfigPath)
+		log.Debug("Using external Kubernetes configuration from file", "path", adminConfigPath)
 	default:
 		clientConfig, err = k8s.NewRestConfig("", true, "")
 		if err != nil {
@@ -166,6 +207,57 @@ func New(configMap string, config *kubevip.Config) (*Manager, error) {
 	// 	}
 	// }
 
+	// listen for interrupts or the Linux SIGTERM signal and cancel
+	// our context, which the leader election code will observe and
+	// step down
+	signalChan := make(chan os.Signal, 1)
+	// Add Notification for Userland interrupt
+	signal.Notify(signalChan, syscall.SIGINT)
+
+	// Add Notification for SIGTERM (sent from Kubernetes)
+	signal.Notify(signalChan, syscall.SIGTERM)
+
+	// Add Notification for SIGUSR1 (for configuration dump)
+	signal.Notify(signalChan, syscall.SIGUSR1)
+
+	intfMgr := networkinterface.NewManager()
+	arpMgr := arp.NewManager(config)
+
+	// create the node label manager
+	// constructor will decide if it should be a noop or not
+	nodeLabelManager := node.NewManager(config, clientset)
+
+	var bgpServer *bgp.Server
+	// If BGP is enabled then we start a server instance that will broadcast VIPs
+	if config.EnableBGP {
+		var err error
+		// If Annotations have been set then we will look them up
+		if config.Annotations != "" {
+			err = annotationsWatcher(ctx, clientset, rwClientSet, config)
+		} else {
+			log.Debug("No Node annotations to parse")
+		}
+
+		if err != nil {
+			return nil, err
+		}
+		bgpServer, err = bgp.NewBGPServer(config.BGPConfig, log.Level(config.Logging))
+		if err != nil {
+			return nil, fmt.Errorf("creating BGP server: %w", err)
+		}
+	}
+
+	electionMgr, err := election.NewManager(config, clientset, rwClientSet)
+	if err != nil {
+		return nil, fmt.Errorf("creating election manager: %w", err)
+	}
+
+	leaseMgr := lease.NewManager()
+	routeMgr := route.NewManager()
+
+	svcProcessor := services.NewServicesProcessor(config, bgpServer, clientset, rwClientSet,
+		intfMgr, arpMgr, nodeLabelManager, electionMgr, leaseMgr, routeMgr)
+
 	return &Manager{
 		clientSet:   clientset,
 		rwClientSet: rwClientSet,
@@ -177,166 +269,192 @@ func New(configMap string, config *kubevip.Config) (*Manager, error) {
 			Name:      "all_services_events",
 			Help:      "Count all events fired by the service watcher categorised by event type",
 		}, []string{"type"}),
-		bgpSessionInfoGauge: prometheus.NewGaugeVec(prometheus.GaugeOpts{
-			Namespace: "kube_vip",
-			Subsystem: "manager",
-			Name:      "bgp_session_info",
-			Help:      "Display state of session by setting metric for label value with current state to 1",
-		}, []string{"state", "peer"}),
+		signalChan:       signalChan,
+		svcProcessor:     svcProcessor,
+		intfMgr:          intfMgr,
+		arpMgr:           arpMgr,
+		bgpServer:        bgpServer,
+		nodeLabelManager: nodeLabelManager,
+		electionMgr:      electionMgr,
+		leaseMgr:         leaseMgr,
+		routeMgr:         routeMgr,
 	}, nil
 }
 
 // Start will begin the Manager, which will start services and watch the configmap
-func (sm *Manager) Start() error {
-	// listen for interrupts or the Linux SIGTERM signal and cancel
-	// our context, which the leader election code will observe and
-	// step down
-	sm.signalChan = make(chan os.Signal, 1)
-	// Add Notification for Userland interrupt
-	signal.Notify(sm.signalChan, syscall.SIGINT)
+func (sm *Manager) Start(ctx context.Context) error {
+	wg := sync.WaitGroup{}
+	defer wg.Wait()
 
-	// Add Notification for SIGTERM (sent from Kubernetes)
-	signal.Notify(sm.signalChan, syscall.SIGTERM)
-
-	// All watchers and other goroutines should have an additional goroutine that blocks on this, to shut things down
-	sm.shutdownChan = make(chan struct{})
-
-	// If BGP is enabled then we start a server instance that will broadcast VIPs
-	if sm.config.EnableBGP {
-
-		// If Annotations have been set then we will look them up
-		err := sm.parseAnnotations()
-		if err != nil {
-			return err
+	// HealthCheck
+	if sm.config.HealthCheckPort != 0 {
+		if sm.config.HealthCheckPort < 1024 {
+			return fmt.Errorf("healthcheck port is using a port that is less than 1024 [%d]", sm.config.HealthCheckPort)
 		}
-
-		log.Infoln("Starting Kube-vip Manager with the BGP engine")
-		return sm.startBGP()
+		http.HandleFunc("/healthz", func(w http.ResponseWriter, _ *http.Request) {
+			fmt.Fprintf(w, "OK")
+		})
+		wg.Go(func() {
+			server := &http.Server{
+				Addr:              fmt.Sprintf(":%d", sm.config.HealthCheckPort),
+				ReadHeaderTimeout: 3 * time.Second,
+			}
+			err := server.ListenAndServe()
+			if err != nil {
+				log.Error("healthcheck", "unable to start", err)
+			}
+		})
 	}
 
-	if sm.config.EnableARP || sm.config.EnableWireguard {
-		// Before starting the leader Election enable any additional functionality
-		upnpEnabled, _ := strconv.ParseBool(os.Getenv("enableUPNP"))
+	// on exit, clean up the node labels
+	defer func() {
+		if err := sm.nodeLabelManager.CleanUpLabels(10 * time.Second); err != nil {
+			log.Error("CleanUpNodeLabels", "unable to cleanup node labels", err)
+		}
+	}()
 
-		if upnpEnabled {
-			sm.upnp = true
-			clients := upnp.GetConnectionClients(context.TODO())
+	if sm.config.EnableARP || sm.config.EnableWireguard {
+		if sm.config.EnableUPNP {
+			clients := upnp.GetConnectionClients(ctx)
 			if len(clients) == 0 {
-				log.Errorf("Error Enabling UPNP. No Clients found")
+				log.Error("Error Enabling UPNP. No Clients found")
 				// Set the struct to false so nothing should use it in future
-				sm.upnp = false
+				sm.config.EnableUPNP = false
 			} else {
 				for _, c := range clients {
 					ip, err := c.GetExternalIPAddress()
-					log.Infof("Found UPNP IGD2 Gateway address[%s] error: [%s]", ip, err)
+					if err != nil {
+						log.Error("unable to find IGD2 Gateway address", "err", err)
+					}
+					log.Info("Found UPNP IGD2 Gateway address", "ip", ip)
+				}
+			}
+			// TODO: It would be nice to run the UPNP refresh only on the leader.
+			wg.Go(func() {
+				sm.svcProcessor.RefreshUPNPForwards(ctx)
+			})
+		}
+	}
+
+	return sm.startMode(ctx)
+}
+
+// Start will begin the Manager, which will start services and watch the configmap
+func (sm *Manager) startMode(ctx context.Context) error {
+	var cpCluster *cluster.Cluster
+	var err error
+
+	w := worker.New(sm.arpMgr, sm.intfMgr, sm.config, &sm.closing, sm.Kill,
+		sm.svcProcessor, &sm.mutex, sm.clientSet, sm.bgpServer, sm.electionMgr,
+		sm.leaseMgr, sm.routeMgr, sm.nodeLabelManager)
+
+	// use a Go context so we can tell the leaderelection code when we
+	// want to step down
+	wg := sync.WaitGroup{}
+	modeCtx, cancel := context.WithCancel(ctx)
+	defer func() {
+
+		wg.Wait()
+		w.Cleanup()
+		cancel()
+		log.Info("Shutting down Kube-Vip")
+	}()
+
+	log.Info("starting Kube-vip Manager", "mode", w.Name())
+	if err := w.Configure(modeCtx, &wg); err != nil {
+		defer cancel()
+		return fmt.Errorf("failed to configure %s mode: %w", w.Name(), err)
+	}
+
+	if sm.config.EnableControlPlane {
+		err = w.InitControlPlane()
+		if err != nil {
+			defer cancel()
+			return err
+		}
+	}
+
+	// Shutdown function that will wait on this signal, unless we call it ourselves
+	wg.Go(func() {
+		sm.waitForShutdown(modeCtx, cancel, cpCluster)
+	})
+
+	if sm.config.EnableControlPlane {
+		wg.Go(func() {
+			w.StartControlPlane(modeCtx, sm.electionMgr)
+		})
+	}
+
+	if sm.config.EnableServices {
+		// This will tidy any dangling kube-vip iptables rules
+		if sm.config.EgressClean {
+			tableName := nftables.EgressTableBaseNameForInstance(sm.config.InstanceName)
+			err := nftables.ClearTablesWithName(tableName)
+			if err != nil {
+				log.Warn("[egress]", "mode", "nftables-internal", "clearing error", err)
+			} else {
+				log.Info("[egress]", "mode", "nftables-internal", "tables", "cleared")
+			}
+			// TODO: Deprecate the iptables code v1.2.x
+			err = vip.ClearIPTables(sm.config.EgressWithNftables, sm.config.ServiceNamespace, iptables.ProtocolIPv4)
+			if err != nil {
+				log.Info("[egress]", "legacy-iptables", sm.config.EgressWithNftables, "mode", "IPv4", "error", err)
+			}
+			err = vip.ClearIPTables(sm.config.EgressWithNftables, sm.config.ServiceNamespace, iptables.ProtocolIPv6)
+			if err != nil {
+				log.Info("[egress]", "legacy-iptables", sm.config.EgressWithNftables, "mode", "IPv6", "error", err)
+			}
+		}
+		w.ConfigureServices()
+
+		for {
+			select {
+			case <-modeCtx.Done():
+				return nil
+			default:
+				if err = w.StartServices(modeCtx); err != nil {
+					if utils.IsPanicError(err) {
+						sm.Kill()
+						return fmt.Errorf("failed to reconcile services, non-recoverable error: %w", err)
+					} else {
+						log.Error("failed to reconcile services, restarting", "error", err)
+					}
 				}
 			}
 		}
-		// TODO: It would be nice to run the UPNP refresh only on the leader.
-		go sm.refreshUPNPForwards()
 	}
 
-	// If ARP is enabled then we start a LeaderElection that will use ARP to advertise VIPs
-	if sm.config.EnableARP {
-		log.Infoln("Starting Kube-vip Manager with the ARP engine")
-		return sm.startARP(sm.config.NodeName)
-	}
-
-	if sm.config.EnableWireguard {
-		log.Infoln("Starting Kube-vip Manager with the Wireguard engine")
-		return sm.startWireguard(sm.config.NodeName)
-	}
-
-	if sm.config.EnableRoutingTable {
-		log.Infoln("Starting Kube-vip Manager with the Routing Table engine")
-		return sm.startTableMode(sm.config.NodeName)
-	}
-
-	log.Errorln("prematurely exiting Load-balancer as no modes [ARP/BGP/Wireguard] are enabled")
 	return nil
 }
 
-func returnNameSpace() (string, error) {
-	if data, err := os.ReadFile("/var/run/secrets/kubernetes.io/serviceaccount/namespace"); err == nil {
-		if ns := strings.TrimSpace(string(data)); len(ns) > 0 {
-			return ns, nil
-		}
-		return "", err
-	}
-	return "", fmt.Errorf("unable to find Namespace")
-}
-
-func (sm *Manager) parseAnnotations() error {
-	if sm.config.Annotations == "" {
-		log.Debugf("No Node annotations to parse")
-		return nil
-	}
-
-	err := sm.annotationsWatcher()
-	if err != nil {
-		return err
-	}
-	return nil
-}
-
-func (sm *Manager) serviceInterface() string {
-	svcIf := sm.config.Interface
-	if sm.config.ServicesInterface != "" {
-		svcIf = sm.config.ServicesInterface
-	}
-	return svcIf
-}
-
-func (sm *Manager) startTrafficMirroringIfEnabled() error {
-	if sm.config.MirrorDestInterface != "" {
-		svcIf := sm.serviceInterface()
-		log.Infof("mirroring traffic from interface %s to interface %s", svcIf, sm.config.MirrorDestInterface)
-		if err := trafficmirror.MirrorTrafficFromNIC(svcIf, sm.config.MirrorDestInterface); err != nil {
-			return err
-		}
-	} else {
-		log.Debug("skip starting traffic mirroring since it's not enabled.")
-	}
-	return nil
-}
-
-func (sm *Manager) stopTrafficMirroringIfEnabled() error {
-	if sm.config.MirrorDestInterface != "" {
-		svcIf := sm.serviceInterface()
-		log.Infof("clean up qdisc config on interface %s", svcIf)
-		if err := trafficmirror.CleanupQDSICFromNIC(svcIf); err != nil {
-			return err
-		}
-	} else {
-		log.Debug("skip stopping traffic mirroring since it's not enabled.")
-	}
-	return nil
-}
-
-func (sm *Manager) findServiceInstance(svc *v1.Service) *Instance {
-	svcUID := string(svc.UID)
-	log.Debugf("service UID: %s", svcUID)
-	for i := range sm.serviceInstances {
-		log.Debugf("saved service instance %d UID: %s", i, sm.serviceInstances[i].UID)
-		if sm.serviceInstances[i].UID == svcUID {
-			return sm.serviceInstances[i]
-		}
-	}
-	return nil
-}
-
-// Refresh UPNP Port Forwards for all Service Instances registered in the SM
-func (sm *Manager) refreshUPNPForwards() {
-	log.Info("Starting UPNP Port Refresher")
+func (sm *Manager) waitForShutdown(ctx context.Context, cancel context.CancelFunc, cpCluster *cluster.Cluster) {
 	for {
-		time.Sleep(300 * time.Second)
-
-		log.Infof("[UPNP] Refreshing %d Instances", len(sm.serviceInstances))
-		for i := range sm.serviceInstances {
-			sm.upnpMap(context.TODO(), sm.serviceInstances[i])
-			if err := sm.updateStatus(sm.serviceInstances[i]); err != nil {
-				log.Warnf("[UPNP] Error updating service IPs %s [%s]", sm.serviceInstances[i].serviceSnapshot.Name, err.Error())
+		sig := <-sm.signalChan
+		switch sig {
+		case syscall.SIGUSR1:
+			log.Info("Received SIGUSR1, dumping configuration")
+			sm.dumpConfiguration(ctx)
+		case syscall.SIGINT, syscall.SIGTERM:
+			sm.closing.Store(true)
+			log.Info("Received kube-vip termination, signaling shutdown")
+			if cpCluster != nil {
+				cpCluster.Stop()
 			}
+			// Cancel the context, which will in turn cancel the leadership and all goroutines
+			cancel()
+			return
 		}
 	}
 }
+
+func (sm *Manager) Kill() {
+	sm.sigint.Do(func() {
+		sm.signalChan <- syscall.SIGINT
+	})
+}
+
+// normalizeNodeName ensures the local machine hostname conforms to
+// Kubernetes RFC1123 node naming conventions (lowercase).
+func normalizeNodeName(name string) string {
+	return strings.ToLower(name)
+}

