Merge pull request #1575 from kube-vip/thebsdbox-patch-1

Update Makefile for new release

.github/workflows/anchore-syft.yml

@@ -26,6 +26,6 @@ jobs:
         with:
           ref: ${{ github.ref_name }}
       - name: Anchore SBOM Action
-        uses: anchore/sbom-action@v0.20.10
+        uses: anchore/sbom-action@v0.24.0
         with:
           format: cyclonedx-json

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

@@ -25,26 +25,26 @@ jobs:
         run: make dockerx86Local
       - name: Run Manifest generation tests
         run: make manifest-test
-      - name: Run ARP mode tests v1.29.0 onwards
-        run: E2E_KEEP_LOGS=true make e2e-tests129-arp
+      - name: Run ARP mode tests
+        run: DOCKER_API_VERSION=1.48 E2E_KEEP_LOGS=true make e2e-tests-arp
         if: matrix.mode== 'arp'
-      - name: Run RT mode tests v1.29.0 onwards
-        run: E2E_KEEP_LOGS=true make e2e-tests129-rt
+      - name: Run RT mode tests
+        run: DOCKER_API_VERSION=1.48 E2E_KEEP_LOGS=true 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 v1.29.0 onwards
-        run: sudo -E PATH=$PATH DOCKER_API_VERSION=1.48 E2E_KEEP_LOGS=true make e2e-tests129-bgp
+      - name: Run BGP mode tests
+        run: sudo -E PATH=$PATH DOCKER_API_VERSION=1.48 E2E_KEEP_LOGS=true make e2e-tests-bgp
         if: matrix.mode== 'bgp'
       - name: Change log directory permissions
-        run: sudo chmod -R 755 /tmp/kube-vip-test-${{ matrix.mode }}*
+        run: sudo chmod -R 755 /tmp/kube-vip-test*
         if: matrix.mode== 'bgp' && always()
       - name: Save logs
-        uses: actions/upload-artifact@v5
+        uses: actions/upload-artifact@v7
         with:
           name: e2e-test-logs-${{ matrix.mode }}-${{ steps.date.outputs.date }}
-          path: /tmp/kube-vip-test-${{ matrix.mode }}*
+          path: /tmp/kube-vip-test*
         if: always()
   service-e2e-tests:
     runs-on: ubuntu-latest
@@ -62,10 +62,10 @@ jobs:
       - name: Build image with iptables
         run: make dockerx86ActionIPTables
       - name: Run tests
-        run: DOCKERTAG=action E2E_KEEP_LOGS=true make service-tests
+        run: DOCKER_API_VERSION=1.48 DOCKERTAG=action E2E_KEEP_LOGS=true make service-tests
       - name: Save logs
-        uses: actions/upload-artifact@v5
+        uses: actions/upload-artifact@v7
         with:
           name: services-test-logs-${{ steps.date.outputs.date }}
-          path: /tmp/kube-vip-service-tests*
+          path: /tmp/kube-vip-services*
         if: always()

.github/workflows/ci.yaml

@@ -10,7 +10,10 @@ jobs:
       - name: Init
         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 v1.64.8
+        uses: golangci/golangci-lint-action@v9
+        with:
+          version: v2.12
+          install-only: true
       - name: Checkout code
         uses: actions/checkout@v6
       - name: Install Go

.github/workflows/main.yaml

@@ -13,23 +13,23 @@ jobs:
       - name: Checkout
         uses: actions/checkout@v6
       - 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

@@ -13,29 +13,29 @@ jobs:
       - name: Checkout
         uses: actions/checkout@v6
       - 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@v5.9.0
+        uses: docker/metadata-action@v6.1.0
         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
@@ -48,7 +48,7 @@ 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

.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

@@ -7,7 +7,26 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Fixed
+- 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:
@@ -26,6 +45,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Changed
 - 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
 

Dockerfile

@@ -1,6 +1,6 @@
 # syntax=docker/dockerfile:experimental
 
-FROM golang:1.25.4-alpine3.22 as dev
+FROM golang:1.26.3-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:1.25.4-alpine3.22 as dev
+FROM golang:1.26.3-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:3.22.2
+FROM alpine:3.23.4
 # 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 := v1.0.2
+VERSION := v1.2.0
 
 BUILD := `git rev-parse HEAD`
 
@@ -16,8 +16,10 @@ TARGETOS=linux
 LDFLAGS=-ldflags "-s -w -X=main.Version=$(VERSION) -X=main.Build=$(BUILD) -extldflags -static"
 DOCKERTAG ?= $(VERSION)
 REPOSITORY ?= docker.io/plndr
+GO_VERSION := 1.25.6
+K8S_VERSION ?= v1.35.0
 
-.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
 
@@ -127,26 +129,27 @@ manifest-test:
 	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:
-	GOMAXPROCS=4 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 -p ./testing/e2e
 
-e2e-tests129-arp:
-	GOMAXPROCS=4 TEST_MODE=arp 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 -p ./testing/e2e
 
-e2e-tests129-rt:
-	GOMAXPROCS=4 TEST_MODE=rt V129=true K8S_IMAGE_PATH=kindest/node:v1.29.0 E2E_IMAGE_PATH=$(REPOSITORY)/$(TARGET):$(DOCKERTAG) go run github.com/onsi/ginkgo/v2/ginkgo --tags=e2e -v -p ./testing/e2e
+e2e-tests-bgp: get-whoami get-gobgp
+	GOMAXPROCS=4 TEST_MODE=bgp K8S_IMAGE_PATH=kindest/node:$(K8S_VERSION) E2E_IMAGE_PATH=$(REPOSITORY)/$(TARGET):$(DOCKERTAG) go run github.com/onsi/ginkgo/v2/ginkgo --tags=e2e -v -p ./testing/e2e
 
-e2e-tests129-bgp:
-	GOMAXPROCS=4 TEST_MODE=bgp 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-tests129: e2e-tests129-arp e2e-tests129-rt e2e-tests129-bgp
+e2e-tests: e2e-tests-arp e2e-tests-rt e2e-tests-bgp
 
 service-tests:
+	$(MAKE) -C testing/e2e/e2e dockerLocal
 	E2E_IMAGE_PATH=$(REPOSITORY)/$(TARGET):$(DOCKERTAG) go run ./testing/services -Services -simple -deployments -leaderActive -leaderFailover -localDeploy -egress -egressIPv6 -dualStack
 
 trivy: dockerx86ActionIPTables
@@ -177,3 +180,5 @@ get-gobgp:
 	wget -nc --directory-prefix=bin https://github.com/osrg/gobgp/releases/download/v3.37.0/gobgp_3.37.0_linux_amd64.tar.gz
 	tar -xvzf bin/gobgp_3.37.0_linux_amd64.tar.gz -C bin
 
+get-whoami:
+	docker pull ghcr.io/traefik/whoami:v1.11

README.md

@@ -58,14 +58,56 @@ 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.
+
 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)

cmd/kube-vip-kubeadm.go

@@ -42,6 +42,10 @@ var kubeKubeadmInit = &cobra.Command{
 			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 == "" {
@@ -65,7 +69,11 @@ var kubeKubeadmInit = &cobra.Command{
 			}
 		}
 
-		cfg := kubevip.GeneratePodManifestFromConfig(&initConfig, image, 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
 	},
 }
@@ -82,6 +90,10 @@ var kubeKubeadmJoin = &cobra.Command{
 			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 == "" {
@@ -110,7 +122,11 @@ var kubeKubeadmJoin = &cobra.Command{
 			}
 		}
 
-		cfg := kubevip.GeneratePodManifestFromConfig(&initConfig, image, 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

@@ -50,11 +50,15 @@ var kubeManifestPod = &cobra.Command{
 			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.Error("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
 		}
 
@@ -67,7 +71,11 @@ var kubeManifestPod = &cobra.Command{
 			}
 		}
 
-		cfg := kubevip.GeneratePodManifestFromConfig(&initConfig, image, 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
 	},
 }
@@ -84,10 +92,14 @@ var kubeManifestDaemon = &cobra.Command{
 			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.Error("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
 		}
 
@@ -100,7 +112,11 @@ var kubeManifestDaemon = &cobra.Command{
 			}
 		}
 
-		cfg := kubevip.GenerateDaemonsetManifestFromConfig(&initConfig, image, 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
 	},
 }
@@ -117,11 +133,15 @@ var kubeManifestRbac = &cobra.Command{
 			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.Error("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
 		}
 

cmd/kube-vip.go

@@ -7,12 +7,13 @@ import (
 	"net/http"
 	"os"
 	"slices"
+	"strconv"
 	"strings"
+	"sync"
 	"time"
 
 	log "log/slog"
 
-	"github.com/prometheus/client_golang/prometheus"
 	"github.com/prometheus/client_golang/prometheus/promhttp"
 	"github.com/spf13/cobra"
 	"github.com/vishvananda/netlink"
@@ -20,6 +21,7 @@ import (
 
 	"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"
 )
@@ -80,9 +82,9 @@ 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")
+	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")
@@ -103,6 +105,11 @@ func init() {
 	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")
@@ -133,6 +140,9 @@ func init() {
 	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.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")
@@ -210,6 +220,10 @@ var kubeVipService = &cobra.Command{
 			log.Error("parsing env", "err", err)
 			return
 		}
+		if err := initConfig.Validate(); err != nil {
+			log.Error("validating configuration", "err", err)
+			return
+		}
 
 		// Change RTN_UNSPEC to default type
 		if initConfig.RoutingProtocol == unix.RTN_UNSPEC {
@@ -230,6 +244,15 @@ var kubeVipService = &cobra.Command{
 			configMap = envConfigMap
 		}
 
+		// 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 {
+				log.Error("checking subnet exists if vip_address defined", "err", err)
+				return
+			}
+		}
+
 		// Ensure there is an address to generate the CIDR from
 		if initConfig.VIPSubnet == "" && initConfig.Address != "" {
 			initConfig.VIPSubnet, err = GenerateCidrRange(initConfig.Address, initConfig.DNSMode)
@@ -239,15 +262,18 @@ var kubeVipService = &cobra.Command{
 			}
 		}
 
+		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.Error("new manager", "err", err)
 			return
 		}
 
 		// 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.Error("manager start", "err", err)
 			return
@@ -274,6 +300,10 @@ var kubeVipManager = &cobra.Command{
 			log.Error("parsing environment", "err", err)
 			return
 		}
+		if err := initConfig.Validate(); err != nil {
+			log.Error("validating configuration", "err", err)
+			return
+		}
 
 		// Change RTN_UNSPEC to default type
 		if initConfig.RoutingProtocol == unix.RTN_UNSPEC {
@@ -283,6 +313,15 @@ var kubeVipManager = &cobra.Command{
 		// Set the logging level for all subsequent functions
 		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 {
+				log.Error("checking subnet exists if vip_address defined", "err", err)
+				return
+			}
+		}
+
 		// Ensure there is an address to generate the CIDR from
 		if initConfig.VIPSubnet == "" && initConfig.Address != "" {
 			initConfig.VIPSubnet, err = GenerateCidrRange(initConfig.Address, initConfig.DNSMode)
@@ -295,10 +334,19 @@ var kubeVipManager = &cobra.Command{
 		// Welome messages
 		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,
+				})
 			})
 		}
 
@@ -382,13 +430,13 @@ var kubeVipManager = &cobra.Command{
 				initConfig.Interface = defaultIF.Name
 				log.Info("kube-vip bind", "interface", initConfig.Interface)
 
-				go func() {
-					if err := vip.MonitorDefaultInterface(context.TODO(), defaultIF); err != nil {
+				wg.Go(func() {
+					if err := vip.MonitorDefaultInterface(ctx, defaultIF); err != nil {
 
 						log.Error("interface monitor", "err", err)
 						return
 					}
-				}()
+				})
 			}
 		}
 		// Perform a check on the state of the interface
@@ -404,16 +452,17 @@ var kubeVipManager = &cobra.Command{
 		}
 
 		// Define the new service manager
-		mgr, err := manager.New(configMap, &initConfig)
+		mgr, err := manager.New(ctx, configMap, &initConfig)
 		if err != nil {
 			log.Error("new manager", "err", err)
 			return
 		}
 
-		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.Error("start manager", "err", err)
 			return
@@ -447,19 +496,20 @@ func servePrometheusHTTPServer(ctx context.Context, config PrometheusHTTPServerC
 		ReadHeaderTimeout: 2 * time.Second,
 	}
 
-	go func() {
+	wg := sync.WaitGroup{}
+
+	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.Info("prometheus HTTP server stopped")
-
+	// create prometheus shutdown context (independent of other contexts)
 	ctxShutDown, cancel := context.WithTimeout(context.Background(), 5*time.Second)
 	defer func() {
 		cancel()
@@ -473,6 +523,10 @@ func servePrometheusHTTPServer(ctx context.Context, config PrometheusHTTPServerC
 	if err == http.ErrServerClosed {
 		err = nil
 	}
+
+	log.Info("prometheus HTTP server stopped")
+
+	wg.Wait()
 }
 
 func GenerateCidrRange(address string, dnsMode string) (string, error) {
@@ -483,23 +537,23 @@ func GenerateCidrRange(address string, dnsMode string) (string, error) {
 		ip := net.ParseIP(a)
 		if ip == nil {
 			// we probably are a DNS name
-			ips, err := utils.LookupHost(a, dnsMode)
+			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, "32")
+					cidrs = append(cidrs, strconv.Itoa(vip.DefaultMaskIPv4))
 				} else {
-					cidrs = append(cidrs, "128")
+					cidrs = append(cidrs, strconv.Itoa(vip.DefaultMaskIPv6))
 				}
 			}
 		} else {
 			if ip.To4() != nil {
-				cidrs = append(cidrs, "32")
+				cidrs = append(cidrs, strconv.Itoa(vip.DefaultMaskIPv4))
 			} else {
-				cidrs = append(cidrs, "128")
+				cidrs = append(cidrs, strconv.Itoa(vip.DefaultMaskIPv6))
 			}
 		}
 	}

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()

go.mod

@@ -1,44 +1,45 @@
 module github.com/kube-vip/kube-vip
 
-go 1.24.4
+go 1.26.0
 
 require (
-	github.com/cloudflare/ipvs v0.11.0
+	github.com/cloudflare/ipvs v0.12.0
 	github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc
 	github.com/docker/docker v28.5.2+incompatible
-	github.com/florianl/go-conntrack v0.4.0
+	github.com/florianl/go-conntrack v0.7.0
 	github.com/google/go-cmp v0.7.0
-	github.com/google/go-containerregistry v0.20.6
+	github.com/google/go-containerregistry v0.21.6
 	github.com/google/nftables v0.3.0
 	github.com/gookit/slog v0.6.0
 	github.com/huin/goupnp v1.3.0
 	github.com/insomniacslk/dhcp v0.0.0-20241224095048-b56fa0d5f25d
 	github.com/jpillora/backoff v1.0.0
 	github.com/mdlayher/ndp v1.1.0
-	github.com/onsi/ginkgo/v2 v2.27.2
-	github.com/onsi/gomega v1.38.2
+	github.com/onsi/ginkgo/v2 v2.29.0
+	github.com/onsi/gomega v1.41.0
 	github.com/osrg/gobgp/v3 v3.37.0
 	github.com/pkg/errors v0.9.1
 	github.com/prometheus/client_golang v1.23.2
-	github.com/sirupsen/logrus v1.9.3
-	github.com/spf13/cobra v1.10.1
+	github.com/sirupsen/logrus v1.9.4
+	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 v3.6.6
-	go.etcd.io/etcd/client/pkg/v3 v3.6.6
-	go.etcd.io/etcd/client/v3 v3.6.6
-	go.uber.org/zap v1.27.1
+	go.etcd.io/etcd/api/v3 v3.6.11
+	go.etcd.io/etcd/client/pkg/v3 v3.6.11
+	go.etcd.io/etcd/client/v3 v3.6.11
+	go.uber.org/zap v1.28.0
 	golang.org/x/exp v0.0.0-20250103183323-7d7fa50e5329
-	golang.org/x/sync v0.18.0
-	golang.org/x/sys v0.38.0
+	golang.org/x/sync v0.20.0
+	golang.org/x/sys v0.45.0
 	golang.zx2c4.com/wireguard/wgctrl v0.0.0-20241231184526-a9ab2273dd10
-	google.golang.org/grpc v1.77.0
-	google.golang.org/protobuf v1.36.10
-	k8s.io/api v0.34.2
-	k8s.io/apimachinery v0.34.2
-	k8s.io/client-go v0.34.1
-	k8s.io/klog/v2 v2.130.1
-	sigs.k8s.io/kind v0.30.0
+	google.golang.org/grpc v1.81.1
+	google.golang.org/protobuf v1.36.12-0.20260120151049-f2248ac996af
+	gopkg.in/yaml.v3 v3.0.1
+	k8s.io/api v0.36.1
+	k8s.io/apimachinery v0.36.1
+	k8s.io/client-go v0.36.1
+	k8s.io/klog/v2 v2.140.0
+	sigs.k8s.io/kind v0.31.0
 	sigs.k8s.io/yaml v1.6.0
 )
 
@@ -51,16 +52,16 @@ require (
 	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 v22.5.0 // indirect
 	github.com/dgryski/go-farm v0.0.0-20240924180020-3414d57e47da // indirect
 	github.com/distribution/reference v0.6.0 // indirect
-	github.com/dlclark/regexp2 v1.11.4 // indirect
-	github.com/docker/go-connections v0.5.0 // 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 v3.12.2 // 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 v1.8.0 // indirect
@@ -74,12 +75,12 @@ require (
 	github.com/gogo/protobuf v1.3.2 // indirect
 	github.com/golang/protobuf v1.5.4 // indirect
 	github.com/google/gnostic-models v0.7.0 // indirect
-	github.com/google/pprof v0.0.0-20250403155104-27863c87afa6 // indirect
+	github.com/google/pprof v0.0.0-20260402051712-545e8a4df936 // indirect
 	github.com/google/uuid v1.6.0 // indirect
 	github.com/gookit/color v1.6.0 // indirect
 	github.com/gookit/goutil v0.7.1 // indirect
 	github.com/gookit/gsr v0.1.1 // indirect
-	github.com/grpc-ecosystem/grpc-gateway/v2 v2.27.1 // indirect
+	github.com/grpc-ecosystem/grpc-gateway/v2 v2.28.0 // indirect
 	github.com/hashicorp/hcl v1.0.0 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
 	github.com/josharian/intern v1.0.0 // indirect
@@ -90,15 +91,16 @@ require (
 	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 v1.7.3-0.20250113171957-fbb4dce95f42 // indirect
+	github.com/mdlayher/netlink v1.8.0 // indirect
 	github.com/mdlayher/packet v1.1.2 // indirect
 	github.com/mdlayher/socket v0.5.1 // indirect
 	github.com/mitchellh/mapstructure v1.5.0 // indirect
 	github.com/moby/docker-image-spec v1.3.1 // indirect
-	github.com/moby/sys/sequential v0.6.0 // 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 v1.0.3-0.20250322232337-35a7c28c31ee // indirect
+	github.com/morikuni/aec v1.1.0 // indirect
 	github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 // indirect
 	github.com/opencontainers/go-digest v1.0.0 // indirect
 	github.com/opencontainers/image-spec v1.1.1 // indirect
@@ -109,57 +111,45 @@ require (
 	github.com/prometheus/client_model v0.6.2 // indirect
 	github.com/prometheus/common v0.66.1 // indirect
 	github.com/prometheus/procfs v0.16.1 // indirect
-	github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec // indirect
 	github.com/sagikazarmark/locafero v0.6.0 // indirect
 	github.com/sagikazarmark/slog-shim v0.1.0 // indirect
 	github.com/sourcegraph/conc v0.3.0 // indirect
 	github.com/spf13/afero v1.11.0 // indirect
 	github.com/spf13/cast v1.7.1 // indirect
-	github.com/spf13/pflag v1.0.9 // indirect
+	github.com/spf13/pflag v1.0.10 // indirect
 	github.com/spf13/viper v1.19.0 // indirect
 	github.com/subosito/gotenv v1.6.0 // indirect
-	github.com/tj/go-spin v1.1.0 // 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/xlab/c-for-go v1.3.0 // indirect
-	github.com/xlab/pkgconfig v0.0.0-20170226114623-cea12a0fd245 // 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 v1.38.0 // indirect
-	go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracehttp v1.37.0 // indirect
-	go.opentelemetry.io/otel/metric v1.38.0 // indirect
-	go.opentelemetry.io/otel/trace v1.38.0 // indirect
+	go.opentelemetry.io/otel v1.43.0 // indirect
+	go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracehttp v1.43.0 // indirect
+	go.opentelemetry.io/otel/metric v1.43.0 // indirect
+	go.opentelemetry.io/otel/trace v1.43.0 // indirect
 	go.uber.org/multierr v1.11.0 // indirect
-	go.yaml.in/yaml/v2 v2.4.2 // indirect
+	go.yaml.in/yaml/v2 v2.4.3 // indirect
 	go.yaml.in/yaml/v3 v3.0.4 // indirect
-	golang.org/x/crypto v0.45.0 // indirect
-	golang.org/x/mod v0.29.0 // indirect
-	golang.org/x/net v0.47.0 // indirect
-	golang.org/x/oauth2 v0.32.0 // indirect
-	golang.org/x/term v0.37.0 // indirect
-	golang.org/x/text v0.31.0 // indirect
-	golang.org/x/time v0.9.0 // indirect
-	golang.org/x/tools v0.38.0 // indirect
+	golang.org/x/crypto v0.51.0 // indirect
+	golang.org/x/mod v0.36.0 // indirect
+	golang.org/x/net v0.54.0 // indirect
+	golang.org/x/oauth2 v0.36.0 // indirect
+	golang.org/x/term v0.43.0 // indirect
+	golang.org/x/text v0.37.0 // indirect
+	golang.org/x/time v0.14.0 // indirect
+	golang.org/x/tools v0.45.0 // indirect
 	golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173 // indirect
-	google.golang.org/genproto/googleapis/api v0.0.0-20251022142026-3a174f9686a8 // indirect
-	google.golang.org/genproto/googleapis/rpc v0.0.0-20251022142026-3a174f9686a8 // indirect
-	gopkg.in/evanphx/json-patch.v4 v4.12.0 // indirect
+	google.golang.org/genproto/googleapis/api v0.0.0-20260401024825-9d38bb4040a9 // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20260401024825-9d38bb4040a9 // 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.v2 v2.4.0 // indirect
-	gopkg.in/yaml.v3 v3.0.1 // indirect
-	k8s.io/kube-openapi v0.0.0-20250710124328-f3f2b991d03b // indirect
-	k8s.io/utils v0.0.0-20250604170112-4c0f3b243397 // indirect
-	modernc.org/cc/v4 v4.24.4 // indirect
-	modernc.org/mathutil v1.7.1 // indirect
-	modernc.org/opt v0.1.4 // indirect
-	modernc.org/sortutil v1.2.1 // indirect
-	modernc.org/strutil v1.2.1 // indirect
-	modernc.org/token v1.1.0 // indirect
-	sigs.k8s.io/json v0.0.0-20241014173422-cfa47c3a1cc8 // 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 v6.3.0 // indirect
+	sigs.k8s.io/structured-merge-diff/v6 v6.3.2 // indirect
 )

go.sum

@@ -2,8 +2,6 @@ al.essio.dev/pkg/shellescape v1.5.1 h1:86HrALUujYS/h+GtqoB26SBEdkWfmMI6FubjXlsXy
 al.essio.dev/pkg/shellescape v1.5.1/go.mod h1:6sIqp7X2P6mThCQ7twERpZTuigpr6KbZWtls1U8I890=
 github.com/Azure/go-ansiterm v0.0.0-20250102033503-faa5f7b0171c h1:udKWzYgxTojEKWjV8V+WSxDXJ4NFATAsZjh8iIbsQIg=
 github.com/Azure/go-ansiterm v0.0.0-20250102033503-faa5f7b0171c/go.mod h1:xomTg63KZ2rFqZQzSB4Vz2SUXa1BpHTVz9L5PTmPC4E=
-github.com/BurntSushi/toml v0.3.1/go.mod h1:xHWCNGjB5oqiDr8zfno3MHue2Ht5sIBksp03qcyfWMU=
-github.com/BurntSushi/toml v0.4.1/go.mod h1:CxXYINrC8qIiEnFrOxCa7Jy5BFHlXnUU2pbicEuybxQ=
 github.com/BurntSushi/toml v1.4.0 h1:kuoIxZQy2WRRk1pttg9asf+WVv6tWQuBNVmK8+nqPr0=
 github.com/BurntSushi/toml v1.4.0/go.mod h1:ukJfTF/6rtPPRCnwkur4qwRxa8vTRFBF0uk2lLoLwho=
 github.com/Masterminds/semver/v3 v3.4.0 h1:Zog+i5UMtVoCU8oKka5P7i9q9HgrJeGzI9SA1Xbatp0=
@@ -12,14 +10,12 @@ github.com/Microsoft/go-winio v0.6.2 h1:F2VQgta7ecxGYO8k3ZZz3RS8fVIXVxONVUPlNERo
 github.com/Microsoft/go-winio v0.6.2/go.mod h1:yd8OoFMLzJbo9gZq8j5qaps8bJ9aShtEA8Ipt1oGCvU=
 github.com/beorn7/perks v1.0.1 h1:VlbKKnNfV8bJzeqoa4cOKqO6bYr3WgKZxO8Z16+hsOM=
 github.com/beorn7/perks v1.0.1/go.mod h1:G2ZrVWU2WbWT9wwq4/hrbKbnv/1ERSJQ0ibhJ6rlkpw=
-github.com/cenkalti/backoff/v5 v5.0.2 h1:rIfFVxEf1QsI7E1ZHfp/B4DF/6QBAUhmgkxc0H7Zss8=
-github.com/cenkalti/backoff/v5 v5.0.2/go.mod h1:rkhZdG3JZukswDf7f0cwqPNk4K0sa+F97BxZthm/crw=
+github.com/cenkalti/backoff/v5 v5.0.3 h1:ZN+IMa753KfX5hd8vVaMixjnqRZ3y8CuJKRKj1xcsSM=
+github.com/cenkalti/backoff/v5 v5.0.3/go.mod h1:rkhZdG3JZukswDf7f0cwqPNk4K0sa+F97BxZthm/crw=
 github.com/cespare/xxhash/v2 v2.3.0 h1:UL815xU9SqsFlibzuggzjXhog7bL6oX9BbNZnL2UFvs=
 github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
-github.com/cilium/ebpf v0.5.0/go.mod h1:4tRaxcgiL706VnOzHOdBlY8IEAIdxINsQBcU4xJJXRs=
-github.com/cilium/ebpf v0.7.0/go.mod h1:/oI2+1shJiTGAMgl6/RgJr36Eo1jzrRcAWbcXO2usCA=
-github.com/cloudflare/ipvs v0.11.0 h1:niLcbqfv8+RSYk+yI+jhiCIHobEdEhO++mcwCX0Znsw=
-github.com/cloudflare/ipvs v0.11.0/go.mod h1:XsbuKcQpqb3rpjYPtsFsM8BNZlgLw+Z0iC6RhocqDd8=
+github.com/cloudflare/ipvs v0.12.0 h1:UeKRM4q82F+XQjw1sGRwsH3IOy6keBYUwlqTUqJhy7Y=
+github.com/cloudflare/ipvs v0.12.0/go.mod h1:SvzXyDX2E33X/Cokbgts7rFXueiSqp6NlEMGDT7Bu24=
 github.com/containerd/errdefs v1.0.0 h1:tg5yIfIlQIrxYtu9ajqY42W3lpS19XqdxRQeEwYG8PI=
 github.com/containerd/errdefs v1.0.0/go.mod h1:+YBYIdtsnF4Iw6nWZhJcqGSg/dwvV7tyJ/kCkyJ2k+M=
 github.com/containerd/errdefs/pkg v0.3.0 h1:9IKJ06FvyNlexW690DXuQNx2KA2cUJXx151Xdx3ZPPE=
@@ -39,29 +35,24 @@ github.com/dgryski/go-farm v0.0.0-20240924180020-3414d57e47da h1:aIftn67I1fkbMa5
 github.com/dgryski/go-farm v0.0.0-20240924180020-3414d57e47da/go.mod h1:SqUrOPUnsFjfmXRMNPybcSiG0BgUW2AuFH8PAnS2iTw=
 github.com/distribution/reference v0.6.0 h1:0IXCQ5g4/QMHHkarYzh5l+u8T3t73zM5QvfrDyIgxBk=
 github.com/distribution/reference v0.6.0/go.mod h1:BbU0aIcezP1/5jX/8MP0YiH4SdvB5Y4f/wlDRiLyi3E=
-github.com/dlclark/regexp2 v1.11.4 h1:rPYF9/LECdNymJufQKmri9gV604RvvABwgOA8un7yAo=
-github.com/dlclark/regexp2 v1.11.4/go.mod h1:DHkYz0B9wPfa6wondMfaivmHpzrQ3v9q8cnmRbL6yW8=
 github.com/docker/docker v28.5.2+incompatible h1:DBX0Y0zAjZbSrm1uzOkdr1onVghKaftjlSWt4AFexzM=
 github.com/docker/docker v28.5.2+incompatible/go.mod h1:eEKB0N0r5NX/I1kEveEz05bcu8tLC/8azJZsviup8Sk=
-github.com/docker/go-connections v0.5.0 h1:USnMq7hx7gwdVZq1L49hLXaFtUdTADjXGp+uj1Br63c=
-github.com/docker/go-connections v0.5.0/go.mod h1:ov60Kzw0kKElRwhNs9UlUHAE/F9Fe6GLaXnqyDdmEXc=
+github.com/docker/go-connections v0.7.0 h1:6SsRfJddP22WMrCkj19x9WKjEDTB+ahsdiGYf0mN39c=
+github.com/docker/go-connections v0.7.0/go.mod h1:no1qkHdjq7kLMGUXYAduOhYPSJxxvgWBh7ogVvptn3Q=
 github.com/docker/go-units v0.5.0 h1:69rxXcBk27SvSaaxTtLh/8llcHD8vYHT7WSdRZ/jvr4=
 github.com/docker/go-units v0.5.0/go.mod h1:fgPhTUdO+D/Jk86RDLlptpiXQzgHJF7gydDDbaIK4Dk=
-github.com/dustin/go-humanize v1.0.1 h1:GzkhY7T5VNhEkwH0PVJgjz+fX1rhBrR7pRT3mDkpeCY=
-github.com/dustin/go-humanize v1.0.1/go.mod h1:Mu1zIs6XwVuF/gI1OepvI0qD18qycQx+mFykh5fBlto=
 github.com/eapache/channels v1.1.0 h1:F1taHcn7/F0i8DYqKXJnyhJcVpp2kgFcNePxXtnyu4k=
 github.com/eapache/channels v1.1.0/go.mod h1:jMm2qB5Ubtg9zLd+inMZd2/NUvXgzmWXsDaLyQIGfH0=
 github.com/eapache/queue v1.1.0 h1:YOEu7KNc61ntiQlcEeUIoDTJ2o8mQznoNvUhiigpIqc=
 github.com/eapache/queue v1.1.0/go.mod h1:6eCeP0CKFpHLu8blIFXhExK/dRa7WDZfr6jVFPTqq+I=
-github.com/emicklei/go-restful/v3 v3.12.2 h1:DhwDP0vY3k8ZzE0RunuJy8GhNpPL6zqLkDf9B/a0/xU=
-github.com/emicklei/go-restful/v3 v3.12.2/go.mod h1:6n3XBCmQQb25CM2LCACGz8ukIrRry+4bhvbpWn3mrbc=
+github.com/emicklei/go-restful/v3 v3.13.0 h1:C4Bl2xDndpU6nJ4bc1jXd+uTmYPVUwkD6bFY/oTyCes=
+github.com/emicklei/go-restful/v3 v3.13.0/go.mod h1:6n3XBCmQQb25CM2LCACGz8ukIrRry+4bhvbpWn3mrbc=
 github.com/evanphx/json-patch/v5 v5.9.0 h1:kcBlZQbplgElYIlo/n1hJbls2z/1awpXxpRi0/FOJfg=
 github.com/evanphx/json-patch/v5 v5.9.0/go.mod h1:VNkHZ/282BpEyt/tObQO8s5CMPmYYq14uClGH4abBuQ=
 github.com/felixge/httpsnoop v1.0.4 h1:NFTV2Zj1bL4mc9sqWACXbQFVBBg2W3GPvqp8/ESS2Wg=
 github.com/felixge/httpsnoop v1.0.4/go.mod h1:m8KPJKqk1gH5J9DgRY2ASl2lWCfGKXixSwevea8zH2U=
-github.com/florianl/go-conntrack v0.4.0 h1:TlYkxytdwgVayfU0cKwkHurQA0Rd1ZSEBRckRYDUu18=
-github.com/florianl/go-conntrack v0.4.0/go.mod h1:iPDx4oIats2T7X7Jm3PFyRCJM1GfZhJaSHOWROYOrE8=
-github.com/frankban/quicktest v1.11.3/go.mod h1:wRf/ReqHper53s+kmmSZizM8NamnL3IM0I9ntUbOk+k=
+github.com/florianl/go-conntrack v0.7.0 h1:kWbRhLRUqBlmi2ncASmygN5yI281wlpnTUJs69tcs7g=
+github.com/florianl/go-conntrack v0.7.0/go.mod h1:7kzfTjtQCRsAVkuornpI8KiwFfa79QmcZXZQex1IXhU=
 github.com/frankban/quicktest v1.14.6 h1:7Xjx+VpznH+oBnejlPUj8oUpdxnVs4f8XU8WnHkI4W8=
 github.com/frankban/quicktest v1.14.6/go.mod h1:4ptaffx2x8+WTWXmUCuVU6aPUX1/Mz7zb5vbUoiM6w0=
 github.com/fsnotify/fsnotify v1.8.0 h1:dAwr6QBTBZIkG8roQaJjGof0pp0EeF+tNV7YBP3F/8M=
@@ -98,22 +89,17 @@ github.com/golang/protobuf v1.5.4 h1:i7eJL8qZTpSEXOPTxNKhASYpMn+8e5Q6AdndVa1dWek
 github.com/golang/protobuf v1.5.4/go.mod h1:lnTiLA8Wa4RWRcIUkrtSVa5nRhsEGBg48fD6rSs7xps=
 github.com/google/gnostic-models v0.7.0 h1:qwTtogB15McXDaNqTZdzPJRHvaVJlAl+HVQnLmJEJxo=
 github.com/google/gnostic-models v0.7.0/go.mod h1:whL5G0m6dmc5cPxKc5bdKdEN3UjI7OUGxBlw57miDrQ=
-github.com/google/go-cmp v0.2.0/go.mod h1:oXzfMopK8JAjlY9xF4vHSVASa0yLyX7SntLO5aqRK0M=
-github.com/google/go-cmp v0.3.1/go.mod h1:8QqcDgzrUqlUb/G2PQTWiueGozuR1884gddMywk6iLU=
-github.com/google/go-cmp v0.4.0/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
-github.com/google/go-cmp v0.5.2/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
-github.com/google/go-cmp v0.5.4/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
-github.com/google/go-cmp v0.5.5/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
-github.com/google/go-cmp v0.5.6/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
+github.com/google/go-cmp v0.5.7/go.mod h1:n+brtR0CgQNWTVd5ZUFpTBC8YFBDLK/h/bpaJ8/DtOE=
+github.com/google/go-cmp v0.5.9/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
 github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
 github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
-github.com/google/go-containerregistry v0.20.6 h1:cvWX87UxxLgaH76b4hIvya6Dzz9qHB31qAwjAohdSTU=
-github.com/google/go-containerregistry v0.20.6/go.mod h1:T0x8MuoAoKX/873bkeSfLD2FAkwCDf9/HZgsFJ02E2Y=
+github.com/google/go-containerregistry v0.21.6 h1:T+yqQIlJXKrM98Om4DlW3GoWQAmhZuLMwoDOvVrtiUM=
+github.com/google/go-containerregistry v0.21.6/go.mod h1:U7MMSBIJynke2MVQrQk19NP9k/uQsGz/h0amIFSHMbo=
 github.com/google/gofuzz v1.0.0/go.mod h1:dBl0BpW6vV/+mYPU4Po3pmUjxk6FQPldtuIdl/M65Eg=
 github.com/google/nftables v0.3.0 h1:bkyZ0cbpVeMHXOrtlFc8ISmfVqq5gPJukoYieyVmITg=
 github.com/google/nftables v0.3.0/go.mod h1:BCp9FsrbF1Fn/Yu6CLUc9GGZFw/+hsxfluNXXmxBfRM=
-github.com/google/pprof v0.0.0-20250403155104-27863c87afa6 h1:BHT72Gu3keYf3ZEu2J0b1vyeLSOYI8bm5wbJM/8yDe8=
-github.com/google/pprof v0.0.0-20250403155104-27863c87afa6/go.mod h1:boTsfXsheKC2y+lKOCMpSfarhxDeIzfZG1jqGcPl3cA=
+github.com/google/pprof v0.0.0-20260402051712-545e8a4df936 h1:EwtI+Al+DeppwYX2oXJCETMO23COyaKGP6fHVpkpWpg=
+github.com/google/pprof v0.0.0-20260402051712-545e8a4df936/go.mod h1:MxpfABSjhmINe3F1It9d+8exIHFvUqtLIRCdOGNXqiI=
 github.com/google/shlex v0.0.0-20191202100458-e7afc7fbc510 h1:El6M4kTTCOh6aBiKaUGG7oYTSPP8MxqL4YI3kZKwcP4=
 github.com/google/shlex v0.0.0-20191202100458-e7afc7fbc510/go.mod h1:pupxD2MaaD3pAXIBCelhxNneeOaAeabZDe5s4K6zSpQ=
 github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
@@ -128,8 +114,8 @@ github.com/gookit/gsr v0.1.1 h1:TaHD3M7qa6lcAf9D2J4mGNg+QjgDtD1bw7uctF8RXOM=
 github.com/gookit/gsr v0.1.1/go.mod h1:7wv4Y4WCnil8+DlDYHBjidzrEzfHhXEoFjEA0pPPWpI=
 github.com/gookit/slog v0.6.0 h1:KEQxOJxbTtk7oyqah6nJOEKjOdI0z5qoqkX7I6G65g4=
 github.com/gookit/slog v0.6.0/go.mod h1:hPlpNi/WIcGmkEjHzQTS7s5JZkHmmnGy9sYo6csa08s=
-github.com/grpc-ecosystem/grpc-gateway/v2 v2.27.1 h1:X5VWvz21y3gzm9Nw/kaUeku/1+uBhcekkmy4IkffJww=
-github.com/grpc-ecosystem/grpc-gateway/v2 v2.27.1/go.mod h1:Zanoh4+gvIgluNqcfMVTJueD4wSS5hT7zTt4Mrutd90=
+github.com/grpc-ecosystem/grpc-gateway/v2 v2.28.0 h1:HWRh5R2+9EifMyIHV7ZV+MIZqgz+PMpZ14Jynv3O2Zs=
+github.com/grpc-ecosystem/grpc-gateway/v2 v2.28.0/go.mod h1:JfhWUomR1baixubs02l85lZYYOm7LV6om4ceouMv45c=
 github.com/hashicorp/hcl v1.0.0 h1:0Anlzjpi4vEasTeNFn2mLJgTSwt0+6sfsiTG8qcWGx4=
 github.com/hashicorp/hcl v1.0.0/go.mod h1:E5yfLk+7swimpb2L/Alb/PJmXilQ/rhwaUYs4T20WEQ=
 github.com/hugelgupf/socketpair v0.0.0-20190730060125-05d35a94e714 h1:/jC7qQFrv8CrSJVmaolDVOxTfS9kc36uB6H40kdbQq8=
@@ -142,35 +128,23 @@ github.com/insomniacslk/dhcp v0.0.0-20241224095048-b56fa0d5f25d h1:VkCNWh6tuQLgD
 github.com/insomniacslk/dhcp v0.0.0-20241224095048-b56fa0d5f25d/go.mod h1:VvGYjkZoJyKqlmT1yzakUs4mfKMNB0XdODP0+rdml6k=
 github.com/josharian/intern v1.0.0 h1:vlS4z54oSdjm0bgjRigI+G1HpF+tI+9rE5LLzOg8HmY=
 github.com/josharian/intern v1.0.0/go.mod h1:5DoeVV0s6jJacbCEi61lwdGj/aVlrQvzHFFd8Hwg//Y=
-github.com/josharian/native v0.0.0-20200817173448-b6b71def0850/go.mod h1:7X/raswPFr05uY3HiLlYeyQntB6OO7E/d2Cu7qoaN2w=
+github.com/josharian/native v1.0.0/go.mod h1:7X/raswPFr05uY3HiLlYeyQntB6OO7E/d2Cu7qoaN2w=
 github.com/josharian/native v1.1.0 h1:uuaP0hAbW7Y4l0ZRQ6C9zfb7Mg1mbFKry/xzDAfmtLA=
 github.com/josharian/native v1.1.0/go.mod h1:7X/raswPFr05uY3HiLlYeyQntB6OO7E/d2Cu7qoaN2w=
 github.com/joshdk/go-junit v1.0.0 h1:S86cUKIdwBHWwA6xCmFlf3RTLfVXYQfvanM5Uh+K6GE=
 github.com/joshdk/go-junit v1.0.0/go.mod h1:TiiV0PqkaNfFXjEiyjWM3XXrhVyCa1K4Zfga6W52ung=
 github.com/jpillora/backoff v1.0.0 h1:uvFg412JmmHBHw7iwprIxkPMI+sGQ4kzOWsMeHnm2EA=
 github.com/jpillora/backoff v1.0.0/go.mod h1:J/6gKK9jxlEcS3zixgDgUAsiuZ7yrSoa/FX5e0EB2j4=
-github.com/jsimonetti/rtnetlink v0.0.0-20190606172950-9527aa82566a/go.mod h1:Oz+70psSo5OFh8DBl0Zv2ACw7Esh6pPUphlvZG9x7uw=
-github.com/jsimonetti/rtnetlink v0.0.0-20200117123717-f846d4f6c1f4/go.mod h1:WGuG/smIU4J/54PblvSbh+xvCZmpJnFgr3ds6Z55XMQ=
-github.com/jsimonetti/rtnetlink v0.0.0-20201009170750-9c6f07d100c1/go.mod h1:hqoO/u39cqLeBLebZ8fWdE96O7FxrAsRYhnVOdgHxok=
-github.com/jsimonetti/rtnetlink v0.0.0-20201216134343-bde56ed16391/go.mod h1:cR77jAZG3Y3bsb8hF6fHJbFoyFukLFOkQ98S0pQz3xw=
-github.com/jsimonetti/rtnetlink v0.0.0-20201220180245-69540ac93943/go.mod h1:z4c53zj6Eex712ROyh8WI0ihysb5j2ROyV42iNogmAs=
-github.com/jsimonetti/rtnetlink v0.0.0-20210122163228-8d122574c736/go.mod h1:ZXpIyOK59ZnN7J0BV99cZUPmsqDRZ3eq5X+st7u/oSA=
-github.com/jsimonetti/rtnetlink v0.0.0-20210212075122-66c871082f2b/go.mod h1:8w9Rh8m+aHZIG69YPGGem1i5VzoyRC8nw2kA8B+ik5U=
-github.com/jsimonetti/rtnetlink v0.0.0-20210525051524-4cc836578190/go.mod h1:NmKSdU4VGSiv1bMsdqNALI4RSvvjtz65tTMCnD05qLo=
-github.com/jsimonetti/rtnetlink v0.0.0-20211022192332-93da33804786/go.mod h1:v4hqbTdfQngbVSZJVWUhGE/lbTFf9jb+ygmNUDQMuOs=
 github.com/json-iterator/go v1.1.12 h1:PV8peI4a0ysnczrg+LtxykD8LfKY9ML6u2jnxaEnrnM=
 github.com/json-iterator/go v1.1.12/go.mod h1:e30LSqwooZae/UwlEbR2852Gd8hjQvJoHmT4TnhNGBo=
 github.com/k-sone/critbitgo v1.4.0 h1:l71cTyBGeh6X5ATh6Fibgw3+rtNT80BA0uNNWgkPrbE=
 github.com/k-sone/critbitgo v1.4.0/go.mod h1:7E6pyoyADnFxlUBEKcnfS49b7SUAQGMK+OAp/UQvo0s=
 github.com/kisielk/errcheck v1.5.0/go.mod h1:pFxgyoBC7bSaBwPgfKdkLd5X25qrDl4LWUI2bnpBCr8=
 github.com/kisielk/gotool v1.0.0/go.mod h1:XhKaO+MFFWcvkIS/tQcRk01m1F5IRFswLeQ+oQHNcck=
-github.com/klauspost/compress v1.18.0 h1:c/Cqfb0r+Yi+JtIEq73FWXVkRonBlf0CRNYc8Zttxdo=
-github.com/klauspost/compress v1.18.0/go.mod h1:2Pp+KzxcywXVXMr50+X0Q/Lsb43OQHYWRCY2AiWywWQ=
-github.com/kr/pretty v0.2.1/go.mod h1:ipq/a2n7PKx3OHsz4KJII5eveXtPO4qwEXGdVfWzfnI=
+github.com/klauspost/compress v1.18.6 h1:2jupLlAwFm95+YDR+NwD2MEfFO9d4z4Prjl1XXDjuao=
+github.com/klauspost/compress v1.18.6/go.mod h1:cwPg85FWrGar70rWktvGQj8/hthj3wpl0PGDogxkrSQ=
 github.com/kr/pretty v0.3.1 h1:flRD4NNwYAUpkphVc1HcthR4KEIFJ65n8Mw5qdRn3LE=
 github.com/kr/pretty v0.3.1/go.mod h1:hoEshYVHaxMs3cyo3Yncou5ZscifuDolrwPKZanG3xk=
-github.com/kr/pty v1.1.1/go.mod h1:pFQYn66WHrOpPYNljwOMqo10TkYh1fy3cYio2l3bCsQ=
-github.com/kr/text v0.1.0/go.mod h1:4Jbv+DJW3UT/LiOwJeYQe1efqtUx/iVham/4vfdArNI=
 github.com/kr/text v0.2.0 h1:5Nx0Ya0ZqY2ygV366QzturHI13Jq95ApcVaJBhpS+AY=
 github.com/kr/text v0.2.0/go.mod h1:eLer722TekiGuMkidMxC/pM04lWEeraHUUmBw8l2grE=
 github.com/kylelemons/godebug v1.1.0 h1:RPNrshWIDI6G2gRW9EHilWtl7Z6Sb1BR0xunSBf0SNc=
@@ -183,31 +157,16 @@ github.com/maruel/natural v1.1.1 h1:Hja7XhhmvEFhcByqDoHz9QZbkWey+COd9xWfCfn1ioo=
 github.com/maruel/natural v1.1.1/go.mod h1:v+Rfd79xlw1AgVBjbO0BEQmptqb5HvL/k9GRHB7ZKEg=
 github.com/mattn/go-isatty v0.0.20 h1:xfD0iDuEKnDkl03q4limB+vH+GxLEtL/jb4xVJSWWEY=
 github.com/mattn/go-isatty v0.0.20/go.mod h1:W+V8PltTTMOvKvAeJH7IuucS94S2C6jfK/D7dTCTo3Y=
-github.com/mdlayher/ethtool v0.0.0-20210210192532-2b88debcdd43/go.mod h1:+t7E0lkKfbBsebllff1xdTmyJt8lH37niI6kwFk9OTo=
-github.com/mdlayher/ethtool v0.0.0-20211028163843-288d040e9d60/go.mod h1:aYbhishWc4Ai3I2U4Gaa2n3kHWSwzme6EsG/46HRQbE=
-github.com/mdlayher/genetlink v1.0.0/go.mod h1:0rJ0h4itni50A86M2kHcgS85ttZazNt7a8H2a2cw0Gc=
 github.com/mdlayher/genetlink v1.3.2 h1:KdrNKe+CTu+IbZnm/GVUMXSqBBLqcGpRDa0xkQy56gw=
 github.com/mdlayher/genetlink v1.3.2/go.mod h1:tcC3pkCrPUGIKKsCsp0B3AdaaKuHtaxoJRz3cc+528o=
 github.com/mdlayher/ndp v1.1.0 h1:QylGKGVtH60sKZUE88+IW5ila1Z/M9/OXhWdsVKuscs=
 github.com/mdlayher/ndp v1.1.0/go.mod h1:FmgESgemgjl38vuOIyAHWUUL6vQKA/pQNkvXdWsdQFM=
-github.com/mdlayher/netlink v0.0.0-20190409211403-11939a169225/go.mod h1:eQB3mZE4aiYnlUsyGGCOpPETfdQq4Jhsgf1fk3cwQaA=
-github.com/mdlayher/netlink v1.0.0/go.mod h1:KxeJAFOFLG6AjpyDkQ/iIhxygIUKD+vcwqcnu43w/+M=
-github.com/mdlayher/netlink v1.1.0/go.mod h1:H4WCitaheIsdF9yOYu8CFmCgQthAPIWZmcKp9uZHgmY=
-github.com/mdlayher/netlink v1.1.1/go.mod h1:WTYpFb/WTvlRJAyKhZL5/uy69TDDpHHu2VZmb2XgV7o=
-github.com/mdlayher/netlink v1.2.0/go.mod h1:kwVW1io0AZy9A1E2YYgaD4Cj+C+GPkU6klXCMzIJ9p8=
-github.com/mdlayher/netlink v1.2.1/go.mod h1:bacnNlfhqHqqLo4WsYeXSqfyXkInQ9JneWI68v1KwSU=
-github.com/mdlayher/netlink v1.2.2-0.20210123213345-5cc92139ae3e/go.mod h1:bacnNlfhqHqqLo4WsYeXSqfyXkInQ9JneWI68v1KwSU=
-github.com/mdlayher/netlink v1.3.0/go.mod h1:xK/BssKuwcRXHrtN04UBkwQ6dY9VviGGuriDdoPSWys=
-github.com/mdlayher/netlink v1.4.0/go.mod h1:dRJi5IABcZpBD2A3D0Mv/AiX8I9uDEu5oGkAVrekmf8=
-github.com/mdlayher/netlink v1.4.1/go.mod h1:e4/KuJ+s8UhfUpO9z00/fDZZmhSrs+oxyqAS9cNgn6Q=
-github.com/mdlayher/netlink v1.5.0/go.mod h1:1Kr8BBFxGyUyNmztC9WLOayqYVAd2wsgOZm18nqGuzQ=
-github.com/mdlayher/netlink v1.7.3-0.20250113171957-fbb4dce95f42 h1:A1Cq6Ysb0GM0tpKMbdCXCIfBclan4oHk1Jb+Hrejirg=
-github.com/mdlayher/netlink v1.7.3-0.20250113171957-fbb4dce95f42/go.mod h1:BB4YCPDOzfy7FniQ/lxuYQ3dgmM2cZumHbK8RpTjN2o=
+github.com/mdlayher/netlink v1.6.2/go.mod h1:O1HXX2sIWSMJ3Qn1BYZk1yZM+7iMki/uYGGiwGyq/iU=
+github.com/mdlayher/netlink v1.8.0 h1:e7XNIYJKD7hUct3Px04RuIGJbBxy1/c4nX7D5YyvvlM=
+github.com/mdlayher/netlink v1.8.0/go.mod h1:UhgKXUlDQhzb09DrCl2GuRNEglHmhYoWAHid9HK3594=
 github.com/mdlayher/packet v1.1.2 h1:3Up1NG6LZrsgDVn6X4L9Ge/iyRyxFEFD9o6Pr3Q1nQY=
 github.com/mdlayher/packet v1.1.2/go.mod h1:GEu1+n9sG5VtiRE4SydOmX5GTwyyYlteZiFU+x0kew4=
-github.com/mdlayher/socket v0.0.0-20210307095302-262dc9984e00/go.mod h1:GAFlyu4/XV68LkQKYzKhIo/WW7j3Zi0YRAz/BOoanUc=
-github.com/mdlayher/socket v0.0.0-20211007213009-516dcbdf0267/go.mod h1:nFZ1EtZYK8Gi/k6QNu7z7CgO20i/4ExeQswwWuPmG/g=
-github.com/mdlayher/socket v0.1.0/go.mod h1:mYV5YIZAfHh4dzDVzI8x8tWLWCliuX8Mon5Awbj+qDs=
+github.com/mdlayher/socket v0.2.3/go.mod h1:bz12/FozYNH/VbvC3q7TRIK/Y6dH1kCKsXaUeXi/FmY=
 github.com/mdlayher/socket v0.5.1 h1:VZaqt6RkGkt2OE9l3GcC6nZkqD3xKeQLyfleW/uBcos=
 github.com/mdlayher/socket v0.5.1/go.mod h1:TjPLHI1UgwEv5J1B5q0zTZq12A/6H7nKmtTanQE37IQ=
 github.com/mfridman/tparse v0.18.0 h1:wh6dzOKaIwkUGyKgOntDW4liXSo37qg5AXbIhkMV3vE=
@@ -230,22 +189,20 @@ github.com/modern-go/concurrent v0.0.0-20180306012644-bacd9c7ef1dd/go.mod h1:6dJ
 github.com/modern-go/reflect2 v1.0.2/go.mod h1:yWuevngMOJpCy52FWWMvUC8ws7m/LJsjYzDa0/r8luk=
 github.com/modern-go/reflect2 v1.0.3-0.20250322232337-35a7c28c31ee h1:W5t00kpgFdJifH4BDsTlE89Zl93FEloxaWZfGcifgq8=
 github.com/modern-go/reflect2 v1.0.3-0.20250322232337-35a7c28c31ee/go.mod h1:yWuevngMOJpCy52FWWMvUC8ws7m/LJsjYzDa0/r8luk=
-github.com/morikuni/aec v1.0.0 h1:nP9CBfwrvYnBRgY6qfDQkygYDmYwOilePFkwzv4dU8A=
-github.com/morikuni/aec v1.0.0/go.mod h1:BbKIizmSmc5MMPqRYbxO4ZU0S0+P200+tUnFx7PXmsc=
+github.com/morikuni/aec v1.1.0 h1:vBBl0pUnvi/Je71dsRrhMBtreIqNMYErSAbEeb8jrXQ=
+github.com/morikuni/aec v1.1.0/go.mod h1:xDRgiq/iw5l+zkao76YTKzKttOp2cwPEne25HDkJnBw=
 github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 h1:C3w9PqII01/Oq1c1nUAm88MOHcQC9l5mIlSMApZMrHA=
 github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822/go.mod h1:+n7T8mK8HuQTcFwEeznm/DIxMOiR9yIdICNftLE1DvQ=
-github.com/onsi/ginkgo/v2 v2.27.2 h1:LzwLj0b89qtIy6SSASkzlNvX6WktqurSHwkk2ipF/Ns=
-github.com/onsi/ginkgo/v2 v2.27.2/go.mod h1:ArE1D/XhNXBXCBkKOLkbsb2c81dQHCRcF5zwn/ykDRo=
-github.com/onsi/gomega v1.38.2 h1:eZCjf2xjZAqe+LeWvKb5weQ+NcPwX84kqJ0cZNxok2A=
-github.com/onsi/gomega v1.38.2/go.mod h1:W2MJcYxRGV63b418Ai34Ud0hEdTVXq9NW9+Sx6uXf3k=
+github.com/onsi/ginkgo/v2 v2.29.0 h1:rfh+ZFjgJhYWRoIqVf3Uwx/W20yLrcrE2h2GmYVRaag=
+github.com/onsi/ginkgo/v2 v2.29.0/go.mod h1:+aXOY+vzZ5mu2iI2HpTZUPmM//oQfsNFX6gU9kNcA44=
+github.com/onsi/gomega v1.41.0 h1:OwKp4pXNgVxf6sCplzYo794OFNuoL2q2SBMU5NSWOjA=
+github.com/onsi/gomega v1.41.0/go.mod h1:M/Uqpu/8qTjtzCLUA2zJHX9Iilrau25x1PdoSRbWh5A=
 github.com/opencontainers/go-digest v1.0.0 h1:apOUWs51W5PlhuyGyz9FCeeBIOUDA/6nW8Oi/yOhh5U=
 github.com/opencontainers/go-digest v1.0.0/go.mod h1:0JzlMkj0TRzQZfJkVvzbP0HBR3IKzErnv2BNG4W4MAM=
 github.com/opencontainers/image-spec v1.1.1 h1:y0fUlFfIZhPF1W537XOLg0/fcx6zcHCJwooC2xJA040=
 github.com/opencontainers/image-spec v1.1.1/go.mod h1:qpqAh3Dmcf36wStyyWU+kCeDgrGnAve2nCC8+7h8Q0M=
 github.com/osrg/gobgp/v3 v3.37.0 h1:+ObuOdvj7G7nxrT0fKFta+EAupdWf/q1WzbXydr8IOY=
 github.com/osrg/gobgp/v3 v3.37.0/go.mod h1:kVHVFy1/fyZHJ8P32+ctvPeJogn9qKwa1YCeMRXXrP0=
-github.com/pbnjay/memory v0.0.0-20210728143218-7b4eea64cf58 h1:onHthvaw9LFnH4t2DcNVpwGmV9E1BkGknEliJkfwQj0=
-github.com/pbnjay/memory v0.0.0-20210728143218-7b4eea64cf58/go.mod h1:DXv8WO4yhMYhSNPKjeNKa5WY9YCIEBRbNzFFPJbWO6Y=
 github.com/pelletier/go-toml v1.9.5 h1:4yBQzkHv+7BHq2PQUZF3Mx0IYxG7LsP222s7Agd3ve8=
 github.com/pelletier/go-toml v1.9.5/go.mod h1:u1nR/EPcESfeI/szUZKdtJ0xRNbUoANCkoOuaOx1Y+c=
 github.com/pelletier/go-toml/v2 v2.2.3 h1:YmeHyLY8mFWbdkNWwpr+qIL2bEqT0o95WSdkNHvL12M=
@@ -265,8 +222,6 @@ github.com/prometheus/common v0.66.1 h1:h5E0h5/Y8niHc5DlaLlWLArTQI7tMrsfQjHV+d9Z
 github.com/prometheus/common v0.66.1/go.mod h1:gcaUsgf3KfRSwHY4dIMXLPV0K/Wg1oZ8+SbZk/HH/dA=
 github.com/prometheus/procfs v0.16.1 h1:hZ15bTNuirocR6u0JZ6BAHHmwS1p8B4P6MRqxtzMyRg=
 github.com/prometheus/procfs v0.16.1/go.mod h1:teAbpZRB1iIAJYREa1LsoWUXykVXA1KlTmWl8x/U+Is=
-github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec h1:W09IVJc94icq4NjY3clb7Lk8O1qJ8BdBEF8z0ibU0rE=
-github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec/go.mod h1:qqbHyh8v60DhA7CoWK5oRCqLrMHRGoxYCSS9EjAz6Eo=
 github.com/rogpeppe/go-internal v1.14.1 h1:UQB4HGPB6osV0SQTLymcB4TgvyWu6ZyliaW0tI/otEQ=
 github.com/rogpeppe/go-internal v1.14.1/go.mod h1:MaRKkUm5W0goXpeCfT7UZI6fk/L7L7so1lCWt35ZSgc=
 github.com/russross/blackfriday/v2 v2.1.0/go.mod h1:+Rmxgy9KzJVeS9/2gXHxylqXiyQDYRxCVz55jmeOWTM=
@@ -274,25 +229,25 @@ github.com/sagikazarmark/locafero v0.6.0 h1:ON7AQg37yzcRPU69mt7gwhFEBwxI6P9T4Qu3
 github.com/sagikazarmark/locafero v0.6.0/go.mod h1:77OmuIc6VTraTXKXIs/uvUxKGUXjE1GbemJYHqdNjX0=
 github.com/sagikazarmark/slog-shim v0.1.0 h1:diDBnUNK9N/354PgrxMywXnAwEr1QZcOr6gto+ugjYE=
 github.com/sagikazarmark/slog-shim v0.1.0/go.mod h1:SrcSrq8aKtyuqEI1uvTDTK1arOWRIczQRv+GVI1AkeQ=
-github.com/sirupsen/logrus v1.9.3 h1:dueUQJ1C2q9oE3F7wvmSGAaVtTmUizReu6fjN8uqzbQ=
-github.com/sirupsen/logrus v1.9.3/go.mod h1:naHLuLoDiP4jHNo9R0sCBMtWGeIprob74mVsIT4qYEQ=
+github.com/sirupsen/logrus v1.9.4 h1:TsZE7l11zFCLZnZ+teH4Umoq5BhEIfIzfRDZ1Uzql2w=
+github.com/sirupsen/logrus v1.9.4/go.mod h1:ftWc9WdOfJ0a92nsE2jF5u5ZwH8Bv2zdeOC42RjbV2g=
 github.com/sourcegraph/conc v0.3.0 h1:OQTbbt6P72L20UqAkXXuLOj79LfEanQ+YQFNpLA9ySo=
 github.com/sourcegraph/conc v0.3.0/go.mod h1:Sdozi7LEKbFPqYX2/J+iBAM6HpqSLTASQIKqDmF7Mt0=
 github.com/spf13/afero v1.11.0 h1:WJQKhtpdm3v2IzqG8VMqrr6Rf3UYpEF239Jy9wNepM8=
 github.com/spf13/afero v1.11.0/go.mod h1:GH9Y3pIexgf1MTIWtNGyogA5MwRIDXGUr+hbWNoBjkY=
 github.com/spf13/cast v1.7.1 h1:cuNEagBQEHWN1FnbGEjCXL2szYEXqfJPbP2HNUaca9Y=
 github.com/spf13/cast v1.7.1/go.mod h1:ancEpBxwJDODSW/UG4rDrAqiKolqNNh2DX3mk86cAdo=
-github.com/spf13/cobra v1.10.1 h1:lJeBwCfmrnXthfAupyUTzJ/J4Nc1RsHC/mSRU2dll/s=
-github.com/spf13/cobra v1.10.1/go.mod h1:7SmJGaTHFVBY0jW4NXGluQoLvhqFQM+6XSKD+P4XaB0=
-github.com/spf13/pflag v1.0.9 h1:9exaQaMOCwffKiiiYk6/BndUBv+iRViNW+4lEMi0PvY=
+github.com/spf13/cobra v1.10.2 h1:DMTTonx5m65Ic0GOoRY2c16WCbHxOOw6xxezuLaBpcU=
+github.com/spf13/cobra v1.10.2/go.mod h1:7C1pvHqHw5A4vrJfjNwvOdzYu0Gml16OCs2GRiTUUS4=
 github.com/spf13/pflag v1.0.9/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=
+github.com/spf13/pflag v1.0.10 h1:4EBh2KAYBwaONj6b2Ye1GiHfwjqyROoF4RwYO+vPwFk=
+github.com/spf13/pflag v1.0.10/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=
 github.com/spf13/viper v1.19.0 h1:RWq5SEjt8o25SROyN3z2OrDB9l7RPd3lwTWU8EcEdcI=
 github.com/spf13/viper v1.19.0/go.mod h1:GQUN9bilAbhU/jgc1bKs99f/suXKeUMct8Adx5+Ntkg=
 github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
 github.com/stretchr/objx v0.5.2 h1:xuMeJ0Sdp5ZMRXx/aWO6RZxdr3beISkG5/G/aIRr3pY=
 github.com/stretchr/objx v0.5.2/go.mod h1:FRsXN1f5AsAjCGJKqEizvkpNtU+EGNCLh3NxZ/8L+MA=
 github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UVUgZn+9EI=
-github.com/stretchr/testify v1.7.0/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
 github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U=
 github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
 github.com/subosito/gotenv v1.6.0 h1:9NlTDc1FTs4qu0DDq7AEtTPNw6SVm7uBMsUCUjABIf8=
@@ -305,8 +260,6 @@ github.com/tidwall/pretty v1.2.1 h1:qjsOFOWWQl+N3RsoF5/ssm1pHmJJwhjlSbZ51I6wMl4=
 github.com/tidwall/pretty v1.2.1/go.mod h1:ITEVvHYasfjBbM0u2Pg8T2nJnzm8xPwvNhhsoaGGjNU=
 github.com/tidwall/sjson v1.2.5 h1:kLy8mja+1c9jlljvWTlSazM7cKDRfJuR/bOJhcY5NcY=
 github.com/tidwall/sjson v1.2.5/go.mod h1:Fvgq9kS/6ociJEDnK0Fk1cpYF4FIW6ZF7LAe+6jwd28=
-github.com/tj/go-spin v1.1.0 h1:lhdWZsvImxvZ3q1C5OIB7d72DuOwP4O2NdBg9PyzNds=
-github.com/tj/go-spin v1.1.0/go.mod h1:Mg1mzmePZm4dva8Qz60H2lHwmJ2loum4VIrLgVnKwh4=
 github.com/u-root/uio v0.0.0-20240224005618-d2acac8f3701 h1:pyC9PaHYZFgEKFdlp3G8RaCKgVpHZnecvArXvPXcFkM=
 github.com/u-root/uio v0.0.0-20240224005618-d2acac8f3701/go.mod h1:P3a5rG4X7tI17Nn3aOIAYr5HbIMukwXG0urG0WuL8OA=
 github.com/valyala/bytebufferpool v1.0.0 h1:GqA5TC/0021Y/b9FG4Oi9Mr3q7XYx6KllzawFIhcdPw=
@@ -317,144 +270,101 @@ github.com/vishvananda/netns v0.0.5 h1:DfiHV+j8bA32MFM7bfEunvT8IAqQ/NzSJHtcmW5zd
 github.com/vishvananda/netns v0.0.5/go.mod h1:SpkAiCQRtJ6TvvxPnOSyH3BMl6unz3xZlaprSwhNNJM=
 github.com/x448/float16 v0.8.4 h1:qLwI1I70+NjRFUR3zs1JPUCgaCXSh3SW62uAKT1mSBM=
 github.com/x448/float16 v0.8.4/go.mod h1:14CWIYCyZA/cWjXOioeEpHeN/83MdbZDRQHoFcYsOfg=
-github.com/xlab/c-for-go v1.3.0 h1:WDc+/E59g1OcZ4dYB4K9V4Doh4NnVBMYxaZFhNi77Lg=
-github.com/xlab/c-for-go v1.3.0/go.mod h1:0qVnGIfoNg7c5gABptM38Y7nnXgAMo5KYWVQPycu63Y=
-github.com/xlab/pkgconfig v0.0.0-20170226114623-cea12a0fd245 h1:Sw125DKxZhPUI4JLlWugkzsrlB50jR9v2khiD9FxuSo=
-github.com/xlab/pkgconfig v0.0.0-20170226114623-cea12a0fd245/go.mod h1:C+diUUz7pxhNY6KAoLgrTYARGWnt82zWTylZlxT92vk=
 github.com/xo/terminfo v0.0.0-20220910002029-abceb7e1c41e h1:JVG44RsyaB9T2KIHavMF/ppJZNG9ZpyihvCd0w101no=
 github.com/xo/terminfo v0.0.0-20220910002029-abceb7e1c41e/go.mod h1:RbqR21r5mrJuqunuUZ/Dhy/avygyECGrLceyNeo4LiM=