pkg/manager/manager_arp.go

@@ -1,157 +0,0 @@
-package manager
-
-import (
-	"context"
-	"os"
-	"syscall"
-	"time"
-
-	log "github.com/sirupsen/logrus"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/client-go/tools/leaderelection"
-	"k8s.io/client-go/tools/leaderelection/resourcelock"
-
-	"github.com/kube-vip/kube-vip/pkg/cluster"
-	"github.com/kube-vip/kube-vip/pkg/iptables"
-	"github.com/kube-vip/kube-vip/pkg/vip"
-)
-
-// Start will begin the Manager, which will start services and watch the configmap
-func (sm *Manager) startARP(id string) error {
-	var cpCluster *cluster.Cluster
-	var ns string
-	var err error
-
-	// use a Go context so we can tell the leaderelection code when we
-	// want to step down
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-
-	// Shutdown function that will wait on this signal, unless we call it ourselves
-	go func() {
-		<-sm.signalChan
-		log.Info("Received kube-vip termination, signaling shutdown")
-		if sm.config.EnableControlPlane {
-			cpCluster.Stop()
-		}
-		// Close all go routines
-		close(sm.shutdownChan)
-		// Cancel the context, which will in turn cancel the leadership
-		cancel()
-	}()
-
-	if sm.config.EnableControlPlane {
-		cpCluster, err = cluster.InitCluster(sm.config, false)
-		if err != nil {
-			return err
-		}
-
-		clusterManager, err := initClusterManager(sm)
-		if err != nil {
-			return err
-		}
-
-		go func() {
-			err := cpCluster.StartCluster(sm.config, clusterManager, nil)
-			if err != nil {
-				log.Errorf("Control Plane Error [%v]", err)
-				// Trigger the shutdown of this manager instance
-				sm.signalChan <- syscall.SIGINT
-
-			}
-		}()
-
-		// Check if we're also starting the services, if not we can sit and wait on the closing channel and return here
-		if !sm.config.EnableServices {
-			<-sm.signalChan
-			log.Infof("Shutting down Kube-Vip")
-
-			return nil
-		}
-
-		ns = sm.config.Namespace
-	} else {
-
-		ns, err = returnNameSpace()
-		if err != nil {
-			log.Warnf("unable to auto-detect namespace, dropping to [%s]", sm.config.Namespace)
-			ns = sm.config.Namespace
-		}
-	}
-
-	// This will tidy any dangling kube-vip iptables rules
-	if os.Getenv("EGRESS_CLEAN") != "" {
-		vip.ClearIPTables(sm.config.EgressWithNftables, sm.config.ServiceNamespace, iptables.ProtocolIPv4)
-	}
-
-	// Start a services watcher (all kube-vip pods will watch services), upon a new service
-	// a lock based upon that service is created that they will all leaderElection on
-	if sm.config.EnableServicesElection {
-		log.Infof("beginning watching services, leaderelection will happen for every service")
-		err = sm.startServicesWatchForLeaderElection(ctx)
-		if err != nil {
-			return err
-		}
-	} else {
-
-		log.Infof("beginning services leadership, namespace [%s], lock name [%s], id [%s]", ns, sm.config.ServicesLeaseName, id)
-		// we use the Lease lock type since edits to Leases are less common
-		// and fewer objects in the cluster watch "all Leases".
-		lock := &resourcelock.LeaseLock{
-			LeaseMeta: metav1.ObjectMeta{
-				Name:      sm.config.ServicesLeaseName,
-				Namespace: ns,
-			},
-			Client: sm.clientSet.CoordinationV1(),
-			LockConfig: resourcelock.ResourceLockConfig{
-				Identity: id,
-			},
-		}
-
-		// start the leader election code loop
-		leaderelection.RunOrDie(ctx, leaderelection.LeaderElectionConfig{
-			Lock: lock,
-			// IMPORTANT: you MUST ensure that any code you have that
-			// is protected by the lease must terminate **before**
-			// you call cancel. Otherwise, you could have a background
-			// loop still running and another process could
-			// get elected before your background loop finished, violating
-			// the stated goal of the lease.
-			ReleaseOnCancel: true,
-			LeaseDuration:   time.Duration(sm.config.LeaseDuration) * time.Second,
-			RenewDeadline:   time.Duration(sm.config.RenewDeadline) * time.Second,
-			RetryPeriod:     time.Duration(sm.config.RetryPeriod) * time.Second,
-			Callbacks: leaderelection.LeaderCallbacks{
-				OnStartedLeading: func(ctx context.Context) {
-					err = sm.servicesWatcher(ctx, sm.syncServices)
-					if err != nil {
-						log.Fatal(err)
-					}
-				},
-				OnStoppedLeading: func() {
-					// we can do cleanup here
-					log.Infof("leader lost: %s", id)
-					for _, instance := range sm.serviceInstances {
-						for _, cluster := range instance.clusters {
-							cluster.Stop()
-						}
-					}
-
-					log.Fatal("lost leadership, restarting kube-vip")
-				},
-				OnNewLeader: func(identity string) {
-					// we're notified when new leader elected
-					if sm.config.EnableNodeLabeling {
-						applyNodeLabel(sm.clientSet, sm.config.Address, id, identity)
-					}
-					if identity == id {
-						// I just got the lock
-						return
-					}
-					log.Infof("new leader elected: %s", identity)
-				},
-			},
-		})
-	}
-	return nil
-}

pkg/manager/manager_bgp.go

@@ -1,141 +0,0 @@
-package manager
-
-import (
-	"context"
-	"fmt"
-	"os"
-	"syscall"
-
-	"github.com/kube-vip/kube-vip/pkg/bgp"
-	"github.com/kube-vip/kube-vip/pkg/cluster"
-	"github.com/kube-vip/kube-vip/pkg/equinixmetal"
-	api "github.com/osrg/gobgp/v3/api"
-	"github.com/packethost/packngo"
-	"github.com/prometheus/client_golang/prometheus"
-	log "github.com/sirupsen/logrus"
-)
-
-// Start will begin the Manager, which will start services and watch the configmap
-func (sm *Manager) startBGP() error {
-	var cpCluster *cluster.Cluster
-	// var ns string
-	var err error
-
-	// If Equinix Metal is enabled then we can begin our preparation work
-	var packetClient *packngo.Client
-	if sm.config.EnableMetal {
-		if sm.config.ProviderConfig != "" {
-			key, project, err := equinixmetal.GetPacketConfig(sm.config.ProviderConfig)
-			if err != nil {
-				return err
-			}
-			// Set the environment variable with the key for the project
-			os.Setenv("PACKET_AUTH_TOKEN", key)
-			// Update the configuration with the project key
-			sm.config.MetalProjectID = project
-
-		}
-		packetClient, err = packngo.NewClient()
-		if err != nil {
-			return err
-		}
-
-		// We're using Equinix Metal with BGP, populate the Peer information from the API
-		if sm.config.EnableBGP {
-			log.Infoln("Looking up the BGP configuration from Equinix Metal")
-			err = equinixmetal.BGPLookup(packetClient, sm.config)
-			if err != nil {
-				return err
-			}
-		}
-	}
-
-	log.Info("Starting the BGP server to advertise VIP routes to BGP peers")
-	sm.bgpServer, err = bgp.NewBGPServer(&sm.config.BGPConfig, func(p *api.WatchEventResponse_PeerEvent) {
-		ipaddr := p.GetPeer().GetState().GetNeighborAddress()
-		port := uint64(179)
-		peerDescription := fmt.Sprintf("%s:%d", ipaddr, port)
-
-		for stateName, stateValue := range api.PeerState_SessionState_value {
-			metricValue := 0.0
-			if stateValue == int32(p.GetPeer().GetState().GetSessionState().Number()) {
-				metricValue = 1
-			}
-
-			sm.bgpSessionInfoGauge.With(prometheus.Labels{
-				"state": stateName,
-				"peer":  peerDescription,
-			}).Set(metricValue)
-		}
-	})
-	if err != nil {
-		return err
-	}
-
-	// use a Go context so we can tell the leaderelection code when we
-	// want to step down
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-
-	// Defer a function to check if the bgpServer has been created and if so attempt to close it
-	defer func() {
-		if sm.bgpServer != nil {
-			sm.bgpServer.Close()
-		}
-	}()
-
-	// Shutdown function that will wait on this signal, unless we call it ourselves
-	go func() {
-		<-sm.signalChan
-		log.Info("Received termination, signaling shutdown")
-		if sm.config.EnableControlPlane {
-			if cpCluster != nil {
-				cpCluster.Stop()
-			}
-		}
-		// Cancel the context, which will in turn cancel the leadership
-		cancel()
-	}()
-
-	if sm.config.EnableControlPlane {
-		cpCluster, err = cluster.InitCluster(sm.config, false)
-		if err != nil {
-			return err
-		}
-
-		clusterManager, err := initClusterManager(sm)
-		if err != nil {
-			return err
-		}
-
-		go func() {
-			if sm.config.EnableLeaderElection {
-				err = cpCluster.StartCluster(sm.config, clusterManager, sm.bgpServer)
-			} else {
-				err = cpCluster.StartVipService(sm.config, clusterManager, sm.bgpServer, packetClient)
-			}
-			if err != nil {
-				log.Errorf("Control Plane Error [%v]", err)
-				// Trigger the shutdown of this manager instance
-				sm.signalChan <- syscall.SIGINT
-			}
-		}()
-
-		// Check if we're also starting the services, if not we can sit and wait on the closing channel and return here
-		if !sm.config.EnableServices {
-			<-sm.signalChan
-			log.Infof("Shutting down Kube-Vip")
-
-			return nil
-		}
-	}
-
-	err = sm.servicesWatcher(ctx, sm.syncServices)
-	if err != nil {
-		return err
-	}
-
-	log.Infof("Shutting down Kube-Vip")
-
-	return nil
-}

pkg/manager/manager_dump.go

@@ -0,0 +1,290 @@
+package manager
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"time"
+
+	"github.com/kube-vip/kube-vip/pkg/nftables"
+	v1 "k8s.io/api/core/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+)
+
+// dumpConfiguration prints the current configuration to stdout when SIGUSR1 is received
+func (sm *Manager) dumpConfiguration(ctx context.Context) {
+	sm.mutex.Lock()
+	defer sm.mutex.Unlock()
+
+	fmt.Printf("\n")
+	fmt.Printf("================================================================================\n")
+	fmt.Printf("                   KUBE-VIP CONFIGURATION DUMP\n")
+	fmt.Printf("================================================================================\n")
+	fmt.Printf("Timestamp: %s\n", time.Now().Format(time.RFC3339))
+	fmt.Printf("Node Name: %s\n", sm.config.NodeName)
+	fmt.Printf("Process ID: %d\n", os.Getpid())
+	fmt.Printf("================================================================================\n")
+	fmt.Printf("\n")
+
+	sm.dumpConfigSection()
+	sm.dumpBGPSection(ctx)
+	sm.dumpARPSection()
+	sm.dumpServicesSection(ctx)
+	sm.dumpNetworkInterfacesSection()
+	sm.dumpLeaderElectionSection()
+	sm.dumpRuntimeSection()
+	sm.dumpNFTablesSection()
+	fmt.Printf("================================================================================\n")
+	fmt.Printf("                   END OF CONFIGURATION DUMP\n")
+	fmt.Printf("================================================================================\n")
+	fmt.Printf("\n")
+}
+
+func (sm *Manager) dumpConfigSection() {
+	fmt.Printf("--- BASIC CONFIGURATION ---\n")
+	fmt.Printf("VIP: %s\n", sm.config.Address)
+	fmt.Printf("VIP Subnet: %s\n", sm.config.VIPSubnet)
+	fmt.Printf("Port: %d\n", sm.config.Port)
+	fmt.Printf("Namespace: %s\n", sm.config.Namespace)
+	fmt.Printf("Instance Name: %s\n", sm.config.InstanceName)
+	fmt.Printf("Service Namespace: %s\n", sm.config.ServiceNamespace)
+	fmt.Printf("Interface: %s\n", sm.config.Interface)
+	fmt.Printf("Services Interface: %s\n", sm.config.ServicesInterface)
+	fmt.Printf("Single Node Mode: %t\n", sm.config.SingleNode)
+	fmt.Printf("Start As Leader: %t\n", sm.config.StartAsLeader)
+	fmt.Printf("\n")
+}
+
+func (sm *Manager) dumpBGPSection(ctx context.Context) {
+	fmt.Printf("--- BGP CONFIGURATION ---\n")
+	fmt.Printf("BGP Enabled: %t\n", sm.config.EnableBGP)
+	if sm.config.EnableBGP {
+		fmt.Printf("BGP AS: %d\n", sm.config.BGPConfig.AS)
+		fmt.Printf("BGP Router ID: %s\n", sm.config.BGPConfig.RouterID)
+		fmt.Printf("BGP Source IP: %s\n", sm.config.BGPConfig.SourceIP)
+		fmt.Printf("BGP Source Interface: %s\n", sm.config.BGPConfig.SourceIF)
+		fmt.Printf("BGP Hold Time: %d\n", sm.config.BGPConfig.HoldTime)
+		fmt.Printf("BGP Keepalive Interval: %d\n", sm.config.BGPConfig.KeepaliveInterval)
+		fmt.Printf("BGP Peers: %d\n", len(sm.config.BGPConfig.Peers))
+		for i, peer := range sm.config.BGPConfig.Peers {
+			fmt.Printf("  Peer %d: %s:%d (AS: %d, MultiHop: %t)\n",
+				i+1, peer.Address, peer.Port, peer.AS, peer.MultiHop)
+		}
+		fmt.Printf("\n--- ACTIVE BGP RIB STATE ---\n")
+		sm.dumpBGPRoutes(ctx)
+	}
+	fmt.Printf("\n")
+}
+
+func (sm *Manager) dumpARPSection() {
+	fmt.Printf("--- ARP/NDP CONFIGURATION ---\n")
+	fmt.Printf("ARP Enabled: %t\n", sm.config.EnableARP)
+	if sm.config.EnableARP {
+		fmt.Printf("ARP Broadcast Rate: %d\n", sm.config.ArpBroadcastRate)
+	}
+	fmt.Printf("Wireguard Enabled: %t\n", sm.config.EnableWireguard)
+	fmt.Printf("Routing Table Enabled: %t\n", sm.config.EnableRoutingTable)
+	if sm.config.EnableRoutingTable {
+		fmt.Printf("Routing Table ID: %d\n", sm.config.RoutingTableID)
+		fmt.Printf("Routing Protocol: %d\n", sm.config.RoutingProtocol)
+		fmt.Printf("Clean Routing Table: %t\n", sm.config.CleanRoutingTable)
+	}
+	fmt.Printf("\n")
+}
+
+func (sm *Manager) dumpServicesSection(ctx context.Context) {
+	fmt.Printf("--- SERVICES CONFIGURATION ---\n")
+	fmt.Printf("Services Enabled: %t\n", sm.config.EnableServices)
+	if sm.config.EnableServices {
+		fmt.Printf("Services Election: %t\n", sm.config.EnableServicesElection)
+		fmt.Printf("Load Balancer Class Only: %t\n", sm.config.LoadBalancerClassOnly)
+		fmt.Printf("Load Balancer Class Name: %s\n", sm.config.LoadBalancerClassName)
+		fmt.Printf("Disable Service Updates: %t\n", sm.config.DisableServiceUpdates)
+		fmt.Printf("Enable Endpoints: %t\n", sm.config.EnableEndpoints)
+		fmt.Printf("Service Security Enabled: %t\n", sm.config.EnableServiceSecurity)
+
+		if sm.svcProcessor != nil {
+			instances := sm.svcProcessor.ServiceInstances
+			fmt.Printf("Kube-vip Active Service Instances: %d\n", len(instances))
+			for i, inst := range instances {
+				if inst.ServiceSnapshot != nil {
+					svc := inst.ServiceSnapshot
+					vipConfigs := ""
+					for j, cfg := range svc.Status.LoadBalancer.Ingress {
+						if j > 0 {
+							vipConfigs += ", "
+						}
+						vipConfigs += cfg.IP
+					}
+					fmt.Printf("  Service %d: %s/%s (Type: %s, VIPs: %s)\n",
+						i+1, svc.Namespace, svc.Name, svc.Spec.Type, vipConfigs)
+				}
+			}
+		}
+		if sm.clientSet != nil {
+			fmt.Println()
+			// Kubernetes configuration
+			fmt.Println("--- KUBERNETES CONFIGURATION (SERVICES/ENDPOINTSLICES) ---")
+
+			fmt.Println("Service Configuration:")
+			svcList, err := sm.clientSet.CoreV1().Services(v1.NamespaceAll).List(ctx, metav1.ListOptions{})
+			if err != nil {
+				fmt.Println("Unable to retrieve all Services")
+			} else {
+				for x := range svcList.Items {
+
+					// Build all addresses
+					vipConfigs := ""
+					for j, cfg := range svcList.Items[x].Status.LoadBalancer.Ingress {
+						if j > 0 {
+							vipConfigs += ", "
+						}
+						vipConfigs += cfg.IP
+					}
+					fmt.Printf("Name=%s, UUID=%s, Addresses=%s\n", svcList.Items[x].Name, string(svcList.Items[x].UID), vipConfigs)
+				}
+				fmt.Println()
+			}
+
+			fmt.Println("EndpointSlice Configuration (note endpoint names have -XXXXX prefixed):")
+			epList, err := sm.clientSet.DiscoveryV1().EndpointSlices(v1.NamespaceAll).List(ctx, metav1.ListOptions{})
+			if err != nil {
+				fmt.Println("Unable to retrieve all EndpointSlices")
+			} else {
+				for x := range epList.Items {
+					// Build all addresses
+					fmt.Printf("  Endpoint Slice Name: %s\n", epList.Items[x].Name)
+					for _, ep := range epList.Items[x].Endpoints {
+						endpoints := ""
+						for i, addresses := range ep.Addresses {
+							if i > 0 {
+								endpoints += ", "
+							}
+							endpoints += addresses
+						}
+						nodeName := "Unknown"
+						targetPod := "Unknown"
+						if ep.NodeName != nil {
+							nodeName = *ep.NodeName
+						}
+						if ep.TargetRef != nil {
+							targetPod = ep.TargetRef.Name
+						}
+						fmt.Printf("\tNode: %s, Target Pod:%s, Addresses: %s\n", nodeName, targetPod, endpoints)
+					}
+				}
+			}
+		}
+	}
+	fmt.Printf("\n")
+}
+
+func (sm *Manager) dumpNetworkInterfacesSection() {
+	fmt.Printf("--- NETWORK INTERFACES ---\n")
+	fmt.Printf("Network Interface Manager: %t\n", sm.intfMgr != nil)
+	fmt.Printf("ARP Manager: %t\n", sm.arpMgr != nil)
+	fmt.Printf("\n")
+}
+
+func (sm *Manager) dumpLeaderElectionSection() {
+	fmt.Printf("--- LEADER ELECTION CONFIGURATION ---\n")
+	fmt.Printf("Control Plane Enabled: %t\n", sm.config.EnableControlPlane)
+	if sm.config.EnableControlPlane {
+		fmt.Printf("Detect Control Plane: %t\n", sm.config.DetectControlPlane)
+	}
+	fmt.Printf("Leader Election Type: %s\n", sm.config.LeaderElectionType)
+	fmt.Printf("Leader Election Enabled: %t\n", sm.config.EnableLeaderElection)
+	if sm.config.EnableLeaderElection {
+		fmt.Printf("Lease Name: %s\n", sm.config.LeaseName)
+		fmt.Printf("Lease Duration: %d seconds\n", sm.config.LeaseDuration)
+		fmt.Printf("Renew Deadline: %d seconds\n", sm.config.RenewDeadline)
+		fmt.Printf("Retry Period: %d seconds\n", sm.config.RetryPeriod)
+	}
+	fmt.Printf("Services Lease Name: %s\n", sm.config.ServicesLeaseName)
+	fmt.Printf("Node Labeling Enabled: %t\n", sm.config.EnableNodeLabeling)
+	fmt.Printf("\n")
+}
+
+func (sm *Manager) dumpRuntimeSection() {
+	fmt.Printf("--- RUNTIME STATISTICS ---\n")
+	fmt.Printf("Load Balancer Enabled: %t\n", sm.config.EnableLoadBalancer)
+	if sm.config.EnableLoadBalancer {
+		fmt.Printf("Load Balancer Port: %d\n", sm.config.LoadBalancerPort)
+		fmt.Printf("Load Balancer Forwarding Method: %s\n", sm.config.LoadBalancerForwardingMethod)
+		fmt.Printf("Load Balancers Configured: %d\n", len(sm.config.LoadBalancers))
+	}
+	fmt.Printf("Prometheus HTTP Server: %s\n", sm.config.PrometheusHTTPServer)
+	fmt.Printf("Health Check Port: %d\n", sm.config.HealthCheckPort)
+	fmt.Printf("UPNP Enabled: %t\n", sm.config.EnableUPNP)
+	fmt.Printf("Egress Clean Enabled: %t\n", sm.config.EgressClean)
+	if sm.config.EgressClean {
+		fmt.Printf("Egress with nftables: %t\n", sm.config.EgressWithNftables)
+		fmt.Printf("Egress nftables table name: %s\n", nftables.EgressTableBaseNameForInstance(sm.config.InstanceName))
+		fmt.Printf("Egress Pod CIDR: %s\n", sm.config.EgressPodCidr)
+		fmt.Printf("Egress Service CIDR: %s\n", sm.config.EgressServiceCidr)
+	}
+	fmt.Printf("\n")
+}
+
+func (sm *Manager) dumpNFTablesSection() {
+	fmt.Printf("--- NFTABLES CONFIGURATION ---\n")
+	chains, err := nftables.ListChains()
+	if err != nil {
+		fmt.Printf("Unable to retrieve NFTables chains, error=%s", err)
+	}
+	for x := range chains {
+		fmt.Printf("Chain: %s\n", chains[x])
+	}
+	fmt.Println()
+}
+
+func (sm *Manager) dumpBGPRoutes(ctx context.Context) {
+	if sm.bgpServer == nil {
+		fmt.Printf("  BGP Server instance is inactive or uninitialized\n")
+		return
+	}
+
+	// Create a short-lived execution window so a stuck BGP loop won't hang the entire SIGUSR1 routine
+	queryCtx, cancel := context.WithTimeout(ctx, 3*time.Second)
+	defer cancel()
+
+	for _, isIPv6 := range []bool{false, true} {
+		label := "IPv4"
+		if isIPv6 {
+			label = "IPv6"
+		}
+
+		routes, err := sm.bgpServer.ListAdvertisedRoutes(queryCtx, isIPv6)
+		if err != nil {
+			fmt.Printf("  Error fetching %s routes: %v\n", label, err)
+			continue
+		}
+
+		if len(routes) == 0 {
+			fmt.Printf("  No %s routes found in global RIB\n", label)
+			continue
+		}
+
+		fmt.Printf("  %-18s | %-15s | %s\n", "Prefix", "Next Hop", "Discovered/Updated")
+		fmt.Printf("  ------------------------------------------------------------\n")
+
+		for _, dest := range routes {
+			for _, path := range dest.Paths {
+				nextHop := "N/A"
+				if path.NeighborIp != "" {
+					nextHop = path.NeighborIp
+				}
+
+				var timeStr string
+				if path.Age != nil {
+					timeStr = path.Age.AsTime().Format("15:04:05")
+				} else {
+					timeStr = "Unknown"
+				}
+
+				fmt.Printf("  %-18s | %-15s | %s\n", dest.Prefix, nextHop, timeStr)
+			}
+		}
+		fmt.Println()
+	}
+}