 github.com/yuin/goldmark v1.1.27/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
 github.com/yuin/goldmark v1.2.1/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
-github.com/yuin/goldmark v1.4.0/go.mod h1:mwnBkeHKe2W/ZEtQ+71ViKU8L12m81fl3OWwC1Zlc8k=
-go.etcd.io/etcd/api/v3 v3.6.6 h1:mcaMp3+7JawWv69p6QShYWS8cIWUOl32bFLb6qf8pOQ=
-go.etcd.io/etcd/api/v3 v3.6.6/go.mod h1:f/om26iXl2wSkcTA1zGQv8reJRSLVdoEBsi4JdfMrx4=
-go.etcd.io/etcd/client/pkg/v3 v3.6.6 h1:uoqgzSOv2H9KlIF5O1Lsd8sW+eMLuV6wzE3q5GJGQNs=
-go.etcd.io/etcd/client/pkg/v3 v3.6.6/go.mod h1:YngfUVmvsvOJ2rRgStIyHsKtOt9SZI2aBJrZiWJhCbI=
-go.etcd.io/etcd/client/v3 v3.6.6 h1:G5z1wMf5B9SNexoxOHUGBaULurOZPIgGPsW6CN492ec=
-go.etcd.io/etcd/client/v3 v3.6.6/go.mod h1:36Qv6baQ07znPR3+n7t+Rk5VHEzVYPvFfGmfF4wBHV8=
+go.etcd.io/etcd/api/v3 v3.6.11 h1:XFGTgrJ8nak3kB4NgMG8t7NT+lEeuuvKQAqUHKVgkWQ=
+go.etcd.io/etcd/api/v3 v3.6.11/go.mod h1:HYfTh0jyh+uFgp6gMbxJteIDYY97yMuYz85Rnw6Gy9o=
+go.etcd.io/etcd/client/pkg/v3 v3.6.11 h1:e41mp315Yn3QMGPmEzCyLsMINgJXTY/dX8kM++1csxU=
+go.etcd.io/etcd/client/pkg/v3 v3.6.11/go.mod h1:DysuMe/inqRyC/1tjRR6hReH/VV9Lufs27YKSKBWWJg=
+go.etcd.io/etcd/client/v3 v3.6.11 h1:LAByD96VmmeuairkvdAcE0RZnrmGz/q3ceeWePo9bwc=
+go.etcd.io/etcd/client/v3 v3.6.11/go.mod h1:vOTDMCo+fGPEClJqcFEFSqZ+8e7WKV7AyqJjX//HR2w=
 go.opentelemetry.io/auto/sdk v1.2.1 h1:jXsnJ4Lmnqd11kwkBV2LgLoFMZKizbCi5fNZ/ipaZ64=
 go.opentelemetry.io/auto/sdk v1.2.1/go.mod h1:KRTj+aOaElaLi+wW1kO/DZRXwkF4C5xPbEe3ZiIhN7Y=
 go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.61.0 h1:F7Jx+6hwnZ41NSFTO5q4LYDtJRXBf2PD0rNBkeB/lus=
 go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.61.0/go.mod h1:UHB22Z8QsdRDrnAtX4PntOl36ajSxcdUMt1sF7Y6E7Q=
-go.opentelemetry.io/otel v1.38.0 h1:RkfdswUDRimDg0m2Az18RKOsnI8UDzppJAtj01/Ymk8=
-go.opentelemetry.io/otel v1.38.0/go.mod h1:zcmtmQ1+YmQM9wrNsTGV/q/uyusom3P8RxwExxkZhjM=
-go.opentelemetry.io/otel/exporters/otlp/otlptrace v1.37.0 h1:Ahq7pZmv87yiyn3jeFz/LekZmPLLdKejuO3NcK9MssM=
-go.opentelemetry.io/otel/exporters/otlp/otlptrace v1.37.0/go.mod h1:MJTqhM0im3mRLw1i8uGHnCvUEeS7VwRyxlLC78PA18M=
-go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracehttp v1.37.0 h1:bDMKF3RUSxshZ5OjOTi8rsHGaPKsAt76FaqgvIUySLc=
-go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracehttp v1.37.0/go.mod h1:dDT67G/IkA46Mr2l9Uj7HsQVwsjASyV9SjGofsiUZDA=
-go.opentelemetry.io/otel/metric v1.38.0 h1:Kl6lzIYGAh5M159u9NgiRkmoMKjvbsKtYRwgfrA6WpA=
-go.opentelemetry.io/otel/metric v1.38.0/go.mod h1:kB5n/QoRM8YwmUahxvI3bO34eVtQf2i4utNVLr9gEmI=
-go.opentelemetry.io/otel/sdk v1.38.0 h1:l48sr5YbNf2hpCUj/FoGhW9yDkl+Ma+LrVl8qaM5b+E=
-go.opentelemetry.io/otel/sdk v1.38.0/go.mod h1:ghmNdGlVemJI3+ZB5iDEuk4bWA3GkTpW+DOoZMYBVVg=
-go.opentelemetry.io/otel/sdk/metric v1.38.0 h1:aSH66iL0aZqo//xXzQLYozmWrXxyFkBJ6qT5wthqPoM=
-go.opentelemetry.io/otel/sdk/metric v1.38.0/go.mod h1:dg9PBnW9XdQ1Hd6ZnRz689CbtrUp0wMMs9iPcgT9EZA=
-go.opentelemetry.io/otel/trace v1.38.0 h1:Fxk5bKrDZJUH+AMyyIXGcFAPah0oRcT+LuNtJrmcNLE=
-go.opentelemetry.io/otel/trace v1.38.0/go.mod h1:j1P9ivuFsTceSWe1oY+EeW3sc+Pp42sO++GHkg4wwhs=
-go.opentelemetry.io/proto/otlp v1.7.0 h1:jX1VolD6nHuFzOYso2E73H85i92Mv8JQYk0K9vz09os=
-go.opentelemetry.io/proto/otlp v1.7.0/go.mod h1:fSKjH6YJ7HDlwzltzyMj036AJ3ejJLCgCSHGj4efDDo=
+go.opentelemetry.io/otel v1.43.0 h1:mYIM03dnh5zfN7HautFE4ieIig9amkNANT+xcVxAj9I=
+go.opentelemetry.io/otel v1.43.0/go.mod h1:JuG+u74mvjvcm8vj8pI5XiHy1zDeoCS2LB1spIq7Ay0=
+go.opentelemetry.io/otel/exporters/otlp/otlptrace v1.43.0 h1:88Y4s2C8oTui1LGM6bTWkw0ICGcOLCAI5l6zsD1j20k=
+go.opentelemetry.io/otel/exporters/otlp/otlptrace v1.43.0/go.mod h1:Vl1/iaggsuRlrHf/hfPJPvVag77kKyvrLeD10kpMl+A=
+go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracehttp v1.43.0 h1:3iZJKlCZufyRzPzlQhUIWVmfltrXuGyfjREgGP3UUjc=
+go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracehttp v1.43.0/go.mod h1:/G+nUPfhq2e+qiXMGxMwumDrP5jtzU+mWN7/sjT2rak=
+go.opentelemetry.io/otel/metric v1.43.0 h1:d7638QeInOnuwOONPp4JAOGfbCEpYb+K6DVWvdxGzgM=
+go.opentelemetry.io/otel/metric v1.43.0/go.mod h1:RDnPtIxvqlgO8GRW18W6Z/4P462ldprJtfxHxyKd2PY=
+go.opentelemetry.io/otel/sdk v1.43.0 h1:pi5mE86i5rTeLXqoF/hhiBtUNcrAGHLKQdhg4h4V9Dg=
+go.opentelemetry.io/otel/sdk v1.43.0/go.mod h1:P+IkVU3iWukmiit/Yf9AWvpyRDlUeBaRg6Y+C58QHzg=
+go.opentelemetry.io/otel/sdk/metric v1.43.0 h1:S88dyqXjJkuBNLeMcVPRFXpRw2fuwdvfCGLEo89fDkw=
+go.opentelemetry.io/otel/sdk/metric v1.43.0/go.mod h1:C/RJtwSEJ5hzTiUz5pXF1kILHStzb9zFlIEe85bhj6A=
+go.opentelemetry.io/otel/trace v1.43.0 h1:BkNrHpup+4k4w+ZZ86CZoHHEkohws8AY+WTX09nk+3A=
+go.opentelemetry.io/otel/trace v1.43.0/go.mod h1:/QJhyVBUUswCphDVxq+8mld+AvhXZLhe+8WVFxiFff0=
+go.opentelemetry.io/proto/otlp v1.10.0 h1:IQRWgT5srOCYfiWnpqUYz9CVmbO8bFmKcwYxpuCSL2g=
+go.opentelemetry.io/proto/otlp v1.10.0/go.mod h1:/CV4QoCR/S9yaPj8utp3lvQPoqMtxXdzn7ozvvozVqk=
 go.uber.org/goleak v1.3.0 h1:2K3zAYmnTNqV73imy9J1T3WC+gmCePx2hEGkimedGto=
 go.uber.org/goleak v1.3.0/go.mod h1:CoHD4mav9JJNrW/WLlf7HGZPjdw8EucARQHekz1X6bE=
 go.uber.org/multierr v1.11.0 h1:blXXJkSxSSfBVBlC76pxqeO+LN3aDfLQo+309xJstO0=
 go.uber.org/multierr v1.11.0/go.mod h1:20+QtiLqy0Nd6FdQB9TLXag12DsQkrbs3htMFfDN80Y=
-go.uber.org/zap v1.27.1 h1:08RqriUEv8+ArZRYSTXy1LeBScaMpVSTBhCeaZYfMYc=
-go.uber.org/zap v1.27.1/go.mod h1:GB2qFLM7cTU87MWRP2mPIjqfIDnGu+VIO4V/SdhGo2E=
-go.yaml.in/yaml/v2 v2.4.2 h1:DzmwEr2rDGHl7lsFgAHxmNz/1NlQ7xLIrlN2h5d1eGI=
-go.yaml.in/yaml/v2 v2.4.2/go.mod h1:081UH+NErpNdqlCXm3TtEran0rJZGxAYx9hb/ELlsPU=
+go.uber.org/zap v1.28.0 h1:IZzaP1Fv73/T/pBMLk4VutPl36uNC+OSUh3JLG3FIjo=
+go.uber.org/zap v1.28.0/go.mod h1:rDLpOi171uODNm/mxFcuYWxDsqWSAVkFdX4XojSKg/Q=
+go.yaml.in/yaml/v2 v2.4.3 h1:6gvOSjQoTB3vt1l+CU+tSyi/HOjfOjRLJ4YwYZGwRO0=
+go.yaml.in/yaml/v2 v2.4.3/go.mod h1:zSxWcmIDjOzPXpjlTTbAsKokqkDNAVtZO0WOMiT90s8=
 go.yaml.in/yaml/v3 v3.0.4 h1:tfq32ie2Jv2UxXFdLJdh3jXuOzWiL1fo0bu/FbuKpbc=
 go.yaml.in/yaml/v3 v3.0.4/go.mod h1:DhzuOOF2ATzADvBadXxruRBLzYTpT36CKvDb3+aBEFg=
 golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
 golang.org/x/crypto v0.0.0-20191011191535-87dc89f01550/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI=
 golang.org/x/crypto v0.0.0-20200622213623-75b288015ac9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
-golang.org/x/crypto v0.45.0 h1:jMBrvKuj23MTlT0bQEOBcAE0mjg8mK9RXFhRH6nyF3Q=
-golang.org/x/crypto v0.45.0/go.mod h1:XTGrrkGJve7CYK7J8PEww4aY7gM3qMCElcJQ8n8JdX4=
+golang.org/x/crypto v0.51.0 h1:IBPXwPfKxY7cWQZ38ZCIRPI50YLeevDLlLnyC5wRGTI=
+golang.org/x/crypto v0.51.0/go.mod h1:8AdwkbraGNABw2kOX6YFPs3WM22XqI4EXEd8g+x7Oc8=
 golang.org/x/exp v0.0.0-20250103183323-7d7fa50e5329 h1:9kj3STMvgqy3YA4VQXBrN7925ICMxD5wzMRcgA30588=
 golang.org/x/exp v0.0.0-20250103183323-7d7fa50e5329/go.mod h1:qj5a5QZpwLU2NLQudwIN5koi3beDhSAlJwa67PuM98c=
 golang.org/x/mod v0.2.0/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA=
 golang.org/x/mod v0.3.0/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA=
-golang.org/x/mod v0.4.2/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA=
-golang.org/x/mod v0.5.1/go.mod h1:5OXOZSfqPIIbmVBIIKWRFfZjPR0E5r58TLhUjH0a2Ro=
-golang.org/x/mod v0.29.0 h1:HV8lRxZC4l2cr3Zq1LvtOsi/ThTgWnUk/y64QSs8GwA=
-golang.org/x/mod v0.29.0/go.mod h1:NyhrlYXJ2H4eJiRy/WDBO6HMqZQ6q9nk4JzS3NuCK+w=
-golang.org/x/net v0.0.0-20190311183353-d8887717615a/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
+golang.org/x/mod v0.36.0 h1:JJjpVx6myfUsUdAzZuOSTTmRE0PfZeNWzzvKrP7amb4=
+golang.org/x/mod v0.36.0/go.mod h1:moc6ELqsWcOw5Ef3xVprK5ul/MvtVvkIXLziUOICjUQ=
 golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
+golang.org/x/net v0.0.0-20190503192946-f4e77d36d62c/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
 golang.org/x/net v0.0.0-20190620200207-3b0461eec859/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
-golang.org/x/net v0.0.0-20190827160401-ba9fcec4b297/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
-golang.org/x/net v0.0.0-20191007182048-72f939374954/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
-golang.org/x/net v0.0.0-20200202094626-16171245cfb2/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
 golang.org/x/net v0.0.0-20200226121028-0de0cce0169b/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
-golang.org/x/net v0.0.0-20201010224723-4f7140c49acb/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
 golang.org/x/net v0.0.0-20201021035429-f5854403a974/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
-golang.org/x/net v0.0.0-20201110031124-69a78807bb2b/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
-golang.org/x/net v0.0.0-20201216054612-986b41b23924/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
-golang.org/x/net v0.0.0-20201224014010-6772e930b67b/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
-golang.org/x/net v0.0.0-20210119194325-5f4716e94777/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
-golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
-golang.org/x/net v0.0.0-20210525063256-abc453219eb5/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
-golang.org/x/net v0.0.0-20210805182204-aaa1db679c0d/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
-golang.org/x/net v0.0.0-20210928044308-7d9f5e0b762b/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
-golang.org/x/net v0.0.0-20211020060615-d418f374d309/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
-golang.org/x/net v0.0.0-20211209124913-491a49abca63/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
-golang.org/x/net v0.0.0-20220107192237-5cfca573fb4d/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
-golang.org/x/net v0.47.0 h1:Mx+4dIFzqraBXUugkia1OOvlD6LemFo1ALMHjrXDOhY=
-golang.org/x/net v0.47.0/go.mod h1:/jNxtkgq5yWUGYkaZGqo27cfGZ1c5Nen03aYrrKpVRU=
-golang.org/x/oauth2 v0.32.0 h1:jsCblLleRMDrxMN29H3z/k1KliIvpLgCkE6R8FXXNgY=
-golang.org/x/oauth2 v0.32.0/go.mod h1:lzm5WQJQwKZ3nwavOZ3IS5Aulzxi68dUSgRHujetwEA=
+golang.org/x/net v0.0.0-20220923203811-8be639271d50/go.mod h1:YDH+HFinaLZZlnHAfSS6ZXJJ9M9t4Dl22yv3iI2vPwk=
+golang.org/x/net v0.54.0 h1:2zJIZAxAHV/OHCDTCOHAYehQzLfSXuf/5SoL/Dv6w/w=
+golang.org/x/net v0.54.0/go.mod h1:Sj4oj8jK6XmHpBZU/zWHw3BV3abl4Kvi+Ut7cQcY+cQ=
+golang.org/x/oauth2 v0.36.0 h1:peZ/1z27fi9hUOFCAZaHyrpWG5lwe0RJEEEeH0ThlIs=
+golang.org/x/oauth2 v0.36.0/go.mod h1:YDBUJMTkDnJS+A4BP4eZBjCqtokkg1hODuPjwiGPO7Q=
 golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20190911185100-cd5d95a43a6e/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20201020160332-67f06af15bc9/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20210220032951-036812b2e83c/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
-golang.org/x/sync v0.18.0 h1:kr88TuHDroi+UVf+0hZnirlk8o8T+4MrK6mr60WkH/I=
-golang.org/x/sync v0.18.0/go.mod h1:9KTHXmSnoGruLpwFjVSX0lNNA75CykiMECbovNTZqGI=
+golang.org/x/sync v0.0.0-20220923202941-7f9b1623fab7/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
+golang.org/x/sync v0.20.0 h1:e0PTpb7pjO8GAtTs2dQ6jYa5BWYlMuX047Dco/pItO4=
+golang.org/x/sync v0.20.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
 golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
-golang.org/x/sys v0.0.0-20190312061237-fead79001313/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20190411185658-b44545bcd369/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20190412213103-97732733099d/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20190826190057-c7b8b68b1456/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20191008105621-543471e840be/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20200202164722-d101bd2416d5/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20200930185726-fdedc70b468f/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20201009025420-dfb3f7c4e634/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20201118182958-a01c418693c7/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20201218084310-7d0127a74742/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20210110051926-789bb1bd4061/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20210119212857-b64e53b001e4/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20210123111255-9b0068b26619/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20210124154548-22da62e12c0c/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20210216163648-f7da38b97c65/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20210305230114-8fe3ee5dd75b/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20210423082822-04245dca01da/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20210525143221-35b2ab0089ea/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20210809222454-d867a43fc93e/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20210906170528-6f6e22806c34/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20210927094055-39ccf1dd6fa6/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20211025201205-69cdffdb9359/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20211210111614-af8b64212486/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20211216021012-1d35b9e2eb4e/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20220715151400-c0bba94af5f8/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.0.0-20210615035016-665e8c7367d1/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.0.0-20220319134239-a9b59b0215f8/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.0.0-20220728004956-3c1f35247d10/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.2.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.10.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.38.0 h1:3yZWxaJjBmCWXqhN1qh02AkOnCQ1poK6oF+a7xWL6Gc=
-golang.org/x/sys v0.38.0/go.mod h1:OgkHotnGiDImocRcuBABYBEXf8A9a87e/uXjp9XT3ks=
-golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
-golang.org/x/term v0.37.0 h1:8EGAD0qCmHYZg6J17DvsMy9/wJ7/D/4pV/wfnld5lTU=
-golang.org/x/term v0.37.0/go.mod h1:5pB4lxRNYYVZuTLmy8oR2BH8dflOR+IbTYFD8fi3254=
+golang.org/x/sys v0.45.0 h1:dO4czNzziLiiXplLQgBCEpCvXQ3dnkn0SdaZSYdQ+FY=
+golang.org/x/sys v0.45.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
+golang.org/x/term v0.0.0-20210927222741-03fcf44c2211/go.mod h1:jbD1KX2456YbFQfuXm/mYQcufACuNUgVhRMnK/tPxf8=
+golang.org/x/term v0.43.0 h1:S4RLU2sB31O/NCl+zFN9Aru9A/Cq2aqKpTZJ6B+DwT4=
+golang.org/x/term v0.43.0/go.mod h1:lrhlHNdQJHO+1qVYiHfFKVuVioJIheAc3fBSMFYEIsk=
 golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
 golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
-golang.org/x/text v0.3.6/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
-golang.org/x/text v0.31.0 h1:aC8ghyu4JhP8VojJ2lEHBnochRno1sgL6nEi9WGFGMM=
-golang.org/x/text v0.31.0/go.mod h1:tKRAlv61yKIjGGHX/4tP1LTbc13YSec1pxVEWXzfoeM=
-golang.org/x/time v0.9.0 h1:EsRrnYcQiGH+5FfbgvV4AP7qEZstoyrHB0DzarOQ4ZY=
-golang.org/x/time v0.9.0/go.mod h1:3BpzKBy/shNhVucY/MWOyx10tF3SFh9QdLuxbVysPQM=
+golang.org/x/text v0.3.7/go.mod h1:u+2+/6zg+i71rQMx5EYifcz6MCKuco9NR6JIITiCfzQ=
+golang.org/x/text v0.37.0 h1:Cqjiwd9eSg8e0QAkyCaQTNHFIIzWtidPahFWR83rTrc=
+golang.org/x/text v0.37.0/go.mod h1:a5sjxXGs9hsn/AJVwuElvCAo9v8QYLzvavO5z2PiM38=
+golang.org/x/time v0.14.0 h1:MRx4UaLrDotUKUdCIqzPC48t1Y9hANFKIRpNx+Te8PI=
+golang.org/x/time v0.14.0/go.mod h1:eL/Oa2bBBK0TkX57Fyni+NgnyQQN4LitPmob2Hjnqw4=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
 golang.org/x/tools v0.0.0-20191119224855-298f0cb1881e/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
 golang.org/x/tools v0.0.0-20200619180055-7c47624df98f/go.mod h1:EkVYQZoAsY45+roYkvgYkIh4xh/qjgUK9TdY2XT94GE=
 golang.org/x/tools v0.0.0-20210106214847-113979e3529a/go.mod h1:emZCQorbCU4vsT4fOWvOPXz4eW1wZW4PmDk9uLelYpA=
-golang.org/x/tools v0.1.0/go.mod h1:xkSsbof2nBLbhDlRMhhhyNLN/zl3eTqcnHD5viDpcZ0=
-golang.org/x/tools v0.1.7/go.mod h1:LGqMHiF4EqQNHR1JncWGqT5BVaXmza+X+BDGol+dOxo=
-golang.org/x/tools v0.38.0 h1:Hx2Xv8hISq8Lm16jvBZ2VQf+RLmbd7wVUsALibYI/IQ=
-golang.org/x/tools v0.38.0/go.mod h1:yEsQ/d/YK8cjh0L6rZlY8tgtlKiBNTL14pGDJPJpYQs=
+golang.org/x/tools v0.45.0 h1:18qN3FAooORvApf5XjCXgsuayZOEtXf6JK18I3+ONa8=
+golang.org/x/tools v0.45.0/go.mod h1:LuUGqqaXcXMEFEruIVJVm5mgDD8vww/z/SR1gQ4uE/0=
 golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191011141410-1b5146add898/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
@@ -463,69 +373,50 @@ golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173 h1:/jFs0duh4rdb8uI
 golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173/go.mod h1:tkCQ4FQXmpAgYVh++1cq16/dH4QJtmvpRv19DWGAHSA=
 golang.zx2c4.com/wireguard/wgctrl v0.0.0-20241231184526-a9ab2273dd10 h1:3GDAcqdIg1ozBNLgPy4SLT84nfcBjr6rhGtXYtrkWLU=
 golang.zx2c4.com/wireguard/wgctrl v0.0.0-20241231184526-a9ab2273dd10/go.mod h1:T97yPqesLiNrOYxkwmhMI0ZIlJDm+p0PMR8eRVeR5tQ=
-gonum.org/v1/gonum v0.16.0 h1:5+ul4Swaf3ESvrOnidPp4GZbzf0mxVQpDCYUQE7OJfk=
-gonum.org/v1/gonum v0.16.0/go.mod h1:fef3am4MQ93R2HHpKnLk4/Tbh/s0+wqD5nfa6Pnwy4E=
-google.golang.org/genproto/googleapis/api v0.0.0-20251022142026-3a174f9686a8 h1:mepRgnBZa07I4TRuomDE4sTIYieg/osKmzIf4USdWS4=
-google.golang.org/genproto/googleapis/api v0.0.0-20251022142026-3a174f9686a8/go.mod h1:fDMmzKV90WSg1NbozdqrE64fkuTv6mlq2zxo9ad+3yo=
-google.golang.org/genproto/googleapis/rpc v0.0.0-20251022142026-3a174f9686a8 h1:M1rk8KBnUsBDg1oPGHNCxG4vc1f49epmTO7xscSajMk=
-google.golang.org/genproto/googleapis/rpc v0.0.0-20251022142026-3a174f9686a8/go.mod h1:7i2o+ce6H/6BluujYR+kqX3GKH+dChPTQU19wjRPiGk=
-google.golang.org/grpc v1.77.0 h1:wVVY6/8cGA6vvffn+wWK5ToddbgdU3d8MNENr4evgXM=
-google.golang.org/grpc v1.77.0/go.mod h1:z0BY1iVj0q8E1uSQCjL9cppRj+gnZjzDnzV0dHhrNig=
-google.golang.org/protobuf v1.36.10 h1:AYd7cD/uASjIL6Q9LiTjz8JLcrh/88q5UObnmY3aOOE=
-google.golang.org/protobuf v1.36.10/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=
+gonum.org/v1/gonum v0.17.0 h1:VbpOemQlsSMrYmn7T2OUvQ4dqxQXU+ouZFQsZOx50z4=
+gonum.org/v1/gonum v0.17.0/go.mod h1:El3tOrEuMpv2UdMrbNlKEh9vd86bmQ6vqIcDwxEOc1E=
+google.golang.org/genproto/googleapis/api v0.0.0-20260401024825-9d38bb4040a9 h1:VPWxll4HlMw1Vs/qXtN7BvhZqsS9cdAittCNvVENElA=
+google.golang.org/genproto/googleapis/api v0.0.0-20260401024825-9d38bb4040a9/go.mod h1:7QBABkRtR8z+TEnmXTqIqwJLlzrZKVfAUm7tY3yGv0M=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20260401024825-9d38bb4040a9 h1:m8qni9SQFH0tJc1X0vmnpw/0t+AImlSvp30sEupozUg=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20260401024825-9d38bb4040a9/go.mod h1:4Hqkh8ycfw05ld/3BWL7rJOSfebL2Q+DVDeRgYgxUU8=
+google.golang.org/grpc v1.81.1 h1:VnnIIZ88UzOOKLukQi+ImGz8O1Wdp8nAGGnvOfEIWQQ=
+google.golang.org/grpc v1.81.1/go.mod h1:xGH9GfzOyMTGIOXBJmXt+BX/V0kcdQbdcuwQ/zNw42I=
+google.golang.org/protobuf v1.36.12-0.20260120151049-f2248ac996af h1:+5/Sw3GsDNlEmu7TfklWKPdQ0Ykja5VEmq2i817+jbI=
+google.golang.org/protobuf v1.36.12-0.20260120151049-f2248ac996af/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=
 gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c/go.mod h1:JHkPIbrfpd72SG/EVd6muEfDQjcINNoR0C8j2r3qZ4Q=
-gopkg.in/evanphx/json-patch.v4 v4.12.0 h1:n6jtcsulIzXPJaxegRbvFNNrZDjbij7ny3gmSPG+6V4=
-gopkg.in/evanphx/json-patch.v4 v4.12.0/go.mod h1:p8EYWUEYMpynmqDbY58zCKCFZw8pRWMG4EsWvDvM72M=
+gopkg.in/evanphx/json-patch.v4 v4.13.0 h1:czT3CmqEaQ1aanPc5SdlgQrrEIb8w/wwCvWWnfEbYzo=
+gopkg.in/evanphx/json-patch.v4 v4.13.0/go.mod h1:p8EYWUEYMpynmqDbY58zCKCFZw8pRWMG4EsWvDvM72M=
 gopkg.in/inf.v0 v0.9.1 h1:73M5CoZyi3ZLMOyDlQh031Cx6N9NDJ2Vvfl76EDAgDc=
 gopkg.in/inf.v0 v0.9.1/go.mod h1:cWUDdTG/fYaXco+Dcufb5Vnc6Gp2YChqWtbxRZE0mXw=
 gopkg.in/ini.v1 v1.67.0 h1:Dgnx+6+nfE+IfzjUEISNeydPJh9AXNNsWbGP9KzCsOA=
 gopkg.in/ini.v1 v1.67.0/go.mod h1:pNLf8WUiyNEtQjuu5G5vTm06TEv9tsIgeAvK8hOrP4k=
-gopkg.in/yaml.v2 v2.4.0 h1:D8xgwECY7CYvx+Y2n4sBz93Jn9JRvxdiyyo8CTfuKaY=
-gopkg.in/yaml.v2 v2.4.0/go.mod h1:RDklbk79AGWmwhnvt/jBztapEOGDOx6ZbXqjP6csGnQ=
-gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
 gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
 gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
 gotest.tools/v3 v3.4.0 h1:ZazjZUfuVeZGLAmlKKuyv3IKP5orXcwtOwDQH6YVr6o=
 gotest.tools/v3 v3.4.0/go.mod h1:CtbdzLSsqVhDgMtKsx03ird5YTGB3ar27v0u/yKBW5g=
-honnef.co/go/tools v0.2.1/go.mod h1:lPVVZ2BS5TfnjLyizF7o7hv7j9/L+8cZY2hLyjP9cGY=
-honnef.co/go/tools v0.2.2/go.mod h1:lPVVZ2BS5TfnjLyizF7o7hv7j9/L+8cZY2hLyjP9cGY=
-k8s.io/api v0.34.2 h1:fsSUNZhV+bnL6Aqrp6O7lMTy6o5x2C4XLjnh//8SLYY=
-k8s.io/api v0.34.2/go.mod h1:MMBPaWlED2a8w4RSeanD76f7opUoypY8TFYkSM+3XHw=
-k8s.io/apimachinery v0.34.2 h1:zQ12Uk3eMHPxrsbUJgNF8bTauTVR2WgqJsTmwTE/NW4=
-k8s.io/apimachinery v0.34.2/go.mod h1:/GwIlEcWuTX9zKIg2mbw0LRFIsXwrfoVxn+ef0X13lw=
-k8s.io/client-go v0.34.1 h1:ZUPJKgXsnKwVwmKKdPfw4tB58+7/Ik3CrjOEhsiZ7mY=
-k8s.io/client-go v0.34.1/go.mod h1:kA8v0FP+tk6sZA0yKLRG67LWjqufAoSHA2xVGKw9Of8=
-k8s.io/klog/v2 v2.130.1 h1:n9Xl7H1Xvksem4KFG4PYbdQCQxqc/tTUyrgXaOhHSzk=
-k8s.io/klog/v2 v2.130.1/go.mod h1:3Jpz1GvMt720eyJH1ckRHK1EDfpxISzJ7I9OYgaDtPE=
-k8s.io/kube-openapi v0.0.0-20250710124328-f3f2b991d03b h1:MloQ9/bdJyIu9lb1PzujOPolHyvO06MXG5TUIj2mNAA=
-k8s.io/kube-openapi v0.0.0-20250710124328-f3f2b991d03b/go.mod h1:UZ2yyWbFTpuhSbFhv24aGNOdoRdJZgsIObGBUaYVsts=
-k8s.io/utils v0.0.0-20250604170112-4c0f3b243397 h1:hwvWFiBzdWw1FhfY1FooPn3kzWuJ8tmbZBHi4zVsl1Y=
-k8s.io/utils v0.0.0-20250604170112-4c0f3b243397/go.mod h1:OLgZIPagt7ERELqWJFomSt595RzquPNLL48iOWgYOg0=
-modernc.org/cc/v4 v4.24.4 h1:TFkx1s6dCkQpd6dKurBNmpo+G8Zl4Sq/ztJ+2+DEsh0=
-modernc.org/cc/v4 v4.24.4/go.mod h1:uVtb5OGqUKpoLWhqwNQo/8LwvoiEBLvZXIQ/SmO6mL0=
-modernc.org/ccorpus2 v1.5.2 h1:Ui+4tc58mf/W+2arcYCJR903y3zl3ecsI7Fpaaqozyw=
-modernc.org/ccorpus2 v1.5.2/go.mod h1:Wifvo4Q/qS/h1aRoC2TffcHsnxwTikmi1AuLANuucJQ=
-modernc.org/mathutil v1.7.1 h1:GCZVGXdaN8gTqB1Mf/usp1Y/hSqgI2vAGGP4jZMCxOU=
-modernc.org/mathutil v1.7.1/go.mod h1:4p5IwJITfppl0G4sUEDtCr4DthTaT47/N3aT6MhfgJg=
-modernc.org/opt v0.1.4 h1:2kNGMRiUjrp4LcaPuLY2PzUfqM/w9N23quVwhKt5Qm8=
-modernc.org/opt v0.1.4/go.mod h1:03fq9lsNfvkYSfxrfUhZCWPk1lm4cq4N+Bh//bEtgns=
-modernc.org/sortutil v1.2.1 h1:+xyoGf15mM3NMlPDnFqrteY07klSFxLElE2PVuWIJ7w=
-modernc.org/sortutil v1.2.1/go.mod h1:7ZI3a3REbai7gzCLcotuw9AC4VZVpYMjDzETGsSMqJE=
-modernc.org/strutil v1.2.1 h1:UneZBkQA+DX2Rp35KcM69cSsNES9ly8mQWD71HKlOA0=
-modernc.org/strutil v1.2.1/go.mod h1:EHkiggD70koQxjVdSBM3JKM7k6L0FbGE5eymy9i3B9A=
-modernc.org/token v1.1.0 h1:Xl7Ap9dKaEs5kLoOQeQmPWevfnk/DM5qcLcYlA8ys6Y=
-modernc.org/token v1.1.0/go.mod h1:UGzOrNV1mAFSEB63lOFHIpNRUVMvYTc6yu1SMY/XTDM=
+k8s.io/api v0.36.1 h1:XbL/EMj8K2aJpJtePmqUyQMsM0D4QI2pvl7YKJ20FTY=
+k8s.io/api v0.36.1/go.mod h1:KOWo4ey3TINlXjeHVuwB3i+tXXnu+UcwFBHlI/9dvEo=
+k8s.io/apimachinery v0.36.1 h1:G63Gjx2W+q0YD+72Vo8oY0nDnePVwnuzTmmy5ENrVSA=
+k8s.io/apimachinery v0.36.1/go.mod h1:ibYOR00vW/I1kzvi5SF0dRuJ52BvKtfvRdOn35GPQ+8=
+k8s.io/client-go v0.36.1 h1:FN/K8QIT2CEDt+2WB2HnWrUANZ50AP5GII43/SP2JR0=
+k8s.io/client-go v0.36.1/go.mod h1:s6rAnCtTGYDQnpNjEhSaISV+2O8jwruZ6m3QOYBFbtU=
+k8s.io/klog/v2 v2.140.0 h1:Tf+J3AH7xnUzZyVVXhTgGhEKnFqye14aadWv7bzXdzc=
+k8s.io/klog/v2 v2.140.0/go.mod h1:o+/RWfJ6PwpnFn7OyAG3QnO47BFsymfEfrz6XyYSSp0=
+k8s.io/kube-openapi v0.0.0-20260317180543-43fb72c5454a h1:xCeOEAOoGYl2jnJoHkC3hkbPJgdATINPMAxaynU2Ovg=
+k8s.io/kube-openapi v0.0.0-20260317180543-43fb72c5454a/go.mod h1:uGBT7iTA6c6MvqUvSXIaYZo9ukscABYi2btjhvgKGZ0=
+k8s.io/utils v0.0.0-20260210185600-b8788abfbbc2 h1:AZYQSJemyQB5eRxqcPky+/7EdBj0xi3g0ZcxxJ7vbWU=
+k8s.io/utils v0.0.0-20260210185600-b8788abfbbc2/go.mod h1:xDxuJ0whA3d0I4mf/C4ppKHxXynQ+fxnkmQH0vTHnuk=
 pgregory.net/rapid v1.1.0 h1:CMa0sjHSru3puNx+J0MIAuiiEV4N0qj8/cMWGBBCsjw=
 pgregory.net/rapid v1.1.0/go.mod h1:PY5XlDGj0+V1FCq0o192FdRhpKHGTRIWBgqjDBTrq04=
-sigs.k8s.io/json v0.0.0-20241014173422-cfa47c3a1cc8 h1:gBQPwqORJ8d8/YNZWEjoZs7npUVDpVXUUOFfW6CgAqE=
-sigs.k8s.io/json v0.0.0-20241014173422-cfa47c3a1cc8/go.mod h1:mdzfpAEoE6DHQEN0uh9ZbOCuHbLK5wOm7dK4ctXE9Tg=
-sigs.k8s.io/kind v0.30.0 h1:2Xi1KFEfSMm0XDcvKnUt15ZfgRPCT0OnCBbpgh8DztY=
-sigs.k8s.io/kind v0.30.0/go.mod h1:FSqriGaoTPruiXWfRnUXNykF8r2t+fHtK0P0m1AbGF8=
+sigs.k8s.io/json v0.0.0-20250730193827-2d320260d730 h1:IpInykpT6ceI+QxKBbEflcR5EXP7sU1kvOlxwZh5txg=
+sigs.k8s.io/json v0.0.0-20250730193827-2d320260d730/go.mod h1:mdzfpAEoE6DHQEN0uh9ZbOCuHbLK5wOm7dK4ctXE9Tg=
+sigs.k8s.io/kind v0.31.0 h1:UcT4nzm+YM7YEbqiAKECk+b6dsvc/HRZZu9U0FolL1g=
+sigs.k8s.io/kind v0.31.0/go.mod h1:FSqriGaoTPruiXWfRnUXNykF8r2t+fHtK0P0m1AbGF8=
 sigs.k8s.io/randfill v1.0.0 h1:JfjMILfT8A6RbawdsK2JXGBR5AQVfd+9TbzrlneTyrU=
 sigs.k8s.io/randfill v1.0.0/go.mod h1:XeLlZ/jmk4i1HRopwe7/aU3H5n1zNUcX6TM94b3QxOY=
-sigs.k8s.io/structured-merge-diff/v6 v6.3.0 h1:jTijUJbW353oVOd9oTlifJqOGEkUw2jB/fXCbTiQEco=
-sigs.k8s.io/structured-merge-diff/v6 v6.3.0/go.mod h1:M3W8sfWvn2HhQDIbGWj3S099YozAsymCo/wrT5ohRUE=
+sigs.k8s.io/structured-merge-diff/v6 v6.3.2 h1:kwVWMx5yS1CrnFWA/2QHyRVJ8jM6dBA80uLmm0wJkk8=
+sigs.k8s.io/structured-merge-diff/v6 v6.3.2/go.mod h1:M3W8sfWvn2HhQDIbGWj3S099YozAsymCo/wrT5ohRUE=
 sigs.k8s.io/yaml v1.6.0 h1:G8fkbMSAFqgEFgh4b1wmtzDnioxFCUgTZhlbj5P9QYs=
 sigs.k8s.io/yaml v1.6.0/go.mod h1:796bPqUfzR/0jLAl6XjHl3Ck7MiyVv8dbTdyT3/pMf4=

pkg/arp/arp.go

@@ -25,6 +25,10 @@ type Instance struct {
 }
 
 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,
 	}
@@ -80,9 +84,8 @@ func (m *Manager) RemoveWithIPDelete(instance *Instance, deleteIP bool) {
 	if i != nil {
 		i.mu.Lock()
 		defer i.mu.Unlock()
-		if i.counter > 1 {
-			i.counter--
-		} else {
+		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 {
@@ -112,13 +115,19 @@ func (m *Manager) Count(name string) int {
 
 func (m *Manager) StartAdvertisement(ctx context.Context) {
 	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
-		default:
+		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 {
@@ -131,11 +140,6 @@ func (m *Manager) StartAdvertisement(ctx context.Context) {
 				return true
 			})
 		}
-		if m.config.ArpBroadcastRate < 500 {
-			log.Warn("[ARP manager] arp broadcast rate is too low", "rate (ms)", m.config.ArpBroadcastRate, "setting to (ms)", "3000")
-			m.config.ArpBroadcastRate = 3000
-		}
-		time.Sleep(time.Duration(m.config.ArpBroadcastRate) * time.Millisecond)
 	}
 }
 

pkg/backend/backend.go

@@ -1,6 +1,7 @@
 package backend
 
 import (
+	"context"
 	"fmt"
 	"time"
 