pkg/manager/manager_dump_test.go

@@ -0,0 +1,242 @@
+package manager
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"os"
+	"testing"
+
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestDumpConfiguration(t *testing.T) {
+	config := &kubevip.Config{
+		Address:            "192.168.1.100",
+		Interface:          "eth0",
+		Port:               6443,
+		EnableARP:          true,
+		EnableBGP:          false,
+		EnableControlPlane: true,
+		EnableServices:     true,
+		LeaderElectionType: "kubernetes",
+		Namespace:          "kube-system",
+		NodeName:           "test-node",
+		KubernetesLeaderElection: kubevip.KubernetesLeaderElection{
+			EnableLeaderElection: true,
+			LeaseName:            "test-lease",
+		},
+	}
+
+	mgr := &Manager{
+		config: config,
+	}
+
+	old := os.Stdout
+	r, w, _ := os.Pipe()
+	os.Stdout = w
+
+	mgr.dumpConfiguration(context.TODO())
+
+	w.Close()
+	os.Stdout = old
+
+	var buf bytes.Buffer
+	_, err := io.Copy(&buf, r)
+	assert.NoError(t, err, "io.Copy should not return error")
+	output := buf.String()
+
+	assert.Contains(t, output, "KUBE-VIP CONFIGURATION DUMP", "should contain header")
+	assert.Contains(t, output, "Node Name: test-node", "should contain node name")
+	assert.Contains(t, output, "VIP: 192.168.1.100", "should contain VIP address")
+	assert.Contains(t, output, "Interface: eth0", "should contain interface")
+	assert.Contains(t, output, "Port: 6443", "should contain port")
+	assert.Contains(t, output, "ARP Enabled: true", "should contain ARP status")
+	assert.Contains(t, output, "BGP Enabled: false", "should contain BGP status")
+}
+
+func TestDumpConfigSection(t *testing.T) {
+	config := &kubevip.Config{
+		Address:            "192.168.1.100",
+		Interface:          "eth0",
+		Port:               6443,
+		VIPSubnet:          "/24",
+		EnableARP:          true,
+		EnableBGP:          false,
+		EnableControlPlane: true,
+		EnableServices:     true,
+		LeaderElectionType: "kubernetes",
+		Namespace:          "kube-system",
+		KubernetesLeaderElection: kubevip.KubernetesLeaderElection{
+			EnableLeaderElection: true,
+			LeaseName:            "test-lease",
+		},
+	}
+
+	mgr := &Manager{config: config}
+
+	old := os.Stdout
+	r, w, _ := os.Pipe()
+	os.Stdout = w
+
+	mgr.dumpConfigSection()
+
+	w.Close()
+	os.Stdout = old
+
+	var buf bytes.Buffer
+	_, err := io.Copy(&buf, r)
+	assert.NoError(t, err, "io.Copy should not return error")
+	output := buf.String()
+
+	assert.Contains(t, output, "--- BASIC CONFIGURATION ---")
+	assert.Contains(t, output, "VIP: 192.168.1.100")
+	assert.Contains(t, output, "VIP Subnet: /24")
+	assert.Contains(t, output, "Interface: eth0")
+	assert.Contains(t, output, "Port: 6443")
+	assert.Contains(t, output, "Namespace: kube-system")
+	assert.Contains(t, output, "Single Node Mode: false")
+	assert.Contains(t, output, "Start As Leader: false")
+}
+
+func TestDumpBGPSection(t *testing.T) {
+	t.Run("BGP disabled", func(t *testing.T) {
+		config := &kubevip.Config{
+			EnableBGP: false,
+		}
+		mgr := &Manager{config: config}
+
+		old := os.Stdout
+		r, w, _ := os.Pipe()
+		os.Stdout = w
+
+		mgr.dumpBGPSection(t.Context())
+
+		w.Close()
+		os.Stdout = old
+
+		var buf bytes.Buffer
+		_, err := io.Copy(&buf, r)
+		assert.NoError(t, err, "io.Copy should not return error")
+		output := buf.String()
+
+		assert.Contains(t, output, "BGP Enabled: false")
+	})
+
+	t.Run("BGP enabled", func(t *testing.T) {
+		config := &kubevip.Config{
+			EnableBGP: true,
+			BGPConfig: kubevip.BGPConfig{
+				RouterID: "192.168.1.1",
+				AS:       65000,
+				Peers: []kubevip.BGPPeer{
+					{Address: "192.168.1.2", AS: 65001},
+					{Address: "192.168.1.3", AS: 65002},
+				},
+			},
+		}
+		mgr := &Manager{config: config}
+
+		old := os.Stdout
+		r, w, _ := os.Pipe()
+		os.Stdout = w
+
+		mgr.dumpBGPSection(t.Context())
+
+		w.Close()
+		os.Stdout = old
+
+		var buf bytes.Buffer
+		_, err := io.Copy(&buf, r)
+		assert.NoError(t, err, "io.Copy should not return error")
+		output := buf.String()
+
+		assert.Contains(t, output, "BGP Enabled: true")
+		assert.Contains(t, output, "BGP Router ID: 192.168.1.1")
+		assert.Contains(t, output, "BGP AS: 65000")
+		assert.Contains(t, output, "BGP Peers: 2")
+
+		assert.Contains(t, output, "--- ACTIVE BGP RIB STATE ---")
+		assert.Contains(t, output, "BGP Server instance is inactive or uninitialized")
+	})
+}
+
+func TestDumpARPSection(t *testing.T) {
+	t.Run("ARP disabled", func(t *testing.T) {
+		config := &kubevip.Config{
+			EnableARP: false,
+		}
+		mgr := &Manager{config: config}
+
+		old := os.Stdout
+		r, w, _ := os.Pipe()
+		os.Stdout = w
+
+		mgr.dumpARPSection()
+
+		w.Close()
+		os.Stdout = old
+
+		var buf bytes.Buffer
+		_, err := io.Copy(&buf, r)
+		assert.NoError(t, err, "io.Copy should not return error")
+		output := buf.String()
+
+		assert.Contains(t, output, "ARP Enabled: false")
+	})
+
+	t.Run("ARP enabled", func(t *testing.T) {
+		config := &kubevip.Config{
+			EnableARP:        true,
+			ArpBroadcastRate: 5,
+		}
+		mgr := &Manager{config: config}
+
+		old := os.Stdout
+		r, w, _ := os.Pipe()
+		os.Stdout = w
+
+		mgr.dumpARPSection()
+
+		w.Close()
+		os.Stdout = old
+
+		var buf bytes.Buffer
+		_, err := io.Copy(&buf, r)
+		assert.NoError(t, err, "io.Copy should not return error")
+		output := buf.String()
+
+		assert.Contains(t, output, "ARP Enabled: true")
+		assert.Contains(t, output, "ARP Broadcast Rate: 5")
+	})
+}
+
+func TestDumpRuntimeSection(t *testing.T) {
+	config := &kubevip.Config{
+		EnableLoadBalancer:   false,
+		PrometheusHTTPServer: "",
+		HealthCheckPort:      0,
+		EnableUPNP:           false,
+		EgressClean:          false,
+	}
+	mgr := &Manager{config: config}
+
+	old := os.Stdout
+	r, w, _ := os.Pipe()
+	os.Stdout = w
+
+	mgr.dumpRuntimeSection()
+
+	w.Close()
+	os.Stdout = old
+
+	var buf bytes.Buffer
+	_, err := io.Copy(&buf, r)
+	assert.NoError(t, err, "io.Copy should not return error")
+	output := buf.String()
+
+	assert.Contains(t, output, "--- RUNTIME STATISTICS ---", "should contain runtime section header")
+	assert.Contains(t, output, "Load Balancer Enabled: false", "should contain load balancer status")
+	assert.Contains(t, output, "UPNP Enabled: false", "should contain UPNP status")
+}

pkg/manager/manager_table.go

@@ -1,214 +0,0 @@
-package manager
-
-import (
-	"context"
-	"fmt"
-	"os"
-	"strconv"
-	"syscall"
-	"time"
-
-	"github.com/kube-vip/kube-vip/pkg/cluster"
-	"github.com/kube-vip/kube-vip/pkg/iptables"
-	"github.com/kube-vip/kube-vip/pkg/vip"
-	log "github.com/sirupsen/logrus"
-	"github.com/vishvananda/netlink"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/client-go/tools/leaderelection"
-	"k8s.io/client-go/tools/leaderelection/resourcelock"
-)
-
-// Start will begin the Manager, which will start services and watch the configmap
-func (sm *Manager) startTableMode(id string) error {
-	var cpCluster *cluster.Cluster
-	var err error
-
-	// use a Go context so we can tell the leaderelection code when we
-	// want to step down
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	log.Infof("all routing table entries will exist in table [%d] with protocol [%d]", sm.config.RoutingTableID, sm.config.RoutingProtocol)
-
-	if sm.config.CleanRoutingTable {
-		go func() {
-			// we assume that after 10s all services should be configured so we can delete redundant routes
-			time.Sleep(time.Second * 10)
-			if err := sm.cleanRoutes(); err != nil {
-				log.Errorf("error checking for old routes: %v", err)
-			}
-		}()
-	}
-
-	egressCleanEnv := os.Getenv("EGRESS_CLEAN")
-	if egressCleanEnv != "" {
-		egressClean, err := strconv.ParseBool(egressCleanEnv)
-		if err != nil {
-			log.Warnf("failed to parse EGRESS_CLEAN env value [%s]. Egress cleaning will not be performed: %s", egressCleanEnv, err.Error())
-		}
-		if egressClean {
-			vip.ClearIPTables(sm.config.EgressWithNftables, sm.config.ServiceNamespace, iptables.ProtocolIPv4)
-			vip.ClearIPTables(sm.config.EgressWithNftables, sm.config.ServiceNamespace, iptables.ProtocolIPv6)
-			log.Debug("IPTables rules cleaned on startup")
-		}
-	}
-
-	// Shutdown function that will wait on this signal, unless we call it ourselves
-	go func() {
-		<-sm.signalChan
-		log.Info("Received kube-vip termination, signaling shutdown")
-		if sm.config.EnableControlPlane {
-			cpCluster.Stop()
-		}
-
-		// Cancel the context, which will in turn cancel the leadership
-		cancel()
-	}()
-
-	if sm.config.EnableControlPlane {
-		cpCluster, err = cluster.InitCluster(sm.config, false)
-		if err != nil {
-			return fmt.Errorf("cluster initialization error: %w", err)
-		}
-
-		clusterManager, err := initClusterManager(sm)
-		if err != nil {
-			return fmt.Errorf("cluster manager initialization error: %w", err)
-		}
-
-		if err := cpCluster.StartVipService(sm.config, clusterManager, nil, nil); err != nil {
-			log.Errorf("Control Plane Error [%v]", err)
-			// Trigger the shutdown of this manager instance
-			sm.signalChan <- syscall.SIGINT
-		}
-	} else {
-		ns, err := returnNameSpace()
-		if err != nil {
-			log.Warnf("unable to auto-detect namespace, dropping to [%s]", sm.config.Namespace)
-			ns = sm.config.Namespace
-		}
-
-		// Start a services watcher (all kube-vip pods will watch services), upon a new service
-		// a lock based upon that service is created that they will all leaderElection on
-		if sm.config.EnableServicesElection {
-			log.Infof("beginning watching services, leaderelection will happen for every service")
-			err = sm.startServicesWatchForLeaderElection(ctx)
-			if err != nil {
-				return err
-			}
-		} else if sm.config.EnableLeaderElection {
-
-			log.Infof("beginning services leadership, namespace [%s], lock name [%s], id [%s]", ns, plunderLock, id)
-			// we use the Lease lock type since edits to Leases are less common
-			// and fewer objects in the cluster watch "all Leases".
-			lock := &resourcelock.LeaseLock{
-				LeaseMeta: metav1.ObjectMeta{
-					Name:      plunderLock,
-					Namespace: ns,
-				},
-				Client: sm.clientSet.CoordinationV1(),
-				LockConfig: resourcelock.ResourceLockConfig{
-					Identity: id,
-				},
-			}
-			// start the leader election code loop
-			leaderelection.RunOrDie(ctx, leaderelection.LeaderElectionConfig{
-				Lock: lock,
-				// IMPORTANT: you MUST ensure that any code you have that
-				// is protected by the lease must terminate **before**
-				// you call cancel. Otherwise, you could have a background
-				// loop still running and another process could
-				// get elected before your background loop finished, violating
-				// the stated goal of the lease.
-				ReleaseOnCancel: true,
-				LeaseDuration:   time.Duration(sm.config.LeaseDuration) * time.Second,
-				RenewDeadline:   time.Duration(sm.config.RenewDeadline) * time.Second,
-				RetryPeriod:     time.Duration(sm.config.RetryPeriod) * time.Second,
-				Callbacks: leaderelection.LeaderCallbacks{
-					OnStartedLeading: func(ctx context.Context) {
-						err = sm.servicesWatcher(ctx, sm.syncServices)
-						if err != nil {
-							log.Fatal(err)
-						}
-					},
-					OnStoppedLeading: func() {
-						// we can do cleanup here
-						log.Infof("leader lost: %s", id)
-						for _, instance := range sm.serviceInstances {
-							for _, cluster := range instance.clusters {
-								cluster.Stop()
-							}
-						}
-
-						log.Fatal("lost leadership, restarting kube-vip")
-					},
-					OnNewLeader: func(identity string) {
-						// we're notified when new leader elected
-						if identity == id {
-							// I just got the lock
-							return
-						}
-						log.Infof("new leader elected: %s", identity)
-					},
-				},
-			})
-		} else {
-			log.Infof("beginning watching services without leader election")
-			err = sm.servicesWatcher(ctx, sm.syncServices)
-			if err != nil {
-				log.Errorf("Cannot watch services, %v", err)
-			}
-		}
-	}
-
-	return nil
-}
-
-func (sm *Manager) cleanRoutes() error {
-	routes, err := vip.ListRoutes(sm.config.RoutingTableID, sm.config.RoutingProtocol)
-	if err != nil {
-		return fmt.Errorf("error getting routes: %w", err)
-	}
-
-	for i := range routes {
-		found := false
-		if sm.config.EnableControlPlane {
-			found = (routes[i].Dst.IP.String() == sm.config.Address)
-		} else {
-			for _, instance := range sm.serviceInstances {
-				for _, cluster := range instance.clusters {
-					for n := range cluster.Network {
-						r := cluster.Network[n].PrepareRoute()
-						if r.Dst.String() == routes[i].Dst.String() {
-							found = true
-						}
-					}
-				}
-			}
-		}
-
-		if !found {
-			err = netlink.RouteDel(&(routes[i]))
-			if err != nil {
-				log.Errorf("[route] error deleting route: %v", routes[i])
-			}
-			log.Debugf("[route] deleted route: %v", routes[i])
-		}
-
-	}
-	return nil
-}
-
-func (sm *Manager) countRouteReferences(route *netlink.Route) int {
-	cnt := 0
-	for _, instance := range sm.serviceInstances {
-		for _, cluster := range instance.clusters {
-			for n := range cluster.Network {
-				r := cluster.Network[n].PrepareRoute()
-				if r.Dst.String() == route.Dst.String() {
-					cnt++
-				}
-			}
-		}
-	}
-	return cnt
-}

pkg/manager/manager_test.go

@@ -0,0 +1,38 @@
+package manager
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestNormalizeNodeName(t *testing.T) {
+	tests := []struct {
+		name     string
+		hostname string
+		expected string
+	}{
+		{
+			name:     "All lowercase hostname remains unchanged",
+			hostname: "worker-node-1",
+			expected: "worker-node-1",
+		},
+		{
+			name:     "Mixed case hostname is lowercased",
+			hostname: "Worker-Node-1",
+			expected: "worker-node-1",
+		},
+		{
+			name:     "All uppercase hostname is lowercased",
+			hostname: "MASTER-NODE",
+			expected: "master-node",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := normalizeNodeName(tt.hostname)
+			assert.Equal(t, tt.expected, result, "The normalized node name did not match the expected RFC1123 compliant name")
+		})
+	}
+}