@@ -65,7 +66,7 @@ func (e *Entry) Check() bool {
 	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 +76,10 @@ 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 +3,78 @@ package bgp
 import (
 	"context"
 	"fmt"
+	log "log/slog"
 	"net"
 
 	api "github.com/osrg/gobgp/v3/api"
 )
 
 // 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(ctx, &api.AddPathRequest{
+			Path: p,
+		}); err != nil {
+			return err
+		}
+		log.Debug("[BGP] added host", "addr", addr, "cnt", len(objects)+1)
 	}
 
-	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)
+		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(ctx, &api.DeletePathRequest{
+			Path: p,
+		}); err != nil {
+			return err
+		}
+		delete(b.tracker, addr)
+		log.Debug("[BGP] deleted host", "addr", addr, "cnt", len(objects))
 	}
 
-	return b.s.DeletePath(context.Background(), &api.DeletePathRequest{
-		Path: p,
-	})
+	return nil
 }

pkg/bgp/peers.go

@@ -4,24 +4,30 @@ import (
 	"context"
 	"fmt"
 	"net"
+	"strconv"
+	"time"
 
-	//nolint
-
+	"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/v3/api"
 
 	"github.com/kube-vip/kube-vip/pkg/utils"
+	"github.com/osrg/gobgp/v3/pkg/config/oc"
 	"github.com/osrg/gobgp/v3/pkg/server"
 	"google.golang.org/protobuf/types/known/anypb"
 )
 
+const defaultBGPPort uint32 = 179
+
 // AddPeer will add peers to the BGP configuration
-func (b *Server) AddPeer(peer kubevip.BGPPeer) (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{
@@ -41,10 +47,21 @@ func (b *Server) AddPeer(peer kubevip.BGPPeer) (err error) {
 		Transport: &api.Transport{
 			MtuDiscovery:  true,
 			RemoteAddress: peer.Address,
-			RemotePort:    uint32(179),
+			RemotePort:    defaultBGPPort,
 		},
 	}
 
+	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,
+		}
+	}
+
 	if b.c.MpbgpNexthop != "" {
 		p.AfiSafis = []*api.AfiSafi{
 			{
@@ -74,16 +91,16 @@ func (b *Server) AddPeer(peer kubevip.BGPPeer) (err error) {
 			return fmt.Errorf("failed to get MP-BGP addresses: %w", err)
 		}
 
-		mask := "128"
+		mask := strconv.Itoa(vip.DefaultMaskIPv6)
 		address := ipv4Address
 		family := api.Family_AFI_IP
 		if utils.IsIPv4(p.Conf.NeighborAddress) {
-			mask = "32"
+			mask = strconv.Itoa(vip.DefaultMaskIPv4)
 			address = ipv6Address
 			family = api.Family_AFI_IP6
 		}
 
-		err = b.s.AddDefinedSet(context.Background(), &api.AddDefinedSetRequest{
+		err = b.s.AddDefinedSet(ctx, &api.AddDefinedSetRequest{
 			DefinedSet: &api.DefinedSet{
 				DefinedType: api.DefinedType_NEIGHBOR,
 				Name:        fmt.Sprintf("peer-%s", p.Conf.NeighborAddress),
@@ -95,7 +112,7 @@ func (b *Server) AddPeer(peer kubevip.BGPPeer) (err error) {
 		}
 
 		if address != "" {
-			if err := insertPolicy(b.s, address, p, family); err != nil {
+			if err := insertPolicy(ctx, b.s, address, p, family); err != nil {
 				return fmt.Errorf("failed to add policy: %w", err)
 			}
 		}
@@ -109,7 +126,7 @@ func (b *Server) AddPeer(peer kubevip.BGPPeer) (err error) {
 		}
 	}
 
-	if err := b.s.AddPeer(context.Background(), &api.AddPeerRequest{Peer: p}); err != nil {
+	if err := b.s.AddPeer(ctx, &api.AddPeerRequest{Peer: p}); err != nil {
 		return fmt.Errorf("failed to add peer: %v", err)
 	}
 
@@ -128,7 +145,7 @@ func (b *Server) getPath(ip net.IP) (path *api.Path) {
 		//nolint
 		nlri, _ := anypb.New(&api.IPAddressPrefix{
 			Prefix:    ip.String(),
-			PrefixLen: 32,
+			PrefixLen: vip.DefaultMaskIPv4,
 		})
 
 		//nolint
@@ -148,7 +165,7 @@ func (b *Server) getPath(ip net.IP) (path *api.Path) {
 		//nolint
 		nlri, _ := anypb.New(&api.IPAddressPrefix{
 			Prefix:    ip.String(),
-			PrefixLen: 128,
+			PrefixLen: vip.DefaultMaskIPv6,
 		})
 
 		v6Family := &api.Family{
@@ -172,7 +189,7 @@ func (b *Server) getPath(ip net.IP) (path *api.Path) {
 	return
 }
 
-func insertPolicy(s *server.BgpServer, address string, p *api.Peer, family api.Family_Afi) error {
+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"
@@ -218,14 +235,14 @@ func insertPolicy(s *server.BgpServer, address string, p *api.Peer, family api.F
 		},
 	}
 
-	err := s.AddPolicy(context.Background(), &api.AddPolicyRequest{
+	err := s.AddPolicy(ctx, &api.AddPolicyRequest{
 		Policy: policy,
 	})
 	if err != nil {
 		return fmt.Errorf("failed to add policy: %w", err)
 	}
 
-	err = s.AddPolicyAssignment(context.Background(), &api.AddPolicyAssignmentRequest{
+	err = s.AddPolicyAssignment(ctx, &api.AddPolicyAssignmentRequest{
 		Assignment: &api.PolicyAssignment{
 			Name:      "global",
 			Direction: api.PolicyDirection_EXPORT,
@@ -242,3 +259,44 @@ func insertPolicy(s *server.BgpServer, address string, p *api.Peer, family api.F
 
 	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 "", 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()
+			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:
+			}
+		}
+
+	}
+}

pkg/bgp/server.go

@@ -3,6 +3,7 @@ package bgp
 import (
 	"context"
 	"fmt"
+	"sync"
 	"time"
 
 	log "log/slog"
@@ -10,17 +11,14 @@ import (
 	"github.com/kube-vip/kube-vip/pkg/kubevip"
 	api "github.com/osrg/gobgp/v3/api"
 	gobgp "github.com/osrg/gobgp/v3/pkg/server"
-	"github.com/prometheus/client_golang/prometheus"
 )
 
 // Server manages a server object
 type Server struct {
-	s *gobgp.BgpServer
-	c *kubevip.BGPConfig
-
-	// This is a prometheus gauge indicating the state of the sessions.
-	// 1 means "ESTABLISHED", 0 means "NOT ESTABLISHED"
-	BGPSessionInfoGauge *prometheus.GaugeVec
+	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
@@ -38,24 +36,18 @@ func NewBGPServer(c kubevip.BGPConfig) (b *Server, err error) {
 	}
 
 	b = &Server{
-		s: gobgp.NewBgpServer(),
-		c: &c,
-
-		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"}),
+		s:       gobgp.NewBgpServer(),
+		c:       &c,
+		tracker: make(map[string]map[string]bool),
 	}
 	return
 }
 
 // Start starts the BGP server
-func (b *Server) Start(peerStateChangeCallback func(*api.WatchEventResponse_PeerEvent)) (err error) {
+func (b *Server) Start(ctx context.Context, peerStateChangeCallback func(*api.WatchEventResponse_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:        b.c.AS,
 			RouterId:   b.c.RouterID,
@@ -65,7 +57,7 @@ func (b *Server) Start(peerStateChangeCallback func(*api.WatchEventResponse_Peer
 		return
 	}
 
-	if err = b.s.WatchEvent(context.Background(), &api.WatchEventRequest{Peer: &api.WatchEventRequest_Peer{}}, func(r *api.WatchEventResponse) {
+	if err = b.s.WatchEvent(ctx, &api.WatchEventRequest{Peer: &api.WatchEventRequest_Peer{}}, func(r *api.WatchEventResponse) {
 		if p := r.GetPeer(); p != nil && p.Type == api.WatchEventResponse_PeerEvent_STATE {
 			log.Info("[BGP]", "peer", p.String())
 			if peerStateChangeCallback != nil {
@@ -77,13 +69,13 @@ func (b *Server) Start(peerStateChangeCallback func(*api.WatchEventResponse_Peer
 	}
 
 	for _, p := range b.c.Peers {
-		if err = b.AddPeer(p); err != nil {
+		if err = b.AddPeer(ctx, p); err != nil {
 			return
 		}
 	}
 
 	if b.c.Zebra.Enabled {
-		if err = b.s.EnableZebra(context.Background(), &api.EnableZebraRequest{
+		if err = b.s.EnableZebra(ctx, &api.EnableZebraRequest{
 			Url:          b.c.Zebra.URL,
 			Version:      b.c.Zebra.Version,
 			SoftwareName: b.c.Zebra.SoftwareName,
@@ -98,7 +90,8 @@ func (b *Server) Start(peerStateChangeCallback func(*api.WatchEventResponse_Peer
 
 // Close will stop a running BGP Server
 func (b *Server) Close() error {
-	ctx, cf := context.WithTimeout(context.Background(), 5*time.Second)
-	defer cf()
+	// create new BGP stop context (independent)
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
 	return b.s.StopBgp(ctx, &api.StopBgpRequest{})
 }

pkg/cluster/cluster.go

@@ -1,28 +1,39 @@
 package cluster
 
 import (
-	"sync"
+	"crypto/tls"
+	"crypto/x509"
+	"fmt"
+	"net/http"
+	"os"
+	"time"
 
 	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/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
-	arpMgr    *arp.Manager
+	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, intfMgr *networkinterface.Manager, arpMgr *arp.Manager) (*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 {
@@ -32,10 +43,22 @@ func InitCluster(c *kubevip.Config, disableVIP bool, intfMgr *networkinterface.M
 			return nil, err
 		}
 	}
+
+	if c.ControlPlaneHealthCheck.Address != "" {
+		healthCheckHTTPClient, err = newHealthCheckHTTPClient(c)
+		if err != nil {
+			return nil, fmt.Errorf("initializing BGP health check client: %w", err)
+		}
+	}
+
 	// Initialise the Cluster structure
 	newCluster := &Cluster{
-		Network: networks,
-		arpMgr:  arpMgr,
+		Network:               networks,
+		arpMgr:                arpMgr,
+		stop:                  make(chan bool),
+		routeMgr:              routeMgr,
+		nodeLabelMgr:          nodeLabelMgr,
+		healthCheckHTTPClient: healthCheckHTTPClient,
 	}
 
 	log.Debug("service security", "enabled", c.EnableServiceSecurity)
@@ -54,9 +77,9 @@ func startNetworking(c *kubevip.Config, intfMgr *networkinterface.Manager) ([]vi
 
 	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,
-			c.EnableLoadBalancer, c.EnableServiceSecurity, intfMgr)
+		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.EnableServiceSecurity, intfMgr)
 		if err != nil {
 			return nil, err
 		}
@@ -70,11 +93,42 @@ func startNetworking(c *kubevip.Config, intfMgr *networkinterface.Manager) ([]vi
 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
 }

pkg/cluster/clusterDDNS.go

@@ -2,6 +2,7 @@ package cluster
 
 import (
 	"context"
+	"sync"
 
 	"github.com/kube-vip/kube-vip/pkg/vip"
 )
@@ -12,9 +13,9 @@ 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, network vip.Network) error {
-	ddnsMgr := vip.NewDDNSManager(ctx, network)
-	ip, err := ddnsMgr.Start()
+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
 	}

pkg/cluster/clusterLeaderElection.go

@@ -3,132 +3,59 @@ 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/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"
 
 	log "log/slog"
-
-	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"
 )
 
-// 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)
+	})
+
+	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.Info("cluster membership", "namespace", c.Namespace, "lock", c.LeaseName, "id", 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
 
@@ -142,310 +69,178 @@ func (cluster *Cluster) StartCluster(c *kubevip.Config, sm *Manager, bgpServer *
 		}
 	}
 
-	// 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 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)
-		if err != nil {
-			log.Error("new BGP server", "err", err)
+	// 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)
 		}
-		if err := bgpServer.Start(nil); err != nil {
-			log.Error("starting BGP server", "err", err)
+
+		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 {
+		objLease.Cancel()
+		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("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 {
+				log.Info("VIP already configured on interface", "IP", cluster.Network[i].IP(), "interface", cluster.Network[i].Interface())
+			}
 		}
 	}
 
-	run := &runConfig{
-		config:  c,
-		leaseID: c.NodeName,
-		sm:      sm,
-		onStartedLeading: func(ctx context.Context) { //nolint TODO: potential clean code
-			// When we become leader, ensure we can take over VIPs even if they're preserved on other nodes
-			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("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 {
-						log.Info("VIP already configured on interface", "IP", cluster.Network[i].IP(), "interface", cluster.Network[i].Interface())
-					}
-				}
-			}
+	// As we're leading lets start the vip service
+	err := cluster.StartVipService(objLease.Ctx, c, em, bgpServer, killFunc)
+	if err != nil {
+		log.Error("starting VIP service on leader", "err", err)
+		killFunc()
+	}
+}
 
-			// Start ARP advertisements now that we have leadership
-			log.Info("Start ARP/NDP advertisement")
-			go cluster.arpMgr.StartAdvertisement(ctxArp)
+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")
 
-			// As we're leading lets start the vip service
-			err := cluster.vipService(ctxArp, ctxDNS, c, sm, bgpServer, cancel)
-			if err != nil {
-				log.Error("starting VIP service on leader", "err", err)
-			}
-		},
-		onStoppedLeading: func() {
-			// 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
+	}
 
-			// Stop the dns context
-			cancelDNS()
-			// Stop the Arp context if it is running
-			cancelArp()
+	// Stop the cluster context if it is running
+	objLease.Cancel()
 
-			// Stop the BGP server
-			if bgpServer != nil {
-				err := bgpServer.Close()
+	// Handle VIP cleanup based on configuration
+	if c.PreserveVIPOnLeadershipLoss {
+		// For IPv6, we must remove VIPs immediately to avoid DAD failures on the new leader
+		// IPv6 Duplicate Address Detection will fail if the new leader tries to add an IP that is still present on this node's interface
+		// We need to check each VIP individually and only remove IPv6 VIPs
+		for i := range cluster.Network {
+			if utils.IsIPv6(cluster.Network[i].IP()) {
+				log.Info("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("close BGP server", "err", err)
+					log.Warn("delete VIP", "err", err)
 				}
-			}
-
-			// Handle VIP cleanup based on configuration
-			if c.PreserveVIPOnLeadershipLoss {
-				// For IPv6, we must remove VIPs immediately to avoid DAD failures on the new leader
-				// IPv6 Duplicate Address Detection will fail if the new leader tries to add an IP that is still present on this node's interface
-				// We need to check each VIP individually and only remove IPv6 VIPs
-				for i := range cluster.Network {
-					if utils.IsIPv6(cluster.Network[i].IP()) {
-						log.Info("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("delete VIP", "err", err)
-						}
-						if deleted {
-							log.Info("deleted address", "IP", cluster.Network[i].IP(), "interface", cluster.Network[i].Interface())
-						}
-					} else {
-						log.Info("Preserving IPv4 VIP address on interface, only stopped ARP broadcasting", "ip", cluster.Network[i].IP())
-					}
+				if deleted {
+					log.Info("deleted address", "IP", cluster.Network[i].IP(), "interface", cluster.Network[i].Interface())
 				}
 			} else {
-				// Legacy behavior: delete VIP addresses on leadership loss
-				log.Info("Deleting VIP addresses on leadership loss (legacy behavior)")
-				for i := range cluster.Network {
-					deleted, err := cluster.Network[i].DeleteIP()
-					if err != nil {
-						log.Warn("delete VIP", "err", err)
-					}
-					if deleted {
-						log.Info("deleted address", "IP", cluster.Network[i].IP(), "interface", cluster.Network[i].Interface())
-					}
-				}
+				log.Info("Preserving IPv4 VIP address on interface, only stopped ARP broadcasting", "ip", cluster.Network[i].IP())
 			}
-
-			log.Error("lost leadership, restarting kube-vip")
-			panic("") // TODO - we could also return here
-		},
-		onNewLeader: func(identity string) {
-			// 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)
-					}
-					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())
-					}
-				}
+		}
+	} else {
+		// Legacy behavior: delete VIP addresses on leadership loss
+		log.Info("Deleting VIP addresses on leadership loss (legacy behavior)")
+		for i := range cluster.Network {
+			deleted, err := cluster.Network[i].DeleteIP()
+			if err != nil {
+				log.Warn("delete VIP", "err", err)
 			}
-		},
-	}
-
-	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(ctxArp 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",
-	}
-
-	rw, err := watchtools.NewRetryWatcherWithContext(ctxArp, "1", &cache.ListWatch{
-		WatchFunc: func(_ metav1.ListOptions) (watch.Interface, error) {
-			return sm.RetryWatcherClient.CoreV1().Nodes().Watch(context.Background(), listOptions)
-		},
-	})
-	if err != nil {
-		return fmt.Errorf("error creating label watcher: %s", err.Error())
-	}
-
-	go func() {
-		<-sm.SignalChan
-		log.Info("Received termination, signaling shutdown")
-		// Cancel the context
-		rw.Stop()
-	}()
-
-	ch := rw.ResultChan()
-	// defer rw.Stop()
-
-	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")
+			if deleted {
+				log.Info("deleted address", "IP", cluster.Network[i].IP(), "interface", cluster.Network[i].Interface())
 			}
-			// 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("add IPVS backend", "err", err)
-						}
-					} else {
-						err = lb.RemoveBackend(node.Status.Addresses[x].Address, port)
-						if err != nil {
-							log.Error("remove IPVS backend", "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("Del IPVS backend", "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")
-
-			// This round trip allows us to handle unstructured status
-			errObject := apierrors.FromObject(event.Object)
-			statusErr, ok := errObject.(*apierrors.StatusError)
-			if !ok {
-				log.Error(spew.Sprintf("Received an error which is not *metav1.Status but %#+v", event.Object))
-			}
-
-			status := statusErr.ErrStatus
-			log.Error("watcher", "status", status)
-		default:
 		}
 	}
 
-	log.Info("Exiting Node watcher")
-	return nil
+	log.Error("lost leadership, restarting kube-vip")
 }
 
-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
+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)
+			}
+			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())
 			}
 		}
 	}
-	return false
+}
+
+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

@@ -6,106 +6,132 @@ import (
 	"fmt"
 	"io/fs"
 	"net"
-	"os"
-	"os/signal"
+	"net/http"
+	"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/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/vishvananda/netlink"
 	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, cancelLeaderElection context.CancelFunc) 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{}
 
-	var arpWG sync.WaitGroup
-
 	for i := range cluster.Network {
 		network := cluster.Network[i]
 
 		if network.IsDDNS() {
-			if err := cluster.StartDDNS(ctxDNS, cluster.Network[i]); err != nil {
+			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 {
-			log.Error("failed to set mask", "subnet", c.VIPSubnet, "err", err)
-			panic("")
+			return fmt.Errorf("failed to set mask for subnet %q: %w", c.VIPSubnet, err)
 		}
 
 		// start the dns updater if address is dns
 		if network.IsDNS() {
 			log.Info("starting the DNS updater", "address", network.DNSName())
 			ipUpdater := vip.NewIPUpdater(network)
-			ipUpdater.Run(ctxDNS)
+			wg.Go(func() {
+				ipUpdater.Run(ctx)
+			})
 		}
 
 		if !c.EnableRoutingTable {
 			// Normal VIP addition, use skipDAD=false for normal DAD process
 			if _, err = network.AddIP(false, false); err != nil {
-				log.Error(err.Error())
+				return fmt.Errorf("failed to add IP address %s: %w", network.IP(), err)
 			}
 		}
 
 		if c.EnableBGP {
-			// 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(network.CIDR())
-			if err != nil {
-				log.Error(err.Error())
+			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 {
-			lb, err := loadbalancer.NewIPVSLB(network.IP(), c.LoadBalancerPort, c.LoadBalancerForwardingMethod, c.BackendHealthCheckInterval, c.Interface, cancelLeaderElection, signalChan)
+			lb, err := loadbalancer.NewIPVSLB(ctx, network.IP(), c.LoadBalancerPort, c.LoadBalancerForwardingMethod,
+				c.BackendHealthCheckInterval, killFunc, &wg)
 			if err != nil {
-				log.Error("Error creating IPVS LoadBalancer", "err", err)
+				return fmt.Errorf("creating IPVS LoadBalance: %w", err)
 			}
 
-			go func() {
-				err = sm.NodeWatcher(ctxArp, lb, c.Port) //TODO: We're using the ctxARP as the context this will change when rkatz finishes his change
-				if err != nil {
-					log.Error("Error watching node labels", "err", 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 errors.Is(err, &utils.PanicError{}) {
+								killFunc()
+								return
+							}
+						}
+					}
 				}
-			}()
+			})
 
 			loadbalancers = append(loadbalancers, lb)
 		}
 
 		if c.EnableARP {
-			arpWG.Add(1)
-			go cluster.layer2Update(ctxArp, network, c, &arpWG)
+			wg.Go(func() {
+				cluster.layer2Update(ctx, network, c)
+			})
 		}
 	}
 
 	if c.EnableLoadBalancer {
 		// Shutdown function that will wait on this signal, unless we call it ourselves
-		<-signalChan
+		<-ctx.Done()
 		for _, lb := range loadbalancers {
 			err = lb.RemoveIPVSLB()
 			if err != nil {
@@ -119,17 +145,10 @@ func (cluster *Cluster) vipService(ctxArp, ctxDNS context.Context, c *kubevip.Co
 		backendMapV6 := backend.Map{}
 		// only check localhost
 
-		nodename := ""
-		if c.NodeName != "" {
-			nodename = c.NodeName
-		} 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 nod", "err", err)
+		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)
 			}
 		}
 
@@ -160,15 +179,7 @@ func (cluster *Cluster) vipService(ctxArp, ctxDNS context.Context, c *kubevip.Co
 			}
 		}
 
-		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() {
+		backend.Watch(ctx, c.BackendHealthCheckInterval, func() {
 			for i := range cluster.Network {
 				network := cluster.Network[i]
 				networkIP := network.IP()
@@ -197,11 +208,11 @@ func (cluster *Cluster) vipService(ctxArp, ctxDNS context.Context, c *kubevip.Co
 							log.Info("added backend", "ip", network.IP())
 						}
 
-						err = network.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.Warn(err.Error())
 						} else if err == nil && !(*backendMap)[entry] {
-							log.Info("added route", "route", network.PrepareRoute().String())
+							log.Info("added route", "route", network.PrepareRoute())
 						}
 