pkg/manager/manager_wireguard.go

@@ -1,121 +0,0 @@
-package manager
-
-import (
-	"context"
-	"time"
-
-	"github.com/kube-vip/kube-vip/pkg/wireguard"
-	log "github.com/sirupsen/logrus"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/client-go/tools/leaderelection"
-	"k8s.io/client-go/tools/leaderelection/resourcelock"
-)
-
-// Start will begin the Manager, which will start services and watch the configmap
-func (sm *Manager) startWireguard(id string) error {
-	var ns string
-	var err error
-
-	// use a Go context so we can tell the leaderelection code when we
-	// want to step down
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	log.Infoln("reading wireguard peer configuration from Kubernetes secret")
-	s, err := sm.clientSet.CoreV1().Secrets(sm.config.Namespace).Get(ctx, "wireguard", metav1.GetOptions{})
-	if err != nil {
-		return err
-	}
-	// parse all the details needed for Wireguard
-	peerPublicKey := s.Data["peerPublicKey"]
-	peerEndpoint := s.Data["peerEndpoint"]
-	privateKey := s.Data["privateKey"]
-
-	// Configure the interface to join the Wireguard VPN
-	err = wireguard.ConfigureInterface(string(privateKey), string(peerPublicKey), string(peerEndpoint))
-	if err != nil {
-		return err
-	}
-
-	// Shutdown function that will wait on this signal, unless we call it ourselves
-	go func() {
-		<-sm.signalChan
-		log.Info("Received termination, signaling shutdown")
-
-		// Cancel the context, which will in turn cancel the leadership
-		cancel()
-	}()
-
-	ns, err = returnNameSpace()
-	if err != nil {
-		log.Warnf("unable to auto-detect namespace, dropping to [%s]", sm.config.Namespace)
-		ns = sm.config.Namespace
-	}
-
-	// Start a services watcher (all kube-vip pods will watch services), upon a new service
-	// a lock based upon that service is created that they will all leaderElection on
-	if sm.config.EnableServicesElection {
-		log.Infof("beginning watching services, leaderelection will happen for every service")
-		err = sm.startServicesWatchForLeaderElection(ctx)
-		if err != nil {
-			return err
-		}
-	} else {
-
-		log.Infof("beginning services leadership, namespace [%s], lock name [%s], id [%s]", ns, plunderLock, id)
-		// we use the Lease lock type since edits to Leases are less common
-		// and fewer objects in the cluster watch "all Leases".
-		lock := &resourcelock.LeaseLock{
-			LeaseMeta: metav1.ObjectMeta{
-				Name:      plunderLock,
-				Namespace: ns,
-			},
-			Client: sm.clientSet.CoordinationV1(),
-			LockConfig: resourcelock.ResourceLockConfig{
-				Identity: id,
-			},
-		}
-
-		// start the leader election code loop
-		leaderelection.RunOrDie(ctx, leaderelection.LeaderElectionConfig{
-			Lock: lock,
-			// IMPORTANT: you MUST ensure that any code you have that
-			// is protected by the lease must terminate **before**
-			// you call cancel. Otherwise, you could have a background
-			// loop still running and another process could
-			// get elected before your background loop finished, violating
-			// the stated goal of the lease.
-			ReleaseOnCancel: true,
-			LeaseDuration:   time.Duration(sm.config.LeaseDuration) * time.Second,
-			RenewDeadline:   time.Duration(sm.config.RenewDeadline) * time.Second,
-			RetryPeriod:     time.Duration(sm.config.RetryPeriod) * time.Second,
-			Callbacks: leaderelection.LeaderCallbacks{
-				OnStartedLeading: func(ctx context.Context) {
-					err = sm.servicesWatcher(ctx, sm.syncServices)
-					if err != nil {
-						log.Fatal(err)
-					}
-				},
-				OnStoppedLeading: func() {
-					// we can do cleanup here
-					log.Infof("leader lost: %s", id)
-					for _, instance := range sm.serviceInstances {
-						for _, cluster := range instance.clusters {
-							cluster.Stop()
-						}
-					}
-
-					log.Fatal("lost leadership, restarting kube-vip")
-				},
-				OnNewLeader: func(identity string) {
-					// we're notified when new leader elected
-					if identity == id {
-						// I just got the lock
-						return
-					}
-					log.Infof("new leader elected: %s", identity)
-				},
-			},
-		})
-	}
-	return nil
-}

pkg/manager/node_labeling.go

@@ -1,73 +0,0 @@
-package manager
-
-import (
-	"context"
-	"encoding/json"
-	"fmt"
-
-	log "github.com/sirupsen/logrus"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/apimachinery/pkg/types"
-	"k8s.io/client-go/kubernetes"
-)
-
-const (
-	nodeLabelIndex    = "kube-vip.io/has-ip"
-	nodeLabelJSONPath = `kube-vip.io~1has-ip`
-)
-
-type patchStringLabel struct {
-	Op    string `json:"op"`
-	Path  string `json:"path"`
-	Value string `json:"value"`
-}
-
-// applyNodeLabel add/remove node label `kube-vip.io/has-ip=<VIP-Address>` to/from
-// the node where the virtual IP was added to/removed from.
-func applyNodeLabel(clientSet *kubernetes.Clientset, address, id, identity string) {
-	ctx := context.Background()
-	node, err := clientSet.CoreV1().Nodes().Get(ctx, id, metav1.GetOptions{})
-	if err != nil {
-		log.Errorf("can't query node %s labels. error: %v", id, err)
-		return
-	}
-
-	log.Debugf("node %s labels: %+v", id, node.Labels)
-
-	value, ok := node.Labels[nodeLabelIndex]
-	path := fmt.Sprintf("/metadata/labels/%s", nodeLabelJSONPath)
-	if (!ok || value != address) && id == identity {
-		log.Debugf("setting node label `has-ip=%s` on %s", address, id)
-		// Append label
-		applyPatchLabels(ctx, clientSet, id, "add", path, address)
-	} else if ok && value == address {
-		log.Debugf("removing node label `has-ip=%s` on %s", address, id)
-		// Remove label
-		applyPatchLabels(ctx, clientSet, id, "remove", path, address)
-	} else {
-		log.Debugf("no node label change needed")
-	}
-}
-
-// applyPatchLabels add/remove node labels
-func applyPatchLabels(ctx context.Context, clientSet *kubernetes.Clientset,
-	name, operation, path, value string) {
-	patchLabels := []patchStringLabel{{
-		Op:    operation,
-		Path:  path,
-		Value: value,
-	}}
-	patchData, err := json.Marshal(patchLabels)
-	if err != nil {
-		log.Errorf("node patch marshaling failed. error: %v", err)
-		return
-	}
-	// patch node
-	node, err := clientSet.CoreV1().Nodes().Patch(ctx,
-		name, types.JSONPatchType, patchData, metav1.PatchOptions{})
-	if err != nil {
-		log.Errorf("can't patch node %s. error: %v", name, err)
-		return
-	}
-	log.Debugf("updated node %s labels: %+v", name, node.Labels)
-}

pkg/manager/prom.go

@@ -1,8 +0,0 @@
-package manager
-
-import "github.com/prometheus/client_golang/prometheus"
-
-// PrometheusCollector defines a service watch event counter.
-func (sm *Manager) PrometheusCollector() []prometheus.Collector {
-	return []prometheus.Collector{sm.countServiceWatchEvent, sm.bgpSessionInfoGauge}
-}

pkg/manager/service_egress.go

@@ -1,338 +0,0 @@
-package manager
-
-import (
-	"bufio"
-	"context"
-	"fmt"
-	"net"
-	"os"
-	"strings"
-
-	"github.com/kube-vip/kube-vip/pkg/iptables"
-	"github.com/kube-vip/kube-vip/pkg/vip"
-	log "github.com/sirupsen/logrus"
-	v1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-)
-
-// DEBUG
-const (
-	defaultPodCIDR     = "10.0.0.0/16"
-	defaultServiceCIDR = "10.96.0.0/12"
-)
-
-func (sm *Manager) iptablesCheck() error {
-	file, err := os.Open("/proc/modules")
-	if err != nil {
-		return err
-	}
-	defer file.Close()
-	scanner := bufio.NewScanner(file)
-	scanner.Split(bufio.ScanLines)
-	var nat, filter, mangle bool
-	for scanner.Scan() {
-		line := strings.Fields(scanner.Text())
-		switch line[0] {
-		case "iptable_filter":
-			filter = true
-		case "iptable_nat":
-			nat = true
-		case "iptable_mangle":
-			mangle = true
-		}
-	}
-
-	if !filter || !nat || !mangle {
-		return fmt.Errorf("missing iptables modules -> nat [%t] -> filter [%t] mangle -> [%t]", nat, filter, mangle)
-	}
-	return nil
-}
-
-func getSameFamilyCidr(sourceCidrs, ip string) string { //Todo: not sure how this ever worked
-	cidrs := strings.Split(sourceCidrs, ",")
-	isV6 := vip.IsIPv6(ip)
-	matchingFamily := []string{}
-	for _, cidr := range cidrs {
-		// Is the ip an IPv6 address
-		if isV6 {
-			if vip.IsIPv6CIDR(cidr) {
-				matchingFamily = append(matchingFamily, cidr)
-				selectedCIDR, err := checkCIDR(ip, cidr)
-				if err != nil {
-					log.Warnf("IPv6 CIDR check failed: %s", err.Error())
-					continue
-				}
-				if selectedCIDR != "" {
-					return selectedCIDR
-				}
-			}
-		} else {
-			if vip.IsIPv4CIDR(cidr) {
-				matchingFamily = append(matchingFamily, cidr)
-				selectedCidr, err := checkCIDR(ip, cidr)
-				if err != nil {
-					log.Warnf("IPv4 CIDR check failed: %s", err.Error())
-					continue
-				}
-				if selectedCidr != "" {
-					return selectedCidr
-				}
-			}
-		}
-	}
-
-	if len(matchingFamily) > 0 {
-		// return first CIDR that has at least the same IP family as processed IP address
-		// (should be better than just returning first CIDR on the list, I think)
-		return matchingFamily[0]
-	}
-
-	// return to the default behaviour of setting the CIDR to the first one (or only one)
-	return cidrs[0]
-}
-
-func checkCIDR(ip, cidr string) (string, error) {
-	_, ipnetA, err := net.ParseCIDR(cidr)
-	if err != nil {
-		return "", fmt.Errorf("failed to parse CIDR [%s]: %w", cidr, err)
-	}
-	parsedIP := net.ParseIP(ip)
-	if parsedIP == nil {
-		return "", fmt.Errorf("failed to parse IP [%s]", ip)
-	}
-	if ipnetA.Contains(parsedIP) {
-		return cidr, nil
-	}
-
-	return "", nil
-}
-
-func (sm *Manager) configureEgress(vipIP, podIP, destinationPorts, namespace string) error {
-	var podCidr, serviceCidr string
-	var autoServiceCIDR, autoPodCIDR string
-	var discoverErr error
-
-	if sm.config.EgressPodCidr == "" || sm.config.EgressServiceCidr == "" {
-		autoServiceCIDR, autoPodCIDR, discoverErr = sm.AutoDiscoverCIDRs()
-	}
-
-	if discoverErr != nil {
-		log.Warn(discoverErr)
-	}
-
-	if sm.config.EgressPodCidr != "" {
-		podCidr = getSameFamilyCidr(sm.config.EgressPodCidr, podIP)
-	} else {
-		if discoverErr == nil {
-			podCidr = getSameFamilyCidr(autoPodCIDR, podIP)
-		}
-	}
-
-	if podCidr == "" {
-		// There's no default IPv6 pod CIDR, therefore we silently back off if CIDR s not specified.
-		if !vip.IsIPv4(podIP) {
-			return nil
-		}
-		podCidr = defaultPodCIDR
-	}
-
-	if sm.config.EgressServiceCidr != "" {
-		serviceCidr = getSameFamilyCidr(sm.config.EgressServiceCidr, vipIP)
-	} else {
-		if discoverErr == nil {
-			serviceCidr = getSameFamilyCidr(autoServiceCIDR, vipIP)
-		}
-	}
-
-	if serviceCidr == "" {
-		// There's no default IPv6 service CIDR, therefore we silently back off if CIDR s not specified.
-		if !vip.IsIPv4(vipIP) {
-			return nil
-		}
-		serviceCidr = defaultServiceCIDR
-	}
-
-	log.Infof("[Egress] pod CIDR [%s], service CIDR [%s] for vip [%s] / pod [%s]", podCidr, serviceCidr, vipIP, podIP)
-
-	// checking if all addresses are of the same IP family
-	if vip.IsIPv4(podIP) != vip.IsIPv4CIDR(podCidr) {
-		log.Errorf("[Egress] pod's IP [%s] and Pod CIDR [%s] family is not matching. Backing off...", podIP, podCidr)
-		return nil
-	}
-
-	if vip.IsIPv4(vipIP) != vip.IsIPv4CIDR(serviceCidr) {
-		log.Errorf("[Egress] VIP's IP [%s] and Service CIDR [%s] family is not matching. Backing off...", podIP, podCidr)
-		return nil
-	}
-
-	if vip.IsIPv4(vipIP) != vip.IsIPv4(podIP) {
-		log.Errorf("[Egress] VIP's IP [%s] and Pod's IP [%s] family is not matching. Backing off...", podIP, podCidr)
-		return nil
-	}
-
-	protocol := iptables.ProtocolIPv4
-	if vip.IsIPv6(vipIP) {
-		protocol = iptables.ProtocolIPv6
-	}
-
-	i, err := vip.CreateIptablesClient(sm.config.EgressWithNftables, namespace, protocol)
-	if err != nil {
-		return fmt.Errorf("error Creating iptables client [%s]", err)
-	}
-
-	// Check if the kube-vip mangle chain exists, if not create it
-	exists, err := i.CheckMangleChain(vip.MangleChainName)
-	if err != nil {
-		return fmt.Errorf("error checking for existence of mangle chain [%s], error [%s]", vip.MangleChainName, err)
-	}
-
-	if !exists {
-		err = i.CreateMangleChain(vip.MangleChainName)
-		if err != nil {
-			return fmt.Errorf("error creating mangle chain [%s], error [%s]", vip.MangleChainName, err)
-		}
-	}
-	err = i.AppendReturnRulesForDestinationSubnet(vip.MangleChainName, podCidr)
-	if err != nil {
-		return fmt.Errorf("error adding rules to mangle chain [%s], error [%s]", vip.MangleChainName, err)
-	}
-
-	err = i.AppendReturnRulesForDestinationSubnet(vip.MangleChainName, serviceCidr)
-	if err != nil {
-		return fmt.Errorf("error adding rules to mangle chain [%s], error [%s]", vip.MangleChainName, err)
-	}
-
-	mask := "/32"
-	if !vip.IsIPv4(podIP) {
-		mask = "/128"
-	}
-
-	err = i.AppendReturnRulesForMarking(vip.MangleChainName, podIP+mask)
-	if err != nil {
-		return fmt.Errorf("error adding marking rules to mangle chain [%s], error [%s]", vip.MangleChainName, err)
-	}
-
-	err = i.InsertMangeTableIntoPrerouting(vip.MangleChainName)
-	if err != nil {
-		return fmt.Errorf("error adding prerouting mangle chain [%s], error [%s]", vip.MangleChainName, err)
-	}
-
-	if destinationPorts != "" {
-		fixedPorts := strings.Split(destinationPorts, ",")
-
-		for _, fixedPort := range fixedPorts {
-			var proto, port string
-
-			data := strings.Split(fixedPort, ":")
-			if len(data) == 0 {
-				continue
-			} else if len(data) == 1 {
-				proto = "tcp"
-				port = data[0]
-			} else {
-				proto = data[0]
-				port = data[1]
-			}
-
-			err = i.InsertSourceNatForDestinationPort(vipIP, podIP, port, proto)
-			if err != nil {
-				return fmt.Errorf("error adding snat rules to nat chain [%s], error [%s]", vip.MangleChainName, err)
-			}
-		}
-	} else {
-		err = i.InsertSourceNat(vipIP, podIP)
-		if err != nil {
-			return fmt.Errorf("error adding snat rules to nat chain [%s], error [%s]", vip.MangleChainName, err)
-		}
-	}
-	//_ = i.DumpChain(vip.MangleChainName)
-
-	err = vip.DeleteExistingSessions(podIP, false, destinationPorts, "")
-	if err != nil {
-		return err
-	}
-
-	return nil
-}
-
-func (sm *Manager) AutoDiscoverCIDRs() (serviceCIDR, podCIDR string, err error) {
-	log.Debugf("Trying to automatically discover Service and Pod CIDRs")
-	options := v1.ListOptions{
-		LabelSelector: "component=kube-controller-manager",
-	}
-	podList, err := sm.clientSet.CoreV1().Pods("kube-system").List(context.TODO(), options)
-	if err != nil {
-		return "", "", fmt.Errorf("[Egress] Unable to get kube-controller-manager pod: %w", err)
-	}
-	if len(podList.Items) < 1 {
-		return "", "", fmt.Errorf("[Egress] Unable to auto-discover the pod/service CIDRs: kube-controller-manager not found")
-	}
-
-	pod := podList.Items[0]
-	for flags := range pod.Spec.Containers[0].Command {
-		if strings.Contains(pod.Spec.Containers[0].Command[flags], "--cluster-cidr=") {
-			podCIDR = strings.ReplaceAll(pod.Spec.Containers[0].Command[flags], "--cluster-cidr=", "")
-		}
-		if strings.Contains(pod.Spec.Containers[0].Command[flags], "--service-cluster-ip-range=") {
-			serviceCIDR = strings.ReplaceAll(pod.Spec.Containers[0].Command[flags], "--service-cluster-ip-range=", "")
-		}
-	}
-	if podCIDR == "" || serviceCIDR == "" {
-		err = fmt.Errorf("unable to fully determine cluster CIDR configurations")
-	}
-
-	return
-}
-
-func (sm *Manager) TeardownEgress(podIP, vipIP, destinationPorts, namespace string) error {
-	protocol := iptables.ProtocolIPv4
-	if vip.IsIPv6(podIP) {
-		protocol = iptables.ProtocolIPv6
-	}
-
-	i, err := vip.CreateIptablesClient(sm.config.EgressWithNftables, namespace, protocol)
-	if err != nil {
-		return fmt.Errorf("error Creating iptables client [%s]", err)
-	}
-
-	// Remove the marking of egress packets
-	err = i.DeleteMangleMarking(podIP, vip.MangleChainName)
-	if err != nil {
-		return fmt.Errorf("error changing iptables rules for egress [%s]", err)
-	}
-
-	// Clear up SNAT rules
-	if destinationPorts != "" {
-		fixedPorts := strings.Split(destinationPorts, ",")
-
-		for _, fixedPort := range fixedPorts {
-			var proto, port string
-
-			data := strings.Split(fixedPort, ":")
-			if len(data) == 0 {
-				continue
-			} else if len(data) == 1 {
-				proto = "tcp"
-				port = data[0]
-			} else {
-				proto = data[0]
-				port = data[1]
-			}
-
-			err = i.DeleteSourceNatForDestinationPort(podIP, vipIP, port, proto)
-			if err != nil {
-				return fmt.Errorf("error changing iptables rules for egress [%s]", err)
-			}
-
-		}
-	} else {
-		err = i.DeleteSourceNat(podIP, vipIP)
-		if err != nil {
-			return fmt.Errorf("error changing iptables rules for egress [%s]", err)
-		}
-	}
-	err = vip.DeleteExistingSessions(podIP, false, destinationPorts, "")
-	if err != nil {
-		return fmt.Errorf("error changing iptables rules for egress [%s]", err)
-	}
-	return nil
-}