 						(*backendMap)[entry] = true
@@ -219,29 +230,108 @@ func (cluster *Cluster) vipService(ctxArp, ctxDNS context.Context, c *kubevip.Co
 				}
 
 				if deleteAddress {
-					err = network.DeleteRoute()
-					if err != nil && !errors.Is(err, fs.ErrNotExist) && !errors.Is(err, syscall.ESRCH) {
+					err = cluster.routeMgr.Delete(c.NodeName, network)
+					if err != nil {
 						log.Warn("deleting route", "err", err)
-					} else if err == nil {
-						log.Info("deleted route", "route", network.PrepareRoute().String())
 					}
 
 					deleted, err := network.DeleteIP()
 					if err != nil {
 						log.Error("error deleting IP", "err", err)
-						panic("")
+						killFunc()
+						return
 					}
 					if deleted {
 						log.Info("deleted address", "IP", network.IP(), "interface", network.Interface())
 					}
 				}
 			}
-		}, c.BackendHealthCheckInterval, stop)
+		})
+	}
+
+	if c.EnableBGP {
+		<-ctx.Done()
 	}
 
 	return nil
 }
 
+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:
+		}
+	}
+}
+
 func isV6(ip string) (bool, error) {
 	ipaddr := net.ParseIP(ip)
 	if ipaddr == nil {
@@ -265,24 +355,37 @@ 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(ctx context.Context, c *kubevip.Config, bgp *bgp.Server, name string, CountRouteReferences func(*netlink.Route) int) {
+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(ctx)
+	lbCtx, lbCancel := context.WithCancel(ctx)
 
-	cluster.stop = make(chan bool, 1)
-	cluster.completed = make(chan bool, 1)
+	var lbWg sync.WaitGroup
 
-	var arpWG sync.WaitGroup
-
-	log.Debug("StartLoadBalancerService")
 	for i := range cluster.Network {
 		network := cluster.Network[i]
+
+		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)
-			panic("")
+			lbCancel()
+			return utils.NewPanicError(fmt.Sprintf("failed to set mask for subnet %q: %s", c.VIPSubnet, err.Error()))
 		}
 		_, err := network.DeleteIP()
 		if err != nil {
@@ -291,7 +394,7 @@ func (cluster *Cluster) StartLoadBalancerService(ctx context.Context, c *kubevip
 		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.Warn(err.Error())
 			} else {
@@ -299,50 +402,62 @@ func (cluster *Cluster) StartLoadBalancerService(ctx context.Context, c *kubevip
 			}
 		}
 
-		// Normal VIP addition, use skipDAD=false for normal DAD process
-		if _, err = network.AddIP(false, false); err != nil {
-			log.Warn(err.Error())
-		} else {
-			log.Info("successful add IP")
+		if !c.EnableRoutingTable && !c.EnableBGP && !c.EnableWireguard {
+			// 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")
+			}
 		}
 
 		if c.EnableARP {
-			arpWG.Add(1)
-			go cluster.layer2Update(ctxArp, network, c, &arpWG)
+			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
 			log.Debug("(svcs) attempting to advertise over BGP", "address", network.CIDR())
-			err = bgp.AddHost(network.CIDR())
+			err = bgp.AddHost(lbCtx, network.CIDR(), name)
 			if err != nil {
 				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]
 
-		arpWG.Wait() // wait for all cluster ARP/NDP to be finished
+			// 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)
+				})
+			}
+		}
+
+		<-cluster.stop
+		// 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 {
-				// chek if route is not  referenced by another service
-				r := cluster.Network[i].PrepareRoute()
-				if CountRouteReferences(r) < 1 {
-					log.Info("[VIP] Deleting Route for VIP", "IP", cluster.Network[i].IP())
-					if err := cluster.Network[i].DeleteRoute(); err != nil {
-						log.Warn(err.Error())
-					}
+				if err := cluster.routeMgr.Delete(name, cluster.Network[i]); err != nil {
+					log.Warn(err.Error())
 				}
 			}
 
-			close(cluster.completed)
 			return
 		}
 		for i := range cluster.Network {
@@ -379,24 +494,21 @@ func (cluster *Cluster) StartLoadBalancerService(ctx context.Context, c *kubevip
 				}
 			}
 		}
+	})
 
-		close(cluster.completed)
-	}()
+	return nil
 }
 
 // Layer2Update, handles the creation of the
-func (cluster *Cluster) layer2Update(ctx context.Context, network vip.Network, c *kubevip.Config, arpWG *sync.WaitGroup) {
-	defer arpWG.Done()
-	log.Info("layer 2 broadcaster starting")
+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 {
-			ndp, err = vip.NewNDPResponder(network.Interface())
+			ndp, err = waitNDPResponder(ctx, network.Interface())
 			if err != nil {
 				log.Error("failed to create new NDP Responder", "error", err)
 			} else {
@@ -407,6 +519,7 @@ func (cluster *Cluster) layer2Update(ctx context.Context, network vip.Network, c
 		}
 	}
 
+	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)
@@ -416,3 +529,30 @@ func (cluster *Cluster) layer2Update(ctx context.Context, network vip.Network, c
 	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_test.go

@@ -0,0 +1,347 @@
+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/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")
+}
+
+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
+}
+
+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
+}
+
+func (m *mockNetwork) AddIP(bool, bool, ...int) (bool, error) { return true, nil }
+func (m *mockNetwork) AddRoute(bool) (bool, error)            { return false, nil }
+func (m *mockNetwork) DeleteIP() (bool, error)                { return false, 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 "" }
+
+// 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

@@ -1,85 +0,0 @@
-package cluster
-
-import (
-	"context"
-
-	log "log/slog"
-
-	"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.Info("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 {
-			deleted, err := cluster.Network[i].DeleteIP()
-			if err != nil {
-				log.Warn("Attempted to clean existing VIP", "err", err)
-			}
-			if deleted {
-				log.Info("deleted address", "IP", cluster.Network[i].IP(), "interface", cluster.Network[i].Interface())
-			}
-
-			// Normal VIP addition for single node, use skipDAD=false for normal DAD process
-			_, err = cluster.Network[i].AddIP(false, false)
-			if err != nil {
-				log.Warn(err.Error())
-			}
-
-		}
-
-		if c.EnableARP {
-			// Gratuitous ARP, will broadcast to new MAC <-> IP
-			err := vip.ARPSendGratuitous(cluster.Network[i].IP(), c.Interface)
-			if err != nil {
-				log.Warn(err.Error())
-			}
-		}
-	}
-
-	go func() {
-		<-cluster.stop
-
-		if !disableVIP {
-			for i := range cluster.Network {
-				log.Info("[VIP] Releasing the VIP", "address", 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())
-				}
-			}
-		}
-		close(cluster.completed)
-	}()
-	log.Info("Started Load Balancer and Virtual IP")
-	return nil
-}
-
-func (cluster *Cluster) StartVipService(c *kubevip.Config, sm *Manager, bgp *bgp.Server) 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, nil)
-}

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/election/election.go

@@ -0,0 +1,267 @@
+package election
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"sync"
+	"time"
+
+	log "log/slog"
+
+	"github.com/davecgh/go-spew/spew"
+	"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"
+	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"
+)
+
+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 errors.Is(err, &utils.PanicError{}) {
+								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")
+
+			// This round trip allows us to handle unstructured status
+			errObject := apierrors.FromObject(event.Object)
+			statusErr, ok := errObject.(*apierrors.StatusError)
+			if !ok {
+				log.Error(spew.Sprintf("Received an error which is not *metav1.Status but %#+v", event.Object))
+			}
+
+			status := statusErr.ErrStatus
+			log.Error("watcher", "status", status)
+			watchErr = fmt.Errorf("node watcher error, status: %s", status.String())
+		default:
+		}
+	}
+
+	log.Info("Exiting Node watcher")
+	return watchErr
+}
+
+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

@@ -4,6 +4,9 @@ import (
 	"context"
 	"fmt"
 	"net"
+	"sync"
+	"sync/atomic"
+	"time"
 
 	log "log/slog"
 
@@ -11,10 +14,14 @@ import (
 	"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/utils"
+	"github.com/kube-vip/kube-vip/pkg/wireguard"
 	v1 "k8s.io/api/core/v1"
-	discoveryv1 "k8s.io/api/discovery/v1"
 	"k8s.io/apimachinery/pkg/watch"
+	"k8s.io/client-go/kubernetes"
 )
 
 type Processor struct {
@@ -23,26 +30,29 @@ type Processor struct {
 	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) *Processor {
+	instances *[]*instance.Instance, leaseMgr *lease.Manager, tunnelMgr *wireguard.TunnelManager, routeMgr *route.Manager) *Processor {
 	return &Processor{
 		config:    config,
 		provider:  provider,
 		bgpServer: bgpServer,
 		instances: instances,
-		worker:    newEndpointWorker(config, provider, bgpServer, instances),
+		leaseMgr:  leaseMgr,
+		worker:    newEndpointWorker(config, provider, bgpServer, instances, leaseMgr, tunnelMgr, routeMgr),
 	}
 }
 
-func (p *Processor) AddOrModify(ctx *servicecontext.Context, event watch.Event,
-	lastKnownGoodEndpoint *string, service *v1.Service, id string, leaderElectionActive *bool,
-	serviceFunc func(context.Context, *v1.Service) error,
-	leaderCtx *context.Context, cancel *context.CancelFunc) (bool, error) {
+func (p *Processor) AddOrModify(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) {
 
 	var err error
-	if err = p.provider.LoadObject(event.Object, *cancel); err != nil {
+	if err = p.provider.LoadObject(event.Object, svcCtx.Cancel); err != nil {
 		return false, fmt.Errorf("[%s] error loading k8s object: %w", p.provider.GetLabel(), err)
 	}
 
@@ -51,7 +61,7 @@ func (p *Processor) AddOrModify(ctx *servicecontext.Context, event watch.Event,
 		return false, err
 	}
 
-	if err := p.worker.setInstanceEndpointsStatus(service, endpoints); err != nil {
+	if err := p.worker.setInstanceEndpointsStatus(svcCtx.Ctx, service, endpoints); err != nil {
 		log.Error("updating instance", "err", err)
 	}
 
@@ -61,53 +71,90 @@ func (p *Processor) AddOrModify(ctx *servicecontext.Context, event watch.Event,
 	// If we have a local endpoint then begin the leader Election, unless it's already running
 	//
 
+	les := atomic.Int64{}
+
 	// Check that we have local endpoints
 	if len(endpoints) != 0 {
 		// Ignore IPv4
-		if service.Annotations[kubevip.EgressIPv6] == "true" && net.ParseIP(endpoints[0]).To4() != nil {
+		if service.Annotations[kubevip.EgressIPv6] == "true" && !hasV6(endpoints) {
 			return true, nil
 		}
 
-		p.updateLastKnownGoodEndpoint(lastKnownGoodEndpoint, endpoints, service, leaderElectionActive, *cancel)
-		// start leader election if it's enabled and not already started
-		if !*leaderElectionActive && p.config.EnableServicesElection {
-			go func() {
-				*leaderCtx, *cancel = context.WithCancel(ctx.Ctx)
-				startLeaderElection(*leaderCtx, leaderElectionActive, service, serviceFunc)
-			}()
+		p.updateLastKnownGoodEndpoint(lastKnownGoodEndpoint, endpoints, service)
+
+		if p.config.EnableServicesElection {
+			wg.Go(func() {
+				les.Add(1)
+				p.startLeaderElection(svcCtx, service, serviceFunc, wg)
+			})
+		} else if p.config.EnableARP || (p.config.EnableRoutingTable && p.config.EnableLeaderElection) {
+			if !svcCtx.Signalled.Load() {
+				inst := instance.FindServiceInstance(service, *p.instances)
+				if inst == nil {
+					return true, fmt.Errorf("[%s] failed to find an instance for service %s/%s", p.provider.GetLabel(), service.Namespace, service.Name)
+				}
+				// global leader election is enabled for services
+				for x := range inst.VIPConfigs {
+					log.Debug("starting loadbalancer for service", "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 true, fmt.Errorf("failed to start lb: %w", err)
+					}
+				}
+			}
 		}
 
+		svcCtx.SignalReadiness()
+
 		// There are local endpoints available on the node
-		if !p.config.EnableServicesElection && !p.config.EnableLeaderElection {
-			if err := p.worker.processInstance(ctx, service, leaderElectionActive); err != nil {
+		// Process immediately if:
+		// - No services/leader election is enabled, OR
+		// - WireGuard is enabled (it always needs immediate DNAT rule updates)
+		if (!p.config.EnableServicesElection && !p.config.EnableLeaderElection) || p.config.EnableWireguard {
+			if err := p.worker.processInstance(svcCtx, service); err != nil {
 				return false, fmt.Errorf("failed to process non-empty instance: %w", err)
 			}
 		}
 	} else {
 		// There are no local endpoints
-		p.worker.clear(ctx, lastKnownGoodEndpoint, service, *cancel, leaderElectionActive)
+		if svcCtx.Signalled.Load() {
+			svcCtx.ResetReadiness()
+			p.worker.clear(svcCtx, lastKnownGoodEndpoint, service)
+			if p.config.EnableARP && !p.config.EnableServicesElection {
+				i := instance.FindServiceInstance(service, *p.instances)
+				for _, c := range i.Clusters {
+					c.Stop()
+				}
+			}
+		}
 	}
 
 	// Set the service accordingly
-	p.updateAnnotations(service, lastKnownGoodEndpoint)
+	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, "active leader election", *leaderElectionActive)
+		p.provider.GetLabel(), "service name", service.Name, "namespace", service.Namespace, "endpoints", len(endpoints), "last endpoint", *lastKnownGoodEndpoint)
 
 	return false, nil
 }
 
-func (p *Processor) Delete(service *v1.Service, id string) error {
-	if err := p.worker.delete(service, id); err != nil {
+func (p *Processor) Delete(ctx context.Context, service *v1.Service, id string) error {
+	if err := p.worker.delete(ctx, service, id); err != nil {
 		return fmt.Errorf("[%s] error deleting service: %w", p.provider.GetLabel(), err)
 	}
 	return nil
 }
 
-func (p *Processor) updateLastKnownGoodEndpoint(lastKnownGoodEndpoint *string, endpoints []string, service *v1.Service, leaderElectionActive *bool, cancel context.CancelFunc) {
+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 = endpoints[0]
+		*lastKnownGoodEndpoint = ep
 		return
 	}
 
@@ -121,45 +168,131 @@ func (p *Processor) updateLastKnownGoodEndpoint(lastKnownGoodEndpoint *string, e
 	}
 	// If the last endpoint no longer exists, we cancel our leader Election, and set another endpoint as last known good
 	if !stillExists {
-		p.worker.removeEgress(service, lastKnownGoodEndpoint)
-		if *leaderElectionActive && (p.config.EnableServicesElection || p.config.EnableLeaderElection) {
-			log.Warn("existing endpoint has been removed, restarting leaderElection", "provider", p.provider.GetLabel(), "endpoint", *lastKnownGoodEndpoint)
-			// Stop the existing leaderElection
-			cancel()
-			// disable last leaderElection flag
-			*leaderElectionActive = false
+		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 = endpoints[0]
+		*lastKnownGoodEndpoint = ep
 	}
 }
 
-func (p *Processor) updateAnnotations(service *v1.Service, lastKnownGoodEndpoint *string) {
+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" {
-		activeEndpointAnnotation := kubevip.ActiveEndpoint
+		ip := net.ParseIP(*lastKnownGoodEndpoint)
 
-		if !p.config.EnableEndpoints && p.provider.GetProtocol() == string(discoveryv1.AddressTypeIPv6) {
-			activeEndpointAnnotation = kubevip.ActiveEndpointIPv6
+		// 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)
+			}
 		}
-		service.Annotations[activeEndpointAnnotation] = *lastKnownGoodEndpoint
 	}
 }
 
-func startLeaderElection(ctx context.Context, leaderElectionActive *bool, service *v1.Service, serviceFunc func(context.Context, *v1.Service) error) {
+func (p *Processor) startLeaderElection(svcCtx *servicecontext.Context, service *v1.Service, serviceFunc func(*servicecontext.Context, *v1.Service, *sync.WaitGroup, bool) error, wg *sync.WaitGroup) {
 	// This is a blocking function, that will restart (in the event of failure)
 	for {
-		// if the context isn't cancelled restart
-		if ctx.Err() != context.Canceled {
-			*leaderElectionActive = true
-			err := serviceFunc(ctx, service)
-			if err != nil {
-				log.Error(err.Error())
+		select {
+		case <-svcCtx.Ctx.Done():
+			return
+		default:
+			leaseNamespace, serviceLease := lease.ServiceName(service)
+			id := lease.NewID(p.config.LeaderElectionType, leaseNamespace, serviceLease)
+			l := p.leaseMgr.Get(id)
+			l.Lock()
+
+			if !l.Elected.Load() {
+				l.Unlock()
+				err := serviceFunc(svcCtx, service, wg, true)
+				if err != nil {
+					log.Error(err.Error())
+				}
+			} else {
+				l.Unlock()
+				time.Sleep(time.Millisecond * 200)
 			}
-			*leaderElectionActive = false
-		} else {
-			*leaderElectionActive = false
-			break
 		}
 	}
 }
+
+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

@@ -7,6 +7,7 @@ import (
 
 	"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"
 )
@@ -23,20 +24,19 @@ func newBGP(generic generic, bgpServer *bgp.Server) endpointWorker {
 	}
 }
 
-func (b *BGP) processInstance(ctx *servicecontext.Context, service *v1.Service, leaderElectionActive *bool) error {
+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 !ctx.IsNetworkConfigured(cluster.Network[i].IP()) {
+				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(cluster.Network[i].CIDR())
+					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)
-						ctx.ConfiguredNetworks.Store(cluster.Network[i].IP(), true)
-						*leaderElectionActive = true
+						svcCtx.ConfiguredNetworks.Store(cluster.Network[i].IP(), true)
 					}
 				}
 			}
@@ -45,20 +45,19 @@ func (b *BGP) processInstance(ctx *servicecontext.Context, service *v1.Service,
 	return nil
 }
 
-func (b *BGP) clear(ctx *servicecontext.Context, lastKnownGoodEndpoint *string, service *v1.Service, cancel context.CancelFunc, leaderElectionActive *bool) {
+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(cluster.Network[i].CIDR())
+					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)
-						ctx.ConfiguredNetworks.Delete(cluster.Network[i].IP())
-						*leaderElectionActive = false
+						svcCtx.ConfiguredNetworks.Delete(cluster.Network[i].IP())
 					}
 				}
 			}
@@ -66,14 +65,18 @@ func (b *BGP) clear(ctx *servicecontext.Context, lastKnownGoodEndpoint *string,
 		}
 	}
 
-	b.clearEgress(lastKnownGoodEndpoint, service, cancel, leaderElectionActive)
+	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) delete(service *v1.Service, id string) error {
+func (b *BGP) delete(ctx context.Context, service *v1.Service, id string) error {
 	// When no-leader-elecition mode
 	if !b.config.EnableServicesElection && !b.config.EnableLeaderElection {
 		// find all existing local endpoints
@@ -84,32 +87,32 @@ func (b *BGP) delete(service *v1.Service, id string) error {
 
 		// If there were local endpoints deleted
 		if len(endpoints) > 0 {
-			b.deleteAction(service)
+			b.deleteAction(ctx, service)
 		}
 	}
 
 	return nil
 }
 
-func (b *BGP) deleteAction(service *v1.Service) {
-	b.clearBGPHosts(service)
+func (b *BGP) deleteAction(ctx context.Context, service *v1.Service) {
+	b.clearBGPHosts(ctx, service)
 }
 
-func (b *BGP) clearBGPHosts(service *v1.Service) {
-	ClearBGPHosts(service, b.instances, b.bgpServer)
+func (b *BGP) clearBGPHosts(ctx context.Context, service *v1.Service) {
+	ClearBGPHosts(ctx, service, b.instances, b.bgpServer)
 }
 
-func (b *BGP) setInstanceEndpointsStatus(_ *v1.Service, _ []string) error {
+func (b *BGP) setInstanceEndpointsStatus(_ context.Context, _ *v1.Service, _ []string) error {
 	return nil
 }
 
-func ClearBGPHosts(service *v1.Service, instances *[]*instance.Instance, bgpServer *bgp.Server) {
+func ClearBGPHosts(ctx context.Context, service *v1.Service, instances *[]*instance.Instance, bgpServer *bgp.Server) {
 	if instance := instance.FindServiceInstance(service, *instances); instance != nil {
-		ClearBGPHostsByInstance(instance, bgpServer)
+		ClearBGPHostsByInstance(ctx, instance, bgpServer)
 	}
 }
 
-func ClearBGPHostsByInstance(instance *instance.Instance, bgpServer *bgp.Server) {
+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
@@ -117,7 +120,7 @@ func ClearBGPHostsByInstance(instance *instance.Instance, bgpServer *bgp.Server)
 	for _, cluster := range instance.Clusters {
 		for i := range cluster.Network {
 			network := cluster.Network[i]
-			err := bgpServer.DelHost(network.CIDR())
+			err := bgpServer.DelHost(ctx, network.CIDR(), lease.ServiceNamespacedName(instance.ServiceSnapshot))
 			if err != nil {
 				log.Error("[endpoint] error deleting BGP host", "err", err)
 			} else {

pkg/endpoints/endpoints_generic.go

@@ -10,24 +10,31 @@ import (
 	"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, leaderElectionActive *bool) error
-	clear(svcCtx *servicecontext.Context, lastKnownGoodEndpoint *string, service *v1.Service, cancel context.CancelFunc, leaderElectionActive *bool)
+	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)
-	delete(service *v1.Service, id string) error
-	setInstanceEndpointsStatus(service *v1.Service, endpoints []string) error
+	delete(ctx context.Context, service *v1.Service, id string) error
+	setInstanceEndpointsStatus(ctx context.Context, service *v1.Service, endpoints []string) error
 }
 
-func newEndpointWorker(config *kubevip.Config, provider providers.Provider, bgpServer *bgp.Server, instances *[]*instance.Instance) endpointWorker {
-	generic := newGeneric(config, provider, instances)
+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)
+		return newRoutingTable(generic, routeMgr)
 	}
 	if config.EnableBGP {
 		return newBGP(generic, bgpServer)
@@ -40,60 +47,49 @@ 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) generic {
+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, _ *bool) error {
+func (g *generic) processInstance(_ *servicecontext.Context, _ *v1.Service) error {
 	return nil
 }
 
-func (g *generic) clear(_ *servicecontext.Context, lastKnownGoodEndpoint *string, service *v1.Service, cancel context.CancelFunc, leaderElectionActive *bool) {
-	g.clearEgress(lastKnownGoodEndpoint, service, cancel, leaderElectionActive)
+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, cancel context.CancelFunc, leaderElectionActive *bool) {
+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)
+		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
-		if g.config.EnableServicesElection || g.config.EnableLeaderElection {
-			cancel() // stop services watcher
-		}
-		*leaderElectionActive = false
 	}
 }
 
-func (g *generic) getEndpoints(_ *v1.Service, id string) ([]string, error) {
-	return g.getLocalEndpoints(id)
-}
-
-func (g *generic) getLocalEndpoints(id string) ([]string, error) {
-	// Build endpoints
-	var endpoints []string
-	var err error
-	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) 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 !g.config.EnableLeaderElection && !g.config.EnableServicesElection &&
-		service.Spec.ExternalTrafficPolicy == v1.ServiceExternalTrafficPolicyTypeCluster {
+	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)
 		}
@@ -109,10 +105,10 @@ func (g *generic) getAllEndpoints(service *v1.Service, id string) ([]string, err
 func (g *generic) removeEgress(_ *v1.Service, _ *string) {
 }
 
-func (g *generic) delete(_ *v1.Service, _ string) error {
+func (g *generic) delete(_ context.Context, _ *v1.Service, _ string) error {
 	return nil
 }
 
-func (g *generic) setInstanceEndpointsStatus(_ *v1.Service, _ []string) error {
+func (g *generic) setInstanceEndpointsStatus(_ context.Context, _ *v1.Service, _ []string) error {
 	return nil
 }

pkg/endpoints/endpoints_routing_table.go

@@ -2,64 +2,46 @@ package endpoints
 
 import (
 	"context"
-	"errors"
 	"fmt"
 	"net"
-	"syscall"
+	"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"
-	"github.com/vishvananda/netlink"
 	v1 "k8s.io/api/core/v1"
 )
 
 type RoutingTable struct {
 	generic
+	mtx      sync.Mutex
+	routeMgr *route.Manager
 }
 
-func newRoutingTable(generic generic) endpointWorker {
+func newRoutingTable(generic generic, routeMgr *route.Manager) endpointWorker {
 	return &RoutingTable{
-		generic: generic,
+		generic:  generic,
+		routeMgr: routeMgr,
 	}
 }
 
-func (rt *RoutingTable) processInstance(ctx *servicecontext.Context, service *v1.Service, leaderElectionActive *bool) error {
-	instance := instance.FindServiceInstance(service, *rt.instances)
-	if instance != nil {
-		for _, cluster := range instance.Clusters {
+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 !ctx.IsNetworkConfigured(cluster.Network[i].IP()) && cluster.Network[i].HasEndpoints() {
-					err := cluster.Network[i].AddRoute(false)
-					if err != nil {
-						if errors.Is(err, syscall.EEXIST) {
-							// If route exists, but protocol is not set (e.g. the route was created by the older version
-							// of kube-vip) try to update it if necessary
-							isUpdated, err := cluster.Network[i].UpdateRoutes()
-							if err != nil {
-								return fmt.Errorf("[%s] error updating existing routes: %w", rt.provider.GetLabel(), err)
-							}
-							if isUpdated {
-								log.Info("updated 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)
-							} else {
-								log.Info("route already present", "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)
-							}
-						} else {
-							// If other error occurs, return error
-							return fmt.Errorf("[%s] error adding route: %s", rt.provider.GetLabel(), err.Error())
-						}
+				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)
-						ctx.ConfiguredNetworks.Store(cluster.Network[i].IP(), true)
-						*leaderElectionActive = true
+						svcCtx.ConfiguredNetworks.Store(cluster.Network[i].IP(), true)
 					}
 				}
 			}
@@ -69,9 +51,11 @@ func (rt *RoutingTable) processInstance(ctx *servicecontext.Context, service *v1
 	return nil
 }
 