pkg/manager/services.go

@@ -1,495 +0,0 @@
-package manager
-
-import (
-	"context"
-	"fmt"
-	"slices"
-	"strings"
-	"sync"
-	"time"
-
-	"github.com/google/go-cmp/cmp"
-	log "github.com/sirupsen/logrus"
-	"github.com/vishvananda/netlink"
-	v1 "k8s.io/api/core/v1"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/client-go/util/retry"
-
-	"github.com/kube-vip/kube-vip/pkg/upnp"
-	"github.com/kube-vip/kube-vip/pkg/vip"
-)
-
-const (
-	hwAddrKey                = "kube-vip.io/hwaddr"
-	requestedIP              = "kube-vip.io/requestedIP"
-	vipHost                  = "kube-vip.io/vipHost"
-	egress                   = "kube-vip.io/egress"
-	egressIPv6               = "kube-vip.io/egress-ipv6"
-	egressDestinationPorts   = "kube-vip.io/egress-destination-ports"
-	egressSourcePorts        = "kube-vip.io/egress-source-ports"
-	activeEndpoint           = "kube-vip.io/active-endpoint"
-	activeEndpointIPv6       = "kube-vip.io/active-endpoint-ipv6"
-	flushContrack            = "kube-vip.io/flush-conntrack"
-	loadbalancerIPAnnotation = "kube-vip.io/loadbalancerIPs"
-	loadbalancerHostname     = "kube-vip.io/loadbalancerHostname"
-	serviceInterface         = "kube-vip.io/serviceInterface"
-	upnpEnabled              = "kube-vip.io/forwardUPNP"
-)
-
-func (sm *Manager) syncServices(ctx context.Context, svc *v1.Service, wg *sync.WaitGroup) error {
-	defer wg.Done()
-
-	log.Debugf("[STARTING] Service Sync")
-
-	// Iterate through the synchronising services
-	foundInstance := false
-	newServiceAddresses := fetchServiceAddresses(svc)
-	newServiceUID := string(svc.UID)
-
-	ingressIPs := []string{}
-
-	for _, ingress := range svc.Status.LoadBalancer.Ingress {
-		ingressIPs = append(ingressIPs, ingress.IP)
-	}
-
-	shouldBreake := false
-
-	for x := range sm.serviceInstances {
-		if shouldBreake {
-			break
-		}
-		for _, newServiceAddress := range newServiceAddresses {
-			log.Debugf("isDHCP: %t, newServiceAddress: %s", sm.serviceInstances[x].isDHCP, newServiceAddress)
-			if sm.serviceInstances[x].UID == newServiceUID {
-				// If the found instance's DHCP configuration doesn't match the new service, delete it.
-				if (sm.serviceInstances[x].isDHCP && newServiceAddress != "0.0.0.0") ||
-					(!sm.serviceInstances[x].isDHCP && newServiceAddress == "0.0.0.0") ||
-					(!sm.serviceInstances[x].isDHCP && len(svc.Status.LoadBalancer.Ingress) > 0 && !slices.Contains(ingressIPs, newServiceAddress)) ||
-					(len(svc.Status.LoadBalancer.Ingress) > 0 && !comparePortsAndPortStatuses(svc)) ||
-					(sm.serviceInstances[x].isDHCP && len(svc.Status.LoadBalancer.Ingress) > 0 && !slices.Contains(ingressIPs, sm.serviceInstances[x].dhcpInterfaceIP)) {
-					if err := sm.deleteService(newServiceUID); err != nil {
-						return err
-					}
-					shouldBreake = true
-					break
-				}
-				foundInstance = true
-			}
-		}
-	}
-
-	// This instance wasn't found, we need to add it to the manager
-	if !foundInstance && len(newServiceAddresses) > 0 {
-		if err := sm.addService(ctx, svc); err != nil {
-			return err
-		}
-	}
-
-	return nil
-}
-
-func comparePortsAndPortStatuses(svc *v1.Service) bool {
-	portsStatus := svc.Status.LoadBalancer.Ingress[0].Ports
-	if len(portsStatus) != len(svc.Spec.Ports) {
-		return false
-	}
-	for i, portSpec := range svc.Spec.Ports {
-		if portsStatus[i].Port != portSpec.Port || portsStatus[i].Protocol != portSpec.Protocol {
-			return false
-		}
-	}
-	return true
-}
-
-func (sm *Manager) addService(ctx context.Context, svc *v1.Service) error {
-	startTime := time.Now()
-
-	newService, err := NewInstance(svc, sm.config)
-	if err != nil {
-		return err
-	}
-
-	for x := range newService.vipConfigs {
-		newService.clusters[x].StartLoadBalancerService(newService.vipConfigs[x], sm.bgpServer)
-	}
-
-	sm.upnpMap(ctx, newService)
-
-	if newService.isDHCP && len(newService.vipConfigs) == 1 {
-		go func() {
-			for ip := range newService.dhcpClient.IPChannel() {
-				log.Debugf("IP %s may have changed", ip)
-				newService.vipConfigs[0].VIP = ip
-				newService.dhcpInterfaceIP = ip
-				if !sm.config.DisableServiceUpdates {
-					if err := sm.updateStatus(newService); err != nil {
-						log.Warnf("error updating svc: %s", err)
-					}
-				}
-			}
-			log.Debugf("IP update channel closed, stopping")
-		}()
-	}
-
-	sm.serviceInstances = append(sm.serviceInstances, newService)
-
-	if !sm.config.DisableServiceUpdates {
-		log.Debugf("(svcs) will update [%s/%s]", newService.serviceSnapshot.Namespace, newService.serviceSnapshot.Name)
-		if err := sm.updateStatus(newService); err != nil {
-			// delete service to collect garbage
-			if deleteErr := sm.deleteService(newService.UID); deleteErr != nil {
-				return deleteErr
-			}
-			return err
-		}
-	}
-
-	serviceIPs := fetchServiceAddresses(svc)
-	// Check if we need to flush any conntrack connections (due to some dangling conntrack connections)
-	if svc.Annotations[flushContrack] == "true" {
-
-		log.Debugf("Flushing conntrack rules for service [%s]", svc.Name)
-		for _, serviceIP := range serviceIPs {
-			err = vip.DeleteExistingSessions(serviceIP, false, svc.Annotations[egressDestinationPorts], svc.Annotations[egressSourcePorts])
-			if err != nil {
-				log.Errorf("Error flushing any remaining egress connections [%s]", err)
-			}
-			err = vip.DeleteExistingSessions(serviceIP, true, svc.Annotations[egressDestinationPorts], svc.Annotations[egressSourcePorts])
-			if err != nil {
-				log.Errorf("Error flushing any remaining ingress connections [%s]", err)
-			}
-		}
-	}
-
-	// Check if egress is enabled on the service, if so we'll need to configure some rules
-	if svc.Annotations[egress] == "true" && len(serviceIPs) > 0 {
-		log.Debugf("Enabling egress for the service [%s]", svc.Name)
-		// We will need to modify the iptables rules
-		err = sm.iptablesCheck()
-		if err != nil {
-			log.Errorf("Error configuring egress for loadbalancer [%s]", err)
-		}
-		var podIP string
-		errList := []error{}
-
-		// Should egress be IPv6
-		if svc.Annotations[egressIPv6] == "true" {
-			// Does the service have an active IPv6 endpoint
-			if svc.Annotations[activeEndpointIPv6] != "" {
-				for _, serviceIP := range serviceIPs {
-					if sm.config.EnableEndpointSlices && vip.IsIPv6(serviceIP) {
-
-						podIP = svc.Annotations[activeEndpointIPv6]
-
-						err = sm.configureEgress(serviceIP, podIP, svc.Annotations[egressDestinationPorts], svc.Namespace)
-						if err != nil {
-							errList = append(errList, err)
-							log.Errorf("Error configuring egress for loadbalancer [%s]", err)
-						}
-					}
-				}
-			}
-		} else if svc.Annotations[activeEndpoint] != "" { // Not expected to be IPv6, so should be an IPv4 address
-			for _, serviceIP := range serviceIPs {
-				podIPs := svc.Annotations[activeEndpoint]
-				if sm.config.EnableEndpointSlices && vip.IsIPv6(serviceIP) {
-					podIPs = svc.Annotations[activeEndpointIPv6]
-				}
-				err = sm.configureEgress(serviceIP, podIPs, svc.Annotations[egressDestinationPorts], svc.Namespace)
-				if err != nil {
-					errList = append(errList, err)
-					log.Errorf("Error configuring egress for loadbalancer [%s]", err)
-				}
-			}
-		}
-		if len(errList) == 0 {
-			var provider epProvider
-			if !sm.config.EnableEndpointSlices {
-				provider = &endpointsProvider{label: "endpoints"}
-			} else {
-				provider = &endpointslicesProvider{label: "endpointslices"}
-			}
-			err = provider.updateServiceAnnotation(svc.Annotations[activeEndpoint], svc.Annotations[activeEndpointIPv6], svc, sm)
-			if err != nil {
-				log.Errorf("error configuring egress annotation for loadbalancer [%s]", err)
-			}
-		}
-	}
-
-	finishTime := time.Since(startTime)
-	log.Infof("[service] synchronised in %dms", finishTime.Milliseconds())
-
-	return nil
-}
-
-func (sm *Manager) deleteService(uid string) error {
-	// protect multiple calls
-	sm.mutex.Lock()
-	defer sm.mutex.Unlock()
-
-	var updatedInstances []*Instance
-	var serviceInstance *Instance
-	found := false
-	for x := range sm.serviceInstances {
-		log.Debugf("Looking for [%s], found [%s]", uid, sm.serviceInstances[x].UID)
-		// Add the running services to the new array
-		if sm.serviceInstances[x].UID != uid {
-			updatedInstances = append(updatedInstances, sm.serviceInstances[x])
-		} else {
-			// Flip the found when we match
-			found = true
-			serviceInstance = sm.serviceInstances[x]
-		}
-	}
-	// If we've been through all services and not found the correct one then error
-	if !found {
-		// TODO: - fix UX
-		// return fmt.Errorf("unable to find/stop service [%s]", uid)
-		return nil
-	}