-func (rt *RoutingTable) clear(svcCtx *servicecontext.Context, lastKnownGoodEndpoint *string, service *v1.Service, cancel context.CancelFunc, leaderElectionActive *bool) {
-	if !rt.config.EnableServicesElection && !rt.config.EnableLeaderElection {
-		if errs := ClearRoutes(service, rt.instances); len(errs) == 0 {
+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 {
@@ -80,7 +64,11 @@ func (rt *RoutingTable) clear(svcCtx *servicecontext.Context, lastKnownGoodEndpo
 		}
 	}
 
-	rt.clearEgress(lastKnownGoodEndpoint, service, cancel, leaderElectionActive)
+	rt.clearEgress(lastKnownGoodEndpoint, service)
+
+	if svcCtx.LeaderCancel != nil {
+		svcCtx.LeaderCancel()
+	}
 }
 
 func (rt *RoutingTable) getEndpoints(service *v1.Service, id string) ([]string, error) {
@@ -94,7 +82,7 @@ func (rt *RoutingTable) removeEgress(service *v1.Service, lastKnownGoodEndpoint
 	}
 }
 
-func (rt *RoutingTable) delete(service *v1.Service, id string) error {
+func (rt *RoutingTable) delete(_ context.Context, service *v1.Service, id string) error {
 	// When no-leader-elecition mode
 	if !rt.config.EnableServicesElection && !rt.config.EnableLeaderElection {
 		// find all existing local endpoints
@@ -113,23 +101,27 @@ func (rt *RoutingTable) delete(service *v1.Service, id string) error {
 }
 
 func (rt *RoutingTable) deleteAction(service *v1.Service) {
-	ClearRoutes(service, rt.instances)
+	ClearRoutes(service, rt.instances, rt.routeMgr)
 }
 
-func (rt *RoutingTable) setInstanceEndpointsStatus(service *v1.Service, endpoints []string) error {
-	instance := instance.FindServiceInstanceWithTimeout(service, *rt.instances)
-	if instance == nil {
+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 instance.Clusters {
+		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)
-				}
-				// check if endpoint are available and are of same IP family as service
-				if len(endpoints) > 0 && ((net.ParseIP(c.Network[n].IP()).To4() == nil) == (net.ParseIP(endpoints[0]).To4() == nil)) {
-					c.Network[n].SetHasEndpoints(true)
+				} 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
+						}
+					}
 				}
 			}
 		}
@@ -138,52 +130,32 @@ func (rt *RoutingTable) setInstanceEndpointsStatus(service *v1.Service, endpoint
 	return nil
 }
 
-func ClearRoutes(service *v1.Service, instances *[]*instance.Instance) []error {
+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)
+		clearErrs := ClearRoutesByInstance(service, svcInst, instances, routeMgr)
 		errs = append(errs, clearErrs...)
 	}
 	return errs
 }
 
-func ClearRoutesByInstance(service *v1.Service, svcInst *instance.Instance, instances *[]*instance.Instance) []error {
+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 {
-			route := cluster.Network[i].PrepareRoute()
-			// check if route we are about to delete is not referenced by more than one service
-			if CountRouteReferences(route, instances) <= 1 {
-				err := cluster.Network[i].DeleteRoute()
-				if err != nil && !errors.Is(err, syscall.ESRCH) {
-					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())
+			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
 }
-
-func CountRouteReferences(route *netlink.Route, instances *[]*instance.Instance) int {
-	cnt := 0
-	for _, instance := range *instances {
-		for _, cluster := range instance.Clusters {
-			for n := range cluster.Network {
-				if cluster.Network[n].HasEndpoints() {
-					r := cluster.Network[n].PrepareRoute()
-					if r.Dst.String() == route.Dst.String() {
-						cnt++
-					}
-				}
-			}
-		}
-	}
-	return cnt
-}

pkg/endpoints/endpoints_wireguard.go

@@ -0,0 +1,289 @@
+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)
+}
+
+// delete removes all DNAT rules for a service
+func (w *wireguardWorker) delete(ctx context.Context, service *v1.Service, id string) error {
+	log.Info("[wireguard] deleting DNAT rules for service", "service", service.Name, "namespace", service.Namespace)
+
+	w.clear(nil, nil, service)
+	return nil
+}
+
+// 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

@@ -97,11 +97,11 @@ func (ep *Endpoints) GetLocalEndpoints(id string, _ *kubevip.Config) ([]string,
 	return localEndpoints, nil
 }
 
-func (ep *Endpoints) UpdateServiceAnnotation(endpoint string, _ string, service *v1.Service, clientSet *kubernetes.Clientset) error {
+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(context.TODO(), service.Name, metav1.GetOptions{})
+		currentService, err := clientSet.CoreV1().Services(service.Namespace).Get(ctx, service.Name, metav1.GetOptions{})
 		if err != nil {
 			return err
 		}
@@ -113,7 +113,7 @@ func (ep *Endpoints) UpdateServiceAnnotation(endpoint string, _ string, service
 
 		currentServiceCopy.Annotations[kubevip.ActiveEndpoint] = endpoint
 
-		_, err = clientSet.CoreV1().Services(currentService.Namespace).Update(context.TODO(), currentServiceCopy, metav1.UpdateOptions{})
+		_, 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
@@ -132,6 +132,15 @@ func (ep *Endpoints) GetLabel() string {
 	return ep.label
 }
 
-func (ep *Endpoints) GetProtocol() string {
-	return ""
+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

@@ -20,8 +20,10 @@ import (
 )
 
 type Endpointslices struct {
-	label     string
-	endpoints *discoveryv1.EndpointSlice
+	label       string
+	endpointsv4 []discoveryv1.Endpoint
+	endpointsv6 []discoveryv1.Endpoint
+	ports       []discoveryv1.EndpointPort
 }
 
 func NewEndpointslices() Provider {
@@ -56,27 +58,42 @@ func (ep *Endpointslices) LoadObject(endpoints runtime.Object, cancel context.Ca
 		cancel()
 		return fmt.Errorf("[%s] error casting endpoints to v1.Endpoints struct", ep.label)
 	}
-	ep.endpoints = eps
+
+	if eps.AddressType == discoveryv1.AddressTypeIPv6 {
+		ep.endpointsv6 = eps.Endpoints
+	} else {
+		ep.endpointsv4 = eps.Endpoints
+	}
+
+	// Store ports for resolving named ports
+	ep.ports = eps.Ports
+
 	return nil
 }
 
 func (ep *Endpointslices) GetAllEndpoints() ([]string, error) {
 	result := []string{}
-	for _, ep := range ep.endpoints.Endpoints {
-		result = append(result, ep.Addresses...)
+	for _, e := range ep.endpointsv4 {
+		result = append(result, e.Addresses...)
+	}
+	for _, e := range ep.endpointsv6 {
+		result = append(result, e.Addresses...)
 	}
 	return result, nil
 }
 
 func (ep *Endpointslices) GetLocalEndpoints(id string, _ *kubevip.Config) ([]string, error) {
 	var localEndpoints []string
-	for _, endpoint := range ep.endpoints.Endpoints {
+	tmpEps := []discoveryv1.Endpoint{}
+
+	tmpEps = append(tmpEps, ep.endpointsv4...)
+	tmpEps = append(tmpEps, ep.endpointsv6...)
+
+	for _, endpoint := range tmpEps {
 		if endpoint.Conditions.Serving == nil || !*endpoint.Conditions.Serving {
 			continue
 		}
 		for _, address := range endpoint.Addresses {
-			log.Debug("processing endpoint", "provider", ep.label, "ip", address)
-
 			// 1. Compare the Nodename
 			if endpoint.NodeName != nil && id == *endpoint.NodeName {
 				if endpoint.Hostname != nil {
@@ -98,11 +115,11 @@ func (ep *Endpointslices) GetLocalEndpoints(id string, _ *kubevip.Config) ([]str
 	return localEndpoints, nil
 }
 
-func (ep *Endpointslices) UpdateServiceAnnotation(endpoint, endpointIPv6 string, service *v1.Service, clientSet *kubernetes.Clientset) error {
+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(context.TODO(), service.Name, metav1.GetOptions{})
+		currentService, err := clientSet.CoreV1().Services(service.Namespace).Get(ctx, service.Name, metav1.GetOptions{})
 		if err != nil {
 			return err
 		}
@@ -115,7 +132,7 @@ func (ep *Endpointslices) UpdateServiceAnnotation(endpoint, endpointIPv6 string,
 		currentServiceCopy.Annotations[kubevip.ActiveEndpoint] = endpoint
 		currentServiceCopy.Annotations[kubevip.ActiveEndpointIPv6] = endpointIPv6
 
-		_, err = clientSet.CoreV1().Services(currentService.Namespace).Update(context.TODO(), currentServiceCopy, metav1.UpdateOptions{})
+		_, 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
@@ -134,6 +151,13 @@ func (ep *Endpointslices) GetLabel() string {
 	return ep.label
 }
 
-func (ep *Endpointslices) GetProtocol() string {
-	return string(ep.endpoints.AddressType)
+func (ep *Endpointslices) ResolvePort(servicePort v1.ServicePort) int32 {
+	return ResolvePortWithLookup(servicePort, func(name string) int32 {
+		for _, p := range ep.ports {
+			if p.Name != nil && *p.Name == name && p.Port != nil {
+				return *p.Port
+			}
+		}
+		return 0
+	})
 }

pkg/endpoints/providers/interface.go

@@ -16,7 +16,24 @@ type Provider interface {
 	GetAllEndpoints() ([]string, error)
 	GetLocalEndpoints(string, *kubevip.Config) ([]string, error)
 	GetLabel() string
-	UpdateServiceAnnotation(string, string, *v1.Service, *kubernetes.Clientset) error
+	UpdateServiceAnnotation(context.Context, string, string, *v1.Service, *kubernetes.Clientset) error
 	LoadObject(runtime.Object, context.CancelFunc) error
-	GetProtocol() string
+	// 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/etcd/election.go

@@ -2,7 +2,9 @@ package etcd
 
 import (
 	"context"
+	"fmt"
 	"hash/fnv"
+	"sync"
 	"time"
 
 	log "log/slog"
@@ -64,10 +66,11 @@ 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.
@@ -116,7 +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
@@ -185,7 +193,7 @@ watcher:
 	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.Error("Failed trying to become the leader", "err", err)
 		// Resign just in case we acquired leadership just before failing
@@ -204,7 +212,9 @@ 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

pkg/etcd/election_test.go

@@ -20,13 +20,20 @@ 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,
@@ -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()
 
@@ -85,7 +108,7 @@ func TestRunElectionWithTwoMembersAndReelection(t *testing.T) {
 		LeaseDurationSeconds: 1,
 	}
 
-	member1Ctx, _ := context.WithCancel(ctx)
+	member1Ctx, cancelMember1 := context.WithCancel(ctx)
 	member2Ctx, cancelMember2 := context.WithCancel(ctx)
 
 	config1 := configBase
@@ -101,6 +124,7 @@ func TestRunElectionWithTwoMembersAndReelection(t *testing.T) {
 		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
@@ -129,6 +153,7 @@ func TestRunElectionWithTwoMembersAndReelection(t *testing.T) {
 	}()
 
 	wg.Wait()
+
 }
 
 func baseCallbacksForName(name string) etcd.LeaderCallbacks {

pkg/etcd/etcd_suite_test.go

@@ -29,7 +29,8 @@ 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

@@ -1,13 +1,14 @@
 package instance
 
 import (
+	"context"
 	"fmt"
 	"net"
+	"slices"
 	"strconv"
 	"strings"
-	"time"
+	"sync"
 
-	"log/slog"
 	log "log/slog"
 
 	"github.com/vishvananda/netlink"
@@ -17,6 +18,8 @@ import (
 	"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"
@@ -31,18 +34,36 @@ type Instance struct {
 	Clusters []*cluster.Cluster
 
 	// Service uses DHCP
-	IsDHCP              bool
+	IsDHCPv4            bool
+	IsDHCPv6            bool
 	DHCPInterface       string
 	DHCPInterfaceHwaddr string
 	DHCPInterfaceIP     string
+	DHCPInterfaceIPv4   string
+	DHCPInterfaceIPv6   string
 	DHCPHostname        string
-	DHCPClient          *vip.DHCPClient
+	DHCPv4Client        vip.DHCPClient
+	DHCPv6Client        vip.DHCPClient
+	macvlanName         string
+
+	// 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 {
@@ -50,17 +71,42 @@ type Port struct {
 	Type string
 }
 
-func NewInstance(svc *v1.Service, config *kubevip.Config, intfMgr *networkinterface.Manager, arpMgr *arp.Manager) (*Instance, error) {
-	instanceAddresses, _ := FetchServiceAddresses(svc)
+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.ServiceInterface] // If the service has a specific interface defined, then use it
+
+		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 {
@@ -131,7 +177,7 @@ func NewInstance(svc *v1.Service, config *kubevip.Config, intfMgr *networkinterf
 				if cidrs[0] != "" && cidrs[0] != kubevip.Auto {
 					subnet = cidrs[0]
 				} else {
-					subnet = "32"
+					subnet = strconv.Itoa(vip.DefaultMaskIPv4)
 				}
 			}
 		} else {
@@ -144,7 +190,7 @@ func NewInstance(svc *v1.Service, config *kubevip.Config, intfMgr *networkinterf
 				if len(cidrs) > 1 && cidrs[1] != "" && cidrs[1] != kubevip.Auto {
 					subnet = cidrs[1]
 				} else {
-					subnet = "128"
+					subnet = strconv.Itoa(vip.DefaultMaskIPv6)
 				}
 			}
 		}
@@ -164,6 +210,8 @@ func NewInstance(svc *v1.Service, config *kubevip.Config, intfMgr *networkinterf
 			ArpBroadcastRate:            config.ArpBroadcastRate,
 			EnableServiceSecurity:       config.EnableServiceSecurity,
 			DNSMode:                     config.DNSMode,
+			DHCPMode:                    config.DHCPMode,
+			DHCPBackoffAttempts:         config.DHCPBackoffAttempts,
 			DisableServiceUpdates:       config.DisableServiceUpdates,
 			EnableServicesElection:      config.EnableServicesElection,
 			PreserveVIPOnLeadershipLoss: config.PreserveVIPOnLeadershipLoss,
@@ -173,23 +221,88 @@ func NewInstance(svc *v1.Service, config *kubevip.Config, intfMgr *networkinterf
 		})
 	}
 
-	// Create new service
-	instance := &Instance{
-		//UID:             instanceUID,
-		//VIPs:            instanceAddresses,
-		ServiceSnapshot: svc,
+	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,
+			ArpBroadcastRate:       config.ArpBroadcastRate,
+			EnableServiceSecurity:  config.EnableServiceSecurity,
+			DNSMode:                config.DNSMode,
+			DHCPMode:               config.DHCPMode,
+			DisableServiceUpdates:  config.DisableServiceUpdates,
+			EnableServicesElection: config.EnableServicesElection,
+			KubernetesLeaderElection: kubevip.KubernetesLeaderElection{
+				EnableLeaderElection: config.EnableLeaderElection,
+			},
+		})
 	}
-	// for _, port := range svc.Spec.Ports {
-	// 	instance.ExternalPorts = append(instance.ExternalPorts, Port{
-	// 		Port: uint16(port.Port), //nolint
-	// 		Type: string(port.Protocol),
-	// 	})
-	// }
 
 	if svc.Annotations != nil {
 		instance.DHCPInterfaceHwaddr = svc.Annotations[kubevip.HwAddrKey]
-		instance.DHCPInterfaceIP = svc.Annotations[kubevip.RequestedIP]
+		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]
 	}
 
 	configPorts := make([]kubevip.Port, 0)
@@ -212,29 +325,82 @@ func NewInstance(svc *v1.Service, config *kubevip.Config, intfMgr *networkinterf
 	// Create Add configuration to the new service
 	instance.VIPConfigs = newVips
 
-	// If this was purposely created with the address 0.0.0.0,
+	// 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
-	// 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
-		}
+	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
+			}
+		}
 
-	for _, vipConfig := range instance.VIPConfigs {
-		c, err := cluster.InitCluster(vipConfig, false, intfMgr, arpMgr)
+		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)
+				}
+			}
+		}
+
+		c, err := cluster.InitCluster(instance.VIPConfigs[i], false, intfMgr, arpMgr, routeMgr, nodeLabelMgr)
 		if err != nil {
-			log.Error("Failed to add Service %s/%s", svc.Namespace, svc.Name)
+			log.Error("failed to add service", "err", err)
 			return nil, err
 		}
 
@@ -243,7 +409,7 @@ func NewInstance(svc *v1.Service, config *kubevip.Config, intfMgr *networkinterf
 		}
 
 		instance.Clusters = append(instance.Clusters, c)
-		log.Info("(svcs) adding VIP", "ip", vipConfig.VIP, "interface", vipConfig.Interface, "namespace", svc.Namespace, "name", svc.Name)
+		log.Info("(svcs) adding VIP", "ip", instance.VIPConfigs[i].VIP, "interface", instance.VIPConfigs[i].Interface, "namespace", svc.Namespace, "name", svc.Name)
 	}
 
 	return instance, nil
@@ -306,17 +472,72 @@ func getAutoInterfaceName(link netlink.Link, defaultInterface string) string {
 	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))
+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
+		}
 	}
-	parent, err := netlink.LinkByName(i.VIPConfigs[0].Interface)
+
+	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)
 	}
 
-	// Generate name from UID
-	interfaceName := fmt.Sprintf("vip-%s", i.ServiceSnapshot.UID[0:8])
+	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)
@@ -361,34 +582,61 @@ func (i *Instance) startDHCP() error {
 		log.Info("Using existing macvlan interface for DHCP", "interface", interfaceName)
 	}
 
-	// Default rp_filter setting (https://github.com/kube-vip/kube-vip/issues/1170)
-	rpfilterSetting := "0"
+	var initRebootFlag bool
+	ip := net.ParseIP(i.VIPConfigs[index].VIP)
 
-	// 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 {
-			slog.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]
+	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 {
-				slog.Error("[DHCP] rp_filter value not within range 0-2", "value", rpFilter)
+				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 {
-		slog.Error("[DHCP] unable to write rp_filter", "value", rpfilterSetting, "err", err)
-	}
-	var initRebootFlag bool
-	if i.DHCPInterfaceIP != "" {
-		initRebootFlag = true
-	}
+		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)
+		}
 
-	client := vip.NewDHCPClient(iface, initRebootFlag, i.DHCPInterfaceIP)
+		if i.DHCPInterfaceIPv4 != "" {
+			initRebootFlag = true
+		}
+
+		client = vip.NewDHCPv4Client(iface, initRebootFlag, i.DHCPInterfaceIPv4, backoffAttempts)
+
+		// 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 != "" {
@@ -396,15 +644,15 @@ func (i *Instance) startDHCP() error {
 		client.WithHostName(i.DHCPHostname)
 	}
 
-	go client.Start()
+	wg.Go(func() {
+		if err := client.Start(ctx); err != nil {
+			log.Error("[instance] DHCP client error: %w")
+		}
+	})
 
-	// 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
 }
@@ -485,7 +733,6 @@ func FetchServiceAddresses(s *v1.Service) ([]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 {
-		log.Debug("saved service", "instance", i, "UID", instances[i].ServiceSnapshot.UID)
 		if instances[i].ServiceSnapshot.UID == svc.UID {
 			return instances[i]
 		}
@@ -493,24 +740,3 @@ func FindServiceInstance(svc *v1.Service, instances []*Instance) *Instance {
 	log.Debug("instance not found", "namespace", svc.Namespace, "name", svc.Name, "UID", svc.UID)
 	return nil
 }
-
-func FindServiceInstanceWithTimeout(svc *v1.Service, instances []*Instance) *Instance {
-	log.Debug("finding service with timeout", "namespace", svc.Namespace, "name", svc.Name, "UID", svc.UID)
-	ticker := time.NewTicker(time.Millisecond * 200)
-	defer ticker.Stop()
-	to := time.NewTimer(time.Second * 60)
-	defer to.Stop()
-	for {
-		select {
-		case <-to.C:
-			return nil
-		case <-ticker.C:
-			for i := range instances {
-				log.Debug("saved service", "instance", i, "UID", instances[i].ServiceSnapshot.UID)
-				if instances[i].ServiceSnapshot.UID == svc.UID {
-					return instances[i]
-				}
-			}
-		}
-	}
-}

pkg/kubevip/annotations.go

@@ -52,12 +52,25 @@ const (
 	// 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"
+
+	// 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"
 )

pkg/kubevip/config_bgp.go

@@ -16,6 +16,7 @@ import (
 type BGPPeer struct {
 	Address      string
 	Port         uint16
+	Interface    string
 	AS           uint32
 	Password     string
 	MultiHop     bool
@@ -65,6 +66,7 @@ func ParseBGPPeerConfig(config string) (bgpPeers []BGPPeer, err error) {
 			continue
 		}
 		isV6Peer := peerStr[0] == '['
+		isUnnumberedPeer := strings.HasPrefix(peerStr, "unnumbered:")
 
 		address := ""
 		if isV6Peer {
@@ -74,20 +76,29 @@ func ParseBGPPeerConfig(config string) (bgpPeers []BGPPeer, err error) {
 			}
 			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 {
+		if len(peer) < 2 && !isUnnumberedPeer {
 			return nil, fmt.Errorf("mandatory peering params <host>:<AS> incomplete")
 		}
 
-		if !isV6Peer {
+		iface := ""
+		if isUnnumberedPeer {
+			iface = peer[0]
+		} else if !isV6Peer {
 			address = peer[0]
 		}
 
-		ASNumber, err := strconv.ParseUint(peer[1], 10, 32)
-		if err != nil {
-			return nil, fmt.Errorf("BGP Peer AS format error [%s]", peer[1])
+		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])
+			}
 		}
 
 		password := ""
@@ -96,10 +107,10 @@ func ParseBGPPeerConfig(config string) (bgpPeers []BGPPeer, err error) {
 		}
 
 		multiHop := false
-		if len(peer) >= 4 {
+		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[1])
+				return nil, fmt.Errorf("BGP MultiHop format error (true/false) [%s]", peer[3])
 			}
 		}
 
@@ -107,9 +118,9 @@ func ParseBGPPeerConfig(config string) (bgpPeers []BGPPeer, err error) {
 		if len(peer) >= 5 {
 			port, err = strconv.ParseUint(peer[4], 10, 16)
 			if err != nil {
-				return nil, fmt.Errorf("BGP Peer AS format error [%s]", peer[1])
+				return nil, fmt.Errorf("BGP Peer Port format error [%s]", peer[4])
 			}
-		} else {
+		} else if !isUnnumberedPeer {
 			port = 179
 		}
 
@@ -119,6 +130,9 @@ func ParseBGPPeerConfig(config string) (bgpPeers []BGPPeer, err error) {
 			configData := strings.Split(config[1], ";")
 			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]
@@ -133,9 +147,11 @@ func ParseBGPPeerConfig(config string) (bgpPeers []BGPPeer, err error) {
 		}
 
 		peerConfig := BGPPeer{
-			Address:      address,
+			Address: address,
+			//nolint:gosec // previously parsed into uint32
 			AS:           uint32(ASNumber),
 			Port:         uint16(port),
+			Interface:    iface,
 			Password:     password,
 			MultiHop:     multiHop,
 			MpbgpNexthop: mpbgpNexthop,

pkg/kubevip/config_bgp_test.go

@@ -1,10 +1,8 @@
-package bgp
+package kubevip
 
 import (
 	"reflect"
 	"testing"
-
-	"github.com/kube-vip/kube-vip/pkg/kubevip"
 )
 
 func TestParseBGPPeerConfig(t *testing.T) {
@@ -14,13 +12,13 @@ func TestParseBGPPeerConfig(t *testing.T) {
 	tests := []struct {
 		name         string
 		args         args
-		wantBgpPeers []kubevip.BGPPeer
+		wantBgpPeers []BGPPeer
 		wantErr      bool
 	}{
 		{
 			name: "IPv4, default port",
 			args: args{config: "192.168.0.10:65000::false,192.168.0.11:65000::false"},
-			wantBgpPeers: []kubevip.BGPPeer{
+			wantBgpPeers: []BGPPeer{
 				{Address: "192.168.0.10", Port: 179, AS: 65000, MultiHop: false},
 				{Address: "192.168.0.11", Port: 179, AS: 65000, MultiHop: false},
 			},
@@ -28,7 +26,7 @@ func TestParseBGPPeerConfig(t *testing.T) {
 		{
 			name: "IPv4, different port",
 			args: args{config: "192.168.0.10:65000::false:180,192.168.0.11:65000::false:190"},
-			wantBgpPeers: []kubevip.BGPPeer{
+			wantBgpPeers: []BGPPeer{
 				{Address: "192.168.0.10", Port: 180, AS: 65000, MultiHop: false},
 				{Address: "192.168.0.11", Port: 190, AS: 65000, MultiHop: false},
 			},
@@ -36,14 +34,51 @@ func TestParseBGPPeerConfig(t *testing.T) {
 		{
 			name: "IPv6, multi-protocol",
 			args: args{config: "[fd00:1111:2222:3333:c7d9:7235:6bf7:5d52]:65501::false/mpbgp_nexthop=auto_sourceif"},
-			wantBgpPeers: []kubevip.BGPPeer{
+			wantBgpPeers: []BGPPeer{
 				{Address: "fd00:1111:2222:3333:c7d9:7235:6bf7:5d52", Port: 179, AS: 65501, MultiHop: false, MpbgpNexthop: "auto_sourceif"},
 			},
 		},
+		{
+			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},
+			},
+		},
+		{
+			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},
+			},
+		},
+		{
+			name: "Unnumbered",
+			args: args{config: "unnumbered:eth0,unnumbered:eth1:65000::true/mpbgp_nexthop=auto_sourceif"},
+			wantBgpPeers: []BGPPeer{
+				{Interface: "eth0", MultiHop: false},
+				{Interface: "eth1", AS: 65000, MultiHop: true, MpbgpNexthop: "auto_sourceif"},
+			},
+		},
+		{
+			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 := kubevip.ParseBGPPeerConfig(tt.args.config)
+			gotBgpPeers, err := ParseBGPPeerConfig(tt.args.config)
 			if (err != nil) != tt.wantErr {
 				t.Errorf("ParseBGPPeerConfig() error = %v, wantErr %v", err, tt.wantErr)
 				return

pkg/kubevip/config_environment.go

@@ -11,6 +11,7 @@ import (
 	"strings"
 