-	shared := false
-	vipSet := make(map[string]interface{})
-	for x := range updatedInstances {
-		for _, vip := range updatedInstances[x].VIPs {
-			vipSet[vip] = nil
-		}
-	}
-	for _, vip := range serviceInstance.VIPs {
-		if _, found := vipSet[vip]; found {
-			shared = true
-		}
-	}
-	if !shared {
-		for x := range serviceInstance.clusters {
-			serviceInstance.clusters[x].Stop()
-		}
-		if serviceInstance.isDHCP {
-			serviceInstance.dhcpClient.Stop()
-			macvlan, err := netlink.LinkByName(serviceInstance.dhcpInterface)
-			if err != nil {
-				return fmt.Errorf("error finding VIP Interface: %v", err)
-			}
-
-			err = netlink.LinkDel(macvlan)
-			if err != nil {
-				return fmt.Errorf("error deleting DHCP Link : %v", err)
-			}
-		}
-		// TODO: Implement dual-stack loadbalancer support if BGP is enabled
-		for i := range serviceInstance.vipConfigs {
-			if serviceInstance.vipConfigs[i].EnableBGP {
-				cidrVip := fmt.Sprintf("%s/%s", serviceInstance.vipConfigs[i].VIP, serviceInstance.vipConfigs[i].VIPCIDR)
-				err := sm.bgpServer.DelHost(cidrVip)
-				if err != nil {
-					return fmt.Errorf("[BGP] error deleting BGP host: %v", err)
-				}
-				log.Debugf("[BGP] deleted host: %s", cidrVip)
-			}
-		}
-
-		// We will need to tear down the egress
-		if serviceInstance.serviceSnapshot.Annotations[egress] == "true" {
-			if serviceInstance.serviceSnapshot.Annotations[activeEndpoint] != "" {
-				log.Infof("service [%s] has an egress re-write enabled", serviceInstance.serviceSnapshot.Name)
-				err := sm.TeardownEgress(serviceInstance.serviceSnapshot.Annotations[activeEndpoint], serviceInstance.serviceSnapshot.Spec.LoadBalancerIP, serviceInstance.serviceSnapshot.Annotations[egressDestinationPorts], serviceInstance.serviceSnapshot.Namespace)
-				if err != nil {
-					log.Errorf("%v", err)
-				}
-			}
-		}
-	}
-
-	// Update the service array
-	sm.serviceInstances = updatedInstances
-
-	log.Infof("Removed [%s] from manager, [%d] advertised services remain", uid, len(sm.serviceInstances))
-
-	return nil
-}
-
-// Set up UPNP forwards for a service
-// We first try to use the more modern Pinhole API introduced in UPNPv2 and fall back to UPNPv2 Port Forwarding if no forward was successful
-func (sm *Manager) upnpMap(ctx context.Context, s *Instance) {
-	if !isUPNPEnabled(s.serviceSnapshot) {
-		// Skip services missing the annotation
-		return
-	}
-	if !sm.upnp {
-		log.Warnf("[UPNP] Found kube-vip.io/forwardUPNP on service while UPNP forwarding is disabled in the kube-vip config. Not forwarding service %s", s.serviceSnapshot.Name)
-	}
-	// If upnp is enabled then update the gateway/router with the address
-	// TODO - check if this implementation for dualstack is correct
-
-	gateways := upnp.GetGatewayClients(ctx)
-
-	// Reset Gateway IPs to remove stale addresses
-	s.upnpGatewayIPs = make([]string, 0)
-
-	for _, vip := range s.VIPs {
-		for _, port := range s.ExternalPorts {
-			for _, gw := range gateways {
-				log.Infof("[UPNP] Adding map to [%s:%d - %s] on gateway %s", vip, port.Port, s.serviceSnapshot.Name, gw.WANIPv6FirewallControlClient.Location)
-
-				forwardSucessful := false
-				if gw.WANIPv6FirewallControlClient != nil {
-					pinholeID, pinholeErr := gw.WANIPv6FirewallControlClient.AddPinholeCtx(ctx, "0.0.0.0", uint16(port.Port), vip, uint16(port.Port), upnp.MapProtocolToIANA(port.Type), 3600)
-					if pinholeErr == nil {
-						forwardSucessful = true
-						log.Infof("[UPNP] Service should be accessible externally on port [%d]; PinholeID is [%d]", port.Port, pinholeID)
-					} else {
-						//TODO: Cleanup
-						log.Errorf("[UPNP] Unable to map port to gateway using Pinhole API[%s]", pinholeErr.Error())
-					}
-				}
-				// Fallback to PortForward
-				if !forwardSucessful {
-					portMappingErr := gw.ConnectionClient.AddPortMapping("0.0.0.0", uint16(port.Port), strings.ToUpper(port.Type), uint16(port.Port), vip, true, s.serviceSnapshot.Name, 3600)
-					if portMappingErr == nil {
-						log.Infof("[UPNP] Service should be accessible externally on port [%d]", port.Port)
-						forwardSucessful = true
-					} else {
-						//TODO: Cleanup
-						log.Errorf("[UPNP] Unable to map port to gateway using PortForward API[%s]", portMappingErr.Error())
-					}
-				}
-
-				if forwardSucessful {
-					ip, err := gw.ConnectionClient.GetExternalIPAddress()
-					if err == nil {
-						s.upnpGatewayIPs = append(s.upnpGatewayIPs, ip)
-					}
-				}
-			}
-		}
-	}
-
-	// Remove duplicate IPs
-	slices.Sort(s.upnpGatewayIPs)
-	s.upnpGatewayIPs = slices.Compact(s.upnpGatewayIPs)
-}
-
-func (sm *Manager) updateStatus(i *Instance) error {
-	retryErr := retry.RetryOnConflict(retry.DefaultRetry, func() error {
-		// Retrieve the latest version of Deployment before attempting update
-		// RetryOnConflict uses exponential backoff to avoid exhausting the apiserver
-		currentService, err := sm.clientSet.CoreV1().Services(i.serviceSnapshot.Namespace).Get(context.TODO(), i.serviceSnapshot.Name, metav1.GetOptions{})
-		if err != nil {
-			return err
-		}
-
-		currentServiceCopy := currentService.DeepCopy()
-		if currentServiceCopy.Annotations == nil {
-			currentServiceCopy.Annotations = make(map[string]string)
-		}
-
-		// If we're using ARP then we can only broadcast the VIP from one place, add an annotation to the service
-		if sm.config.EnableARP {
-			// Add the current host
-			currentServiceCopy.Annotations[vipHost] = sm.config.NodeName
-		}
-		if i.dhcpInterfaceHwaddr != "" || i.dhcpInterfaceIP != "" {
-			currentServiceCopy.Annotations[hwAddrKey] = i.dhcpInterfaceHwaddr
-			currentServiceCopy.Annotations[requestedIP] = i.dhcpInterfaceIP
-		}
-
-		if !cmp.Equal(currentService, currentServiceCopy) {
-			currentService, err = sm.clientSet.CoreV1().Services(currentServiceCopy.Namespace).Update(context.TODO(), currentServiceCopy, metav1.UpdateOptions{})
-			if err != nil {
-				log.Errorf("Error updating Service Spec [%s] : %v", i.serviceSnapshot.Name, err)
-				return err
-			}
-		}
-
-		ports := make([]v1.PortStatus, 0, len(i.serviceSnapshot.Spec.Ports))
-		for _, port := range i.serviceSnapshot.Spec.Ports {
-			ports = append(ports, v1.PortStatus{
-				Port:     port.Port,
-				Protocol: port.Protocol,
-			})
-		}
-
-		ingresses := []v1.LoadBalancerIngress{}
-
-		for _, c := range i.vipConfigs {
-			if !vip.IsIP(c.VIP) {
-				ips, err := vip.LookupHost(c.VIP, sm.config.DNSMode)
-				if err != nil {
-					return err
-				}
-				for _, ip := range ips {
-					i := v1.LoadBalancerIngress{
-						IP:    ip,
-						Ports: ports,
-					}
-					ingresses = append(ingresses, i)
-				}
-			} else {
-				i := v1.LoadBalancerIngress{
-					IP:    c.VIP,
-					Ports: ports,
-				}
-				ingresses = append(ingresses, i)
-			}
-			if isUPNPEnabled(currentService) {
-				for _, ip := range i.upnpGatewayIPs {
-					i := v1.LoadBalancerIngress{
-						IP:    ip,
-						Ports: ports,
-					}
-					ingresses = append(ingresses, i)
-				}
-			}
-		}
-		if !cmp.Equal(currentService.Status.LoadBalancer.Ingress, ingresses) {
-			currentService.Status.LoadBalancer.Ingress = ingresses
-			_, err = sm.clientSet.CoreV1().Services(currentService.Namespace).UpdateStatus(context.TODO(), currentService, metav1.UpdateOptions{})
-			if err != nil {
-				log.Errorf("Error updating Service %s/%s Status: %v", i.serviceSnapshot.Namespace, i.serviceSnapshot.Name, err)
-				return err
-			}
-		}
-		return nil
-	})
-
-	if retryErr != nil {
-		log.Errorf("Failed to set Services: %v", retryErr)
-		return retryErr
-	}
-	return nil
-}
-
-// fetchServiceAddresses tries to get the addresses from annotations
-// kube-vip.io/loadbalancerIPs, then from spec.loadbalancerIP
-func fetchServiceAddresses(s *v1.Service) []string {
-	annotationAvailable := false
-	if s.Annotations != nil {
-		if v, annotationAvailable := s.Annotations[loadbalancerIPAnnotation]; annotationAvailable {
-			ips := strings.Split(v, ",")
-			var trimmedIPs []string
-			for _, ip := range ips {
-				trimmedIPs = append(trimmedIPs, strings.TrimSpace(ip))
-			}
-			return trimmedIPs
-		}
-	}
-
-	if !annotationAvailable {
-		if len(s.Status.LoadBalancer.Ingress) > 0 {
-			addresses := []string{}
-			for _, ingress := range s.Status.LoadBalancer.Ingress {
-				addresses = append(addresses, ingress.IP)
-			}
-			return addresses
-		}
-	}
-
-	if s.Spec.LoadBalancerIP != "" {
-		return []string{s.Spec.LoadBalancerIP}
-	}
-
-	return []string{}
-}
-
-func isUPNPEnabled(s *v1.Service) bool {
-	return metav1.HasAnnotation(s.ObjectMeta, upnpEnabled) && s.Annotations[upnpEnabled] == "true"
-}

pkg/manager/servicesLeader.go

@@ -1,102 +0,0 @@
-package manager
-
-import (
-	"context"
-	"fmt"
-	"sync"
-	"time"
-
-	log "github.com/sirupsen/logrus"
-	v1 "k8s.io/api/core/v1"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/client-go/tools/leaderelection"
-	"k8s.io/client-go/tools/leaderelection/resourcelock"
-)
-
-// The startServicesWatchForLeaderElection function will start a services watcher, the
-func (sm *Manager) startServicesWatchForLeaderElection(ctx context.Context) error {
-	err := sm.servicesWatcher(ctx, sm.StartServicesLeaderElection)
-	if err != nil {
-		return err
-	}
-
-	for _, instance := range sm.serviceInstances {
-		for _, cluster := range instance.clusters {
-			for i := range cluster.Network {
-				_ = cluster.Network[i].DeleteRoute()
-			}
-			cluster.Stop()
-		}
-	}
-
-	log.Infof("Shutting down kube-Vip")
-
-	return nil
-}
-
-// The startServicesWatchForLeaderElection function will start a services watcher, the
-func (sm *Manager) StartServicesLeaderElection(ctx context.Context, service *v1.Service, wg *sync.WaitGroup) error {
-	serviceLease := fmt.Sprintf("kubevip-%s", service.Name)
-	log.Infof("(svc election) service [%s], namespace [%s], lock name [%s], host id [%s]", service.Name, service.Namespace, serviceLease, sm.config.NodeName)
-	// we use the Lease lock type since edits to Leases are less common
-	// and fewer objects in the cluster watch "all Leases".
-	lock := &resourcelock.LeaseLock{
-		LeaseMeta: metav1.ObjectMeta{
-			Name:      serviceLease,
-			Namespace: service.Namespace,
-		},
-		Client: sm.clientSet.CoordinationV1(),
-		LockConfig: resourcelock.ResourceLockConfig{
-			Identity: sm.config.NodeName,
-		},
-	}
-
-	activeService[string(service.UID)] = true
-	// start the leader election code loop
-	leaderelection.RunOrDie(ctx, leaderelection.LeaderElectionConfig{
-		Lock: lock,
-		// IMPORTANT: you MUST ensure that any code you have that
-		// is protected by the lease must terminate **before**
-		// you call cancel. Otherwise, you could have a background
-		// loop still running and another process could
-		// get elected before your background loop finished, violating
-		// the stated goal of the lease.
-		ReleaseOnCancel: true,
-		LeaseDuration:   time.Duration(sm.config.LeaseDuration) * time.Second,
-		RenewDeadline:   time.Duration(sm.config.RenewDeadline) * time.Second,
-		RetryPeriod:     time.Duration(sm.config.RetryPeriod) * time.Second,
-		Callbacks: leaderelection.LeaderCallbacks{
-			OnStartedLeading: func(ctx context.Context) {
-				// Mark this service as active (as we've started leading)
-				// we run this in background as it's blocking
-				wg.Add(1)
-				go func() {
-					if err := sm.syncServices(ctx, service, wg); err != nil {
-						log.Errorln(err)
-					}
-				}()
-			},
-			OnStoppedLeading: func() {
-				// we can do cleanup here
-				log.Infof("(svc election) service [%s] leader lost: [%s]", service.Name, sm.config.NodeName)
-				if activeService[string(service.UID)] {
-					if err := sm.deleteService(string(service.UID)); err != nil {
-						log.Errorln(err)
-					}
-				}
-				// Mark this service is inactive
-				activeService[string(service.UID)] = false
-			},
-			OnNewLeader: func(identity string) {
-				// we're notified when new leader elected
-				if identity == sm.config.NodeName {
-					// I just got the lock
-					return
-				}
-				log.Infof("(svc election) new leader elected: %s", identity)
-			},
-		},
-	})
-	log.Infof("(svc election) for service [%s] stopping", service.Name)
-	return nil
-}

pkg/manager/watch_annotations.go

@@ -8,14 +8,15 @@ import (
 	"strconv"
 	"strings"
 
-	"github.com/kube-vip/kube-vip/pkg/bgp"
-	log "github.com/sirupsen/logrus"
+	log "log/slog"
+
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/utils"
 
-	"github.com/davecgh/go-spew/spew"
 	v1 "k8s.io/api/core/v1"
-	apierrors "k8s.io/apimachinery/pkg/api/errors"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/labels"
+	"k8s.io/client-go/kubernetes"
 	watchtools "k8s.io/client-go/tools/watch"
 
 	"k8s.io/apimachinery/pkg/watch"
@@ -24,18 +25,19 @@ import (
 
 // This file handles the watching of node annotations for configuration, it will exit once the annotations are
 // present
-func (sm *Manager) annotationsWatcher() error {
+func annotationsWatcher(ctx context.Context, clientSet,
+	rwClientSet kubernetes.Interface, config *kubevip.Config) error {
 	// Use a restartable watcher, as this should help in the event of etcd or timeout issues
-	log.Infof("Kube-Vip is waiting for annotation prefix [%s] to be present on this node", sm.config.Annotations)
+	log.Info("Kube-Vip is waiting for annotation prefix to be present on this node", "prefix", config.Annotations)
 
-	labelSelector := metav1.LabelSelector{MatchLabels: map[string]string{"kubernetes.io/hostname": sm.config.NodeName}}
+	labelSelector := metav1.LabelSelector{MatchLabels: map[string]string{"kubernetes.io/hostname": config.NodeName}}
 	listOptions := metav1.ListOptions{
 		LabelSelector: labels.Set(labelSelector.MatchLabels).String(),
 	}
 
 	// First we'll check the annotations for the node and if
 	// they aren't what are expected, we'll drop into the watch until they are
-	nodeList, err := sm.clientSet.CoreV1().Nodes().List(context.Background(), listOptions)
+	nodeList, err := clientSet.CoreV1().Nodes().List(ctx, listOptions)
 	if err != nil {
 		return err
 	}
@@ -44,41 +46,31 @@ func (sm *Manager) annotationsWatcher() error {
 	// there's probably bigger problems
 	node := nodeList.Items[0]
 
-	bgpConfig, bgpPeer, err := parseBgpAnnotations(sm.config.BGPConfig, &node, sm.config.Annotations)
+	bgpConfig, bgpPeer, err := parseBgpAnnotations(config.BGPConfig, &node, config.Annotations)
 	if err == nil {
 		// No error, the annotations already exist
-		sm.config.BGPConfig = bgpConfig
-		sm.config.BGPPeerConfig = bgpPeer
+		config.BGPConfig = bgpConfig
+		config.BGPPeerConfig = bgpPeer
 		return nil
 	}
 
 	// We got an error with the annotations, falling back to the watch until
 	// they're as needed
-	log.Warn(err)
+	log.Warn(err.Error())
 
-	rw, err := watchtools.NewRetryWatcher(node.ResourceVersion, &cache.ListWatch{
-		WatchFunc: func(options metav1.ListOptions) (watch.Interface, error) {
-			return sm.rwClientSet.CoreV1().Nodes().Watch(context.Background(), listOptions)
+	rw, err := watchtools.NewRetryWatcherWithContext(ctx, node.ResourceVersion, &cache.ListWatch{
+		WatchFunc: func(_ metav1.ListOptions) (watch.Interface, error) {
+			return rwClientSet.CoreV1().Nodes().Watch(ctx, listOptions)
 		},
 	})
+
 	if err != nil {
 		return fmt.Errorf("error creating annotations watcher: %s", err.Error())
 	}
 
-	exitFunction := make(chan struct{})
-	go func() {
-		select {
-		case <-sm.shutdownChan:
-			log.Debug("[annotations] shutdown called")
-			// Stop the retry watcher
-			rw.Stop()
-			return
-		case <-exitFunction:
-			log.Debug("[annotations] function ending")
-			// Stop the retry watcher
-			rw.Stop()
-			return
-		}
+	defer func() {
+		rw.Stop()
+		log.Debug("[annotations] watcher stopped")
 	}()
 
 	ch := rw.ResultChan()
@@ -92,46 +84,38 @@ func (sm *Manager) annotationsWatcher() error {
 				return fmt.Errorf("unable to parse Kubernetes Node from Annotation watcher")
 			}
 
-			bgpConfig, bgpPeer, err := parseBgpAnnotations(sm.config.BGPConfig, node, sm.config.Annotations)
+			bgpConfig, bgpPeer, err := parseBgpAnnotations(config.BGPConfig, node, config.Annotations)
 			if err != nil {
-				log.Error(err)
+				log.Error(err.Error())
 				continue
 			}
 
-			sm.config.BGPConfig = bgpConfig
-			sm.config.BGPPeerConfig = bgpPeer
+			config.BGPConfig = bgpConfig
+			config.BGPPeerConfig = bgpPeer
 
-			rw.Stop()
+			log.Info("[annotations] exiting Annotations watcher - annotations found")
+			return nil
 		case watch.Deleted:
 			node, ok := event.Object.(*v1.Node)
 			if !ok {
 				return fmt.Errorf("unable to parse Kubernetes Node from Kubernetes watcher")
 			}
 
-			log.Infof("Node [%s] has been deleted", node.Name)
+			log.Info("Node has been deleted", "name", node.Name)
 
 		case watch.Bookmark:
 			// Un-used
 		case watch.Error:
 			log.Error("Error attempting to watch Kubernetes Nodes")
-
-			// This round trip allows us to handle unstructured status
-			errObject := apierrors.FromObject(event.Object)
-			statusErr, ok := errObject.(*apierrors.StatusError)
-			if !ok {
-				log.Errorf(spew.Sprintf("Received an error which is not *metav1.Status but %#+v", event.Object))
-
-			}
-
-			status := statusErr.ErrStatus
-			log.Errorf("%v", status)
+			log.Error("annotations watcher failed", "err", utils.WatchError(event.Object))
 		default:
 		}
 	}
-	close(exitFunction)
-	log.Infoln("Exiting Annotations watcher")
-	return nil
-
+	log.Info("[annotations] exiting annotations watcher")
+	if ctx.Err() != nil {
+		return nil
+	}
+	return utils.NewPanicError("annotations watcher channel closed unexpectedly")
 }
 
 // parseNodeAnnotations parses the annotations on the node and updates the configuration
@@ -149,12 +133,14 @@ func (sm *Manager) annotationsWatcher() error {
 // * `<info>` is the relevant information, such as `node-asn` or `peer-ip`
 // * `{{n}}` is the number of the peer, always starting with `0`
 // * kube-vip is only designed to manage one peer, just look for {{n}} == 0
-func parseBgpAnnotations(bgpConfig bgp.Config, node *v1.Node, prefix string) (bgp.Config, bgp.Peer, error) {
-	bgpPeer := bgp.Peer{}
+func parseBgpAnnotations(bgpConfig kubevip.BGPConfig, node *v1.Node, prefix string) (kubevip.BGPConfig, kubevip.BGPPeer, error) {
+	bgpPeer := kubevip.BGPPeer{}
+
+	prefix = regexp.QuoteMeta(prefix)
 
 	nodeASN := ""
+	regex := regexp.MustCompile(fmt.Sprintf("^%s/(bgp-peers-0-)?node-asn$", prefix))
 	for k, v := range node.Annotations {
-		regex := regexp.MustCompile(fmt.Sprintf("^%s/(bgp-peers-0-)?node-asn", prefix))
 		if regex.Match([]byte(k)) {
 			nodeASN = v
 		}
@@ -171,8 +157,8 @@ func parseBgpAnnotations(bgpConfig bgp.Config, node *v1.Node, prefix string) (bg
 	bgpConfig.AS = uint32(u64)
 
 	srcIP := ""
+	regex = regexp.MustCompile(fmt.Sprintf("^%s/(bgp-peers-0-)?src-ip$", prefix))
 	for k, v := range node.Annotations {
-		regex := regexp.MustCompile(fmt.Sprintf("^%s/(bgp-peers-0-)?src-ip", prefix))
 		if regex.Match([]byte(k)) {
 			srcIP = v
 		}
@@ -186,8 +172,8 @@ func parseBgpAnnotations(bgpConfig bgp.Config, node *v1.Node, prefix string) (bg
 	bgpConfig.RouterID, bgpConfig.SourceIP = srcIP, srcIP
 
 	peerASN := ""
+	regex = regexp.MustCompile(fmt.Sprintf("^%s/(bgp-peers-0-)?peer-asn$", prefix))
 	for k, v := range node.Annotations {
-		regex := regexp.MustCompile(fmt.Sprintf("^%s/(bgp-peers-0-)?peer-asn", prefix))
 		if regex.Match([]byte(k)) {
 			peerASN = v
 		}
@@ -204,8 +190,8 @@ func parseBgpAnnotations(bgpConfig bgp.Config, node *v1.Node, prefix string) (bg
 	bgpPeer.AS = uint32(u64)
 
 	peerIPString := ""
+	regex = regexp.MustCompile(fmt.Sprintf("^%s/(bgp-peers-[0-9]+-)?peer-ip$", prefix))
 	for k, v := range node.Annotations {
-		regex := regexp.MustCompile(fmt.Sprintf("^%s/(bgp-peers-[0-9]+-)?peer-ip", prefix))
 		if regex.Match([]byte(k)) {
 			peerIPString += v + ","
 		}
@@ -213,8 +199,13 @@ func parseBgpAnnotations(bgpConfig bgp.Config, node *v1.Node, prefix string) (bg
 	peerIPString = strings.TrimRight(peerIPString, ",")
 
 	peerIPs := strings.Split(peerIPString, ",")
+	if len(peerIPs) >= 1 && peerIPs[0] == "" || len(peerIPs) == 0 {
+		return bgpConfig, bgpPeer, fmt.Errorf("peer-ip value missing or empty")
+	}
 
-	bgpConfig.Peers = make([]bgp.Peer, 0, len(peerIPs))
+	bgpConfig.Peers = make([]kubevip.BGPPeer, 0, len(peerIPs))
+	regexPass := regexp.MustCompile(fmt.Sprintf("^%s/(bgp-peers-0-)?bgp-pass$", prefix))
+	regexMultiHop := regexp.MustCompile(fmt.Sprintf("^%s/(bgp-peers-0-)?peer-multi-hop$", prefix))
 	for _, peerIP := range peerIPs {
 		ipAddr := strings.TrimSpace(peerIP)
 
@@ -223,8 +214,7 @@ func parseBgpAnnotations(bgpConfig bgp.Config, node *v1.Node, prefix string) (bg
 			// Check if we're also expecting a password for this peer
 			base64BGPPassword := ""
 			for k, v := range node.Annotations {
-				regex := regexp.MustCompile(fmt.Sprintf("^%s/(bgp-peers-0-)?bgp-pass", prefix))
-				if regex.Match([]byte(k)) {
+				if regexPass.Match([]byte(k)) {
 					base64BGPPassword = v
 				}
 			}
@@ -237,6 +227,21 @@ func parseBgpAnnotations(bgpConfig bgp.Config, node *v1.Node, prefix string) (bg
 				// Set the password for each peer
 				bgpPeer.Password = string(decodedPassword)
 			}
+
+			// Check if multi-hop is enabled.
+			for k, v := range node.Annotations {
+				if regexMultiHop.MatchString(k) {
+					switch v {
+					case "true":
+						bgpPeer.MultiHop = true
+					case "false":
+						bgpPeer.MultiHop = false
+					default:
+						return bgpConfig, bgpPeer,
+							fmt.Errorf("invalid %q annotation value: %q, must be \"true\" or \"false\"", k, v)
+					}
+				}
+			}
 			bgpConfig.Peers = append(bgpConfig.Peers, bgpPeer)
 		}
 	}