 	"github.com/kube-vip/kube-vip/pkg/detector"
+	"github.com/kube-vip/kube-vip/pkg/utils"
 	"sigs.k8s.io/yaml"
 )
 
@@ -25,7 +26,7 @@ func ParseEnvironment(c *Config) error {
 	if env != "" {
 		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 = int32(logLevel)
 	}
@@ -301,9 +302,6 @@ func ParseEnvironment(c *Config) error {
 			return err
 		}
 		c.PreserveVIPOnLeadershipLoss = b
-	} else {
-		// Default to false for backward compatibility
-		c.PreserveVIPOnLeadershipLoss = false
 	}
 
 	// Wireguard Mode
@@ -379,6 +377,30 @@ func ParseEnvironment(c *Config) error {
 		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 != "" {
@@ -517,6 +539,40 @@ func ParseEnvironment(c *Config) error {
 		c.BGPConfig.KeepaliveInterval = u64
 	}
 
+	// BGP health check options
+	env = os.Getenv(controlPlaneHealthCheckAddress)
+	if 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)
@@ -868,6 +924,9 @@ func mergeConfigValues(baseConfig, fileConfig *Config) {
 	// 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
@@ -878,7 +937,17 @@ func mergeConfigValues(baseConfig, fileConfig *Config) {
 		baseConfig.DNSMode = fileConfig.DNSMode
 	}
 
-	// Health check configuration
+	// 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
 	}
@@ -965,3 +1034,22 @@ func mergeLeaderElectionConfig(base, file *KubernetesLeaderElection) {
 		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_envvar.go

@@ -18,16 +18,16 @@ const (
 	// 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)
@@ -112,6 +112,16 @@ 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"
@@ -212,6 +222,12 @@ 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"
 

pkg/kubevip/config_generator.go

@@ -167,10 +167,10 @@ func GenerateRoleBinding(rolebinding bool, saCfg *applyCoreV1.ServiceAccountAppl
 }
 
 // generatePodSpec will take a kube-vip config and generate a Pod spec
-func generatePodSpec(c *Config, image, 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 {
-		panic(fmt.Errorf("Cannot parse %q: %w", image, err))
+		return nil, fmt.Errorf("cannot parse %q: %w", image, err)
 	}
 	command := "manager"
 
@@ -255,6 +255,28 @@ func generatePodSpec(c *Config, image, imageVersion string, inCluster bool) *cor
 		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{
@@ -485,6 +507,40 @@ func generatePodSpec(c *Config, image, imageVersion string, inCluster bool) *cor
 
 	}
 
+	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{
@@ -565,7 +621,7 @@ func generatePodSpec(c *Config, image, imageVersion string, inCluster bool) *cor
 
 	var securityContext *corev1.SecurityContext
 	if c.LoadBalancerForwardingMethod == "masquerade" {
-		var privileged = true
+		privileged := true
 		securityContext = &corev1.SecurityContext{
 			Privileged: &privileged,
 		}
@@ -583,6 +639,16 @@ func generatePodSpec(c *Config, image, imageVersion string, inCluster bool) *cor
 		}
 	}
 
+	if c.PreserveVIPOnLeadershipLoss {
+		preserveVIPOnLeadershipLoss := []corev1.EnvVar{
+			{
+				Name:  vipPreserveOnLeadershipLoss,
+				Value: strconv.FormatBool(c.PreserveVIPOnLeadershipLoss),
+			},
+		}
+		newEnvironment = append(newEnvironment, preserveVIPOnLeadershipLoss...)
+	}
+
 	newManifest := &corev1.Pod{
 		TypeMeta: metav1.TypeMeta{
 			Kind:       "Pod",
@@ -637,18 +703,24 @@ func generatePodSpec(c *Config, image, imageVersion string, inCluster bool) *cor
 		newManifest.Spec.HostAliases = append(newManifest.Spec.HostAliases, hostAlias)
 	}
 
-	return newManifest
+	return newManifest, nil
 }
 
 // GeneratePodManifestFromConfig will take a kube-vip config and generate a manifest
-func GeneratePodManifestFromConfig(c *Config, image, imageVersion string, inCluster bool) string {
-	newManifest := generatePodSpec(c, image, 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, image, 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 != "" {
@@ -657,7 +729,11 @@ func GenerateDaemonsetManifestFromConfig(c *Config, image, imageVersion string,
 		namespace = metav1.NamespaceSystem
 	}
 
-	podSpec := generatePodSpec(c, image, imageVersion, inCluster).Spec
+	pod, err := generatePodSpec(c, image, imageVersion, inCluster)
+	if err != nil {
+		return "", err
+	}
+
 	newManifest := &appv1.DaemonSet{
 		TypeMeta: metav1.TypeMeta{
 			Kind:       "DaemonSet",
@@ -684,7 +760,7 @@ func GenerateDaemonsetManifestFromConfig(c *Config, image, imageVersion string,
 						"app.kubernetes.io/version": imageVersion,
 					},
 				},
-				Spec: podSpec,
+				Spec: pod.Spec,
 			},
 		},
 	}
@@ -734,5 +810,5 @@ func GenerateDaemonsetManifestFromConfig(c *Config, image, imageVersion string,
 	delete(m, "status")
 
 	b, _ = yaml.Marshal(m)
-	return string(b)
+	return string(b), nil
 }

pkg/kubevip/config_manager.go

@@ -12,6 +12,14 @@ const (
 	Auto = "auto"
 )
 
+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 {

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

@@ -132,6 +132,10 @@ type Config struct {
 	BGPPeerConfig BGPPeer
 	BGPPeers      []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"`
 
@@ -158,6 +162,16 @@ 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"`
 
@@ -189,6 +203,9 @@ type Config struct {
 
 	// ConfigFile defines the path to a JSON/YAML configuration file
 	ConfigFile string `yaml:"configFile"`
+
+	// DHCPBackoffAttempts defaines how many times will DHCP client try to obtain address (unlimited when 0)
+	DHCPBackoffAttempts uint `yaml:"dhcpBackoffAttempts"`
 }
 
 // KubernetesLeaderElection defines all of the settings for Kubernetes KubernetesLeaderElection
@@ -220,12 +237,29 @@ 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 +1,35 @@
+package kubevip
+
+import (
+	"fmt"
+	"net/url"
+	"strings"
+)
+
+// 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
+	}
+
+	return nil
+}
+
+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 +1,32 @@
+package kubevip
+
+import (
+	"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)
+			}
+		})
+	}
+}

pkg/kubevip/constants.go

@@ -2,4 +2,6 @@ package kubevip
 
 const (
 	LBClassName = "kube-vip.io/kube-vip-class"
+
+	DefaultDHCPBackoffAttempts = 3
 )

pkg/kubevip/labels.go

@@ -1,7 +1,22 @@
 package kubevip
 
-const (
-	// label used on nodes, which announce the LoadBalancer IP
-	HasIP         = "kube-vip.io/has-ip"
-	HasIPJSONPath = `kube-vip.io~1has-ip`
+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

@@ -3,7 +3,11 @@ 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"
@@ -22,106 +26,209 @@ func NewManager() *Manager {
 	}
 }
 
-// Add adds lease or incerements counter if lease is alreay used.
-func (m *Manager) Add(service *v1.Service) (*Lease, bool) {
+// 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()
-	_, id := GetName(service)
-	if _, exist := m.leases[id]; !exist {
-		ctx, cancel := context.WithCancel(context.Background())
-		m.leases[id] = newLease(ctx, cancel)
-		return m.leases[id], true
+	if _, exists := m.leases[id.NamespacedName()]; !exists {
+		leaseCtx, leaseCancel := context.WithCancel(ctx)
+		m.leases[id.NamespacedName()] = newLease(leaseCtx, leaseCancel)
 	}
 
-	m.leases[id].increment()
-	return m.leases[id], false
+	return m.leases[id.NamespacedName()]
 }
 
-// Delete decrements lease counter and removes the lease if counter equals 0.
-func (m *Manager) Delete(service *v1.Service) {
+// Delete removes the lease and cancels it if the lease counter equals 0.
+func (m *Manager) Delete(id ID, objectName string) {
 	m.lock.Lock()
 	defer m.lock.Unlock()
-	_, id := GetName(service)
-	if _, exist := m.leases[id]; exist {
-		m.leases[id].decrement()
-		if m.leases[id].cnt < 1 {
-			delete(m.leases, id)
+	if _, exist := m.leases[id.NamespacedName()]; exist {
+		m.leases[id.NamespacedName()].delete(objectName)
+		if m.leases[id.NamespacedName()].cnt.Load() < 1 {
+			m.leases[id.NamespacedName()].Cancel()
+			delete(m.leases, id.NamespacedName())
 		}
 	}
 }
 
 // Get returns lease for the service.
-func (m *Manager) Get(service *v1.Service) *Lease {
+func (m *Manager) Get(id ID) *Lease {
 	m.lock.Lock()
 	defer m.lock.Unlock()
-	_, id := GetName(service)
 
-	if lease, exist := m.leases[id]; exist {
+	if lease, exist := m.leases[id.NamespacedName()]; exist {
 		return lease
 	}
 	return nil
 }
 
-// GetLeaderContext returns leder context for the service.
-func (m *Manager) GetLeaderContext(service *v1.Service) context.Context {
-	m.lock.Lock()
-	defer m.lock.Unlock()
-	_, id := GetName(service)
-	if _, ok := m.leases[id]; !ok {
-		return nil
-	}
-	return m.leases[id].Ctx
-}
-
 // Lease holds lease data.
 type Lease struct {
-	cnt     uint
-	Lock    *sync.Mutex
-	Ctx     context.Context
-	Cancel  context.CancelFunc
-	Started chan any
+	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,
-		cnt:     1,
-		Lock:    new(sync.Mutex),
 		Started: make(chan any),
 	}
 }
 
-func (l *Lease) increment() {
-	l.Lock.Lock()
-	defer l.Lock.Unlock()
-	l.cnt++
+// 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.Load(name); !exists {
+		l.services.Store(name, true)
+		l.cnt.Add(1)
+		return true
+	}
+	return false
 }
 
-func (l *Lease) decrement() {
-	l.Lock.Lock()
-	defer l.Lock.Unlock()
-	if l.cnt == 0 {
-		return
-	}
-	l.cnt--
-	if l.cnt < 1 {
-		l.Cancel()
+// 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)
 	}
 }
 
-// GetName gets lease name and id for the service.
-func GetName(service *v1.Service) (string, string) {
-	serviceLease, exists := service.Annotations[kubevip.ServiceLease]
-	if !exists || serviceLease == "" {
-		serviceLease = fmt.Sprintf("kubevip-%s", service.Name)
-	}
-	serviceLeaseID := fmt.Sprintf("%s/%s", serviceLease, service.Namespace)
-	return serviceLease, serviceLeaseID
+func (l *Lease) Lock() {
+	l.Mtx.Lock()
+	l.locked = true
 }
 
-// UsesCommon checks if service uses common lease feature.
-func UsesCommon(service *v1.Service) bool {
-	_, common := service.Annotations[kubevip.ServiceLease]
-	return common
+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/lease/lease_test.go

@@ -0,0 +1,939 @@
+package lease
+
+import (
+	"context"
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	v1 "k8s.io/api/core/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+)
+
+func createTestService(name, namespace string, annotations map[string]string) *v1.Service {
+	return &v1.Service{
+		ObjectMeta: metav1.ObjectMeta{
+			Name:        name,
+			Namespace:   namespace,
+			Annotations: annotations,
+		},
+	}
+}
+
+func getSvcID(svc *v1.Service) ID {
+	namespace, name := ServiceName(svc)
+	id := NewID("kubernetes", namespace, name)
+	return id
+}
+
+func getSvcData(svc *v1.Service) (context.Context, ID) {
+	return context.TODO(), getSvcID(svc)
+}
+
+const serviceLeaseAnnotation = kubevip.ServiceLease
+
+// TestManager_Add_NewLease tests adding a new service with a new lease
+func TestManager_Add_NewLease(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("test-svc", "default", nil)
+
+	leaseID := mgr.Add(getSvcData(svc))
+	isNew := leaseID.Add(ServiceNamespacedName(svc))
+
+	if !isNew {
+		t.Error("expected isNew to be true for first Add")
+	}
+	if leaseID == nil {
+		t.Fatal("expected lease to be non-nil")
+	}
+	if leaseID.Ctx == nil {
+		t.Error("expected lease context to be non-nil")
+	}
+	if leaseID.Cancel == nil {
+		t.Error("expected lease cancel func to be non-nil")
+	}
+	if leaseID.Started == nil {
+		t.Error("expected lease Started channel to be non-nil")
+	}
+
+}
+
+// TestManager_Add_ExistingLease tests adding a service with an existing lease
+func TestManager_Add_ExistingLease(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("test-svc", "default", nil)
+
+	leaseID1 := mgr.Add(getSvcData(svc))
+	isNew1 := leaseID1.Add(ServiceNamespacedName(svc))
+
+	leaseID2 := mgr.Add(getSvcData(svc))
+	isNew2 := leaseID2.Add(ServiceNamespacedName(svc))
+
+	if !isNew1 {
+		t.Error("expected first Add to return isNew=true")
+	}
+	if isNew2 {
+		t.Error("expected second Add to return isNew=false")
+	}
+	if leaseID1 != leaseID2 {
+		t.Error("expected same lease to be returned for same service")
+	}
+}
+
+// TestManager_Delete_DecrementCounter tests the decrement counter functionality
+func TestManager_Delete_DecrementCounter(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("test-svc", "default", nil)
+
+	// Add twice (simulating adding same service twice)
+	objectName := ServiceNamespacedName(svc)
+
+	ctx1, leaseID1 := getSvcData(svc)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	_ = lease1.Add(objectName)
+
+	ctx2, leaseID2 := getSvcData(svc)
+	lease2 := mgr.Add(ctx2, leaseID2)
+	_ = lease2.Add(objectName)
+
+	// Delete once - should remove the lease
+
+	mgr.Delete(leaseID1, objectName)
+
+	lease := mgr.Get(getSvcID(svc))
+	if lease != nil {
+		t.Error("expected lease to be removed after first delete if same service was processed twice")
+	}
+}
+
+// TestManager_Delete_CancelsContext tests the context cancellation on delete
+func TestManager_Delete_CancelsContext(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("test-svc", "default", nil)
+
+	objectName := ServiceNamespacedName(svc)
+
+	ctx1, leaseID1 := getSvcData(svc)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	_ = lease1.Add(objectName)
+
+	// Verify context is not cancelled
+	select {
+	case <-lease1.Ctx.Done():
+		t.Fatal("expected context to not be cancelled initially")
+	default:
+		// Expected
+	}
+
+	// Delete the lease
+	mgr.Delete(leaseID1, objectName)
+
+	// Verify context is cancelled
+	select {
+	case <-lease1.Ctx.Done():
+		// Expected
+	case <-time.After(100 * time.Millisecond):
+		t.Error("expected context to be cancelled after delete")
+	}
+}
+
+// TestManager_Add_AfterDelete_CreatesNewLease tests adding a service after deleting it
+func TestManager_Add_AfterDelete_CreatesNewLease(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("test-svc", "default", nil)
+
+	objectName := ServiceNamespacedName(svc)
+
+	ctx1, leaseID1 := getSvcData(svc)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	_ = lease1.Add(objectName)
+
+	mgr.Delete(leaseID1, objectName)
+
+	ctx2, leaseID2 := getSvcData(svc)
+	lease2 := mgr.Add(ctx2, leaseID2)
+	isNew := lease2.Add(objectName)
+
+	if !isNew {
+		t.Error("expected isNew to be true after delete and re-add")
+	}
+	if lease1 == lease2 {
+		t.Error("expected new lease to be different from old lease")
+	}
+}
+
+// TestManager_Add_DifferentServices tests adding services with different names
+func TestManager_Add_DifferentServices(t *testing.T) {
+	mgr := NewManager()
+	svc1 := createTestService("svc1", "default", nil)
+	svc2 := createTestService("svc2", "default", nil)
+
+	objectName1 := ServiceNamespacedName(svc1)
+
+	ctx1, leaseID1 := getSvcData(svc1)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	isNew1 := lease1.Add(objectName1)
+
+	objectName2 := ServiceNamespacedName(svc2)
+
+	ctx2, leaseID2 := getSvcData(svc2)
+	lease2 := mgr.Add(ctx2, leaseID2)
+	isNew2 := lease1.Add(objectName2)
+
+	if !isNew1 || !isNew2 {
+		t.Error("expected both adds to return isNew=true")
+	}
+	if lease1 == lease2 {
+		t.Error("expected different leases for different services")
+	}
+}
+
+// TestManager_Add_SameNameDifferentNamespace tests adding services with the same name but different namespaces
+func TestManager_Add_SameNameDifferentNamespace(t *testing.T) {
+	mgr := NewManager()
+	svc1 := createTestService("test-svc", "namespace1", nil)
+	svc2 := createTestService("test-svc", "namespace2", nil)
+
+	objectName1 := ServiceNamespacedName(svc1)
+
+	ctx1, leaseID1 := getSvcData(svc1)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	isNew1 := lease1.Add(objectName1)
+
+	objectName2 := ServiceNamespacedName(svc2)
+
+	ctx2, leaseID2 := getSvcData(svc2)
+	lease2 := mgr.Add(ctx2, leaseID2)
+	isNew2 := lease1.Add(objectName2)
+
+	if !isNew1 || !isNew2 {
+		t.Error("expected both adds to return isNew=true")
+	}
+	if lease1 == lease2 {
+		t.Error("expected different leases for services in different namespaces")
+	}
+}
+
+// TestManager_ConcurrentAccess tests concurrent access to the lease manager
+func TestManager_ConcurrentAccess(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("test-svc", "default", nil)
+
+	var wg sync.WaitGroup
+	const numGoroutines = 100
+
+	objectName1 := ServiceNamespacedName(svc)
+
+	ctx1, leaseID1 := getSvcData(svc)
+	lease1 := mgr.Add(ctx1, leaseID1)
+
+	added := lease1.Add(objectName1)
+	if !added {
+		t.Error("expected lease to be added")
+	}
+
+	// Concurrent adds
+	for range numGoroutines {
+		wg.Go(func() {
+			added := lease1.Add(objectName1)
+			if added {
+				t.Error("expected lease to already exist")
+			}
+		})
+	}
+	wg.Wait()
+
+	mgr.Delete(leaseID1, objectName1)
+
+	// After a one delete, lease should be gone
+	lease := mgr.Get(getSvcID(svc))
+	if lease != nil {
+		t.Error("expected lease to be removed after all concurrent deletes")
+	}
+}
+
+// TestLease_StartedChannel tests the Started channel behavior
+func TestLease_StartedChannel(t *testing.T) {
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	lease := newLease(ctx, cancel)
+
+	// Started channel should be open initially
+	select {
+	case <-lease.Started:
+		t.Fatal("expected Started channel to be open initially")
+	default:
+		// Expected
+	}
+
+	// Close the channel
+	close(lease.Started)
+
+	// Now it should be closed
+	select {
+	case <-lease.Started:
+		// Expected
+	default:
+		t.Error("expected Started channel to be closed after close()")
+	}
+}
+
+// TestGetName_WithoutAnnotation tests with no annotation
+func TestGetName_WithoutAnnotation(t *testing.T) {
+	svc := createTestService("my-service", "my-namespace", nil)
+
+	namespace, name := ServiceName(svc)
+	id := NewID("kubernetes", namespace, name)
+
+	expectedName := "kubevip-my-service"
+	expectedID := "my-namespace/kubevip-my-service"
+
+	if id.Name() != expectedName {
+		t.Errorf("expected name %q, got %q", expectedName, id.Name())
+	}
+	if id.NamespacedName() != expectedID {
+		t.Errorf("expected id %q, got %q", expectedID, id.NamespacedName())
+	}
+}
+
+// TestGetName_WithAnnotation tests with a shared lease annotation
+func TestGetName_WithAnnotation(t *testing.T) {
+	svc := createTestService("my-service", "my-namespace", map[string]string{
+		serviceLeaseAnnotation: "shared-lease",
+	})
+
+	namespace, name := ServiceName(svc)
+	id := NewID("kubernetes", namespace, name)
+
+	expectedName := "shared-lease"
+	expectedID := "my-namespace/shared-lease"
+
+	if id.Name() != expectedName {
+		t.Errorf("expected name %q, got %q", expectedName, id.Name())
+	}
+	if id.NamespacedName() != expectedID {
+		t.Errorf("expected id %q, got %q", expectedID, id.NamespacedName())
+	}
+}
+
+// TestGetName_WithAnnotation tests with a shared lease annotation
+func TestGetName_WithAnnotationAndOverriddenNamespace(t *testing.T) {
+	svc := createTestService("my-service", "my-namespace", map[string]string{
+		serviceLeaseAnnotation: "other-namespace/shared-lease",
+	})
+
+	namespace, name := ServiceName(svc)
+	id := NewID("kubernetes", namespace, name)
+
+	expectedName := "shared-lease"
+	expectedID := "other-namespace/shared-lease"
+	expectedNamespace := "other-namespace"
+
+	if id.Name() != expectedName {
+		t.Errorf("expected name %q, got %q", expectedName, id.Name())
+	}
+	if id.NamespacedName() != expectedID {
+		t.Errorf("expected id %q, got %q", expectedID, id.NamespacedName())
+	}
+	if id.Namespace() != expectedNamespace {
+		t.Errorf("expected namespace %q, got %q", expectedNamespace, id.Namespace())
+	}
+}
+
+// TestGetName_WithoutAnnotation_Etcd tests with no annotation
+func TestGetName_WithoutAnnotation_Etcd(t *testing.T) {
+	svc := createTestService("my-service", "my-namespace", nil)
+
+	namespace, name := ServiceName(svc)
+	id := NewID("etcd", namespace, name)
+
+	expectedName := "kubevip-my-service"
+	expectedID := "my-namespace-kubevip-my-service"
+
+	if id.Name() != expectedName {
+		t.Errorf("expected name %q, got %q", expectedName, id.Name())
+	}
+	if id.NamespacedName() != expectedID {
+		t.Errorf("expected id %q, got %q", expectedID, id.NamespacedName())
+	}
+}
+
+// TestGetName_WithAnnotation_Etcd tests with a shared lease annotation
+func TestGetName_WithAnnotation_Etcd(t *testing.T) {
+	svc := createTestService("my-service", "my-namespace", map[string]string{
+		serviceLeaseAnnotation: "shared-lease",
+	})
+
+	namespace, name := ServiceName(svc)
+	id := NewID("etcd", namespace, name)
+
+	expectedName := "shared-lease"
+	expectedID := "my-namespace-shared-lease"
+
+	if id.Name() != expectedName {
+		t.Errorf("expected name %q, got %q", expectedName, id.Name())
+	}
+	if id.NamespacedName() != expectedID {
+		t.Errorf("expected id %q, got %q", expectedID, id.NamespacedName())
+	}
+}
+
+// TestGetName_WithAnnotation_Etcd tests with a shared lease annotation
+func TestGetName_WithAnnotationAndOverriddenNamespace_Etcd(t *testing.T) {
+	svc := createTestService("my-service", "my-namespace", map[string]string{
+		serviceLeaseAnnotation: "other-namespace/shared-lease",
+	})
+
+	namespace, name := ServiceName(svc)
+	id := NewID("etcd", namespace, name)
+
+	expectedName := "shared-lease"
+	expectedID := "other-namespace-shared-lease"
+	expectedNamespace := "other-namespace"
+
+	if id.Name() != expectedName {
+		t.Errorf("expected name %q, got %q", expectedName, id.Name())
+	}
+	if id.NamespacedName() != expectedID {
+		t.Errorf("expected id %q, got %q", expectedID, id.NamespacedName())
+	}
+	if id.Namespace() != expectedNamespace {
+		t.Errorf("expected namespace %q, got %q", expectedNamespace, id.Namespace())
+	}
+}
+
+// TestManager_LeaderElectionRestartScenario simulates the bug scenario where
+// leadership is lost and the restartable service watcher tries to restart
+// the leader election. This test verifies that after deleting the lease,
+// a new lease can be created.
+func TestManager_LeaderElectionRestartScenario_etcd(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("traefik", "traefik", nil)
+
+	objectName1 := ServiceNamespacedName(svc)
+
+	ctx1, leaseID1 := getSvcData(svc)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	isNew1 := lease1.Add(objectName1)
+
+	if !isNew1 {
+		t.Fatal("expected first add to return isNew=true")
+	}
+
+	// Simulate leadership acquired - close Started channel
+	close(lease1.Started)
+
+	// Simulate leadership lost - the leader election function should delete the lease
+	// This is the fix: delete the lease when RunOrDie returns
+	mgr.Delete(leaseID1, objectName1)
+
+	// Verify lease is removed
+	if mgr.Get(getSvcID(svc)) != nil {
+		t.Error("expected lease to be removed after delete")
+	}
+
+	// Simulate restartable service watcher calling StartServicesLeaderElection again
+	ctx2, leaseID2 := getSvcData(svc)
+	lease2 := mgr.Add(ctx2, leaseID2)
+	isNew2 := lease1.Add(objectName1)
+	if !isNew2 {
+		t.Fatal("expected second add after delete to return isNew=true")
+	}
+
+	// Verify we got a new lease with a fresh Started channel
+	if lease1 == lease2 {
+		t.Error("expected new lease to be different from old lease")
+	}
+
+	// Verify the new Started channel is not closed
+	select {
+	case <-lease2.Started:
+		t.Error("expected new lease's Started channel to be open")
+	default:
+		// Expected
+	}
+}
+
+// TestManager_CommonLeaseScenario tests the common lease feature where
+// multiple services share the same lease.
+func TestManager_CommonLeaseScenario(t *testing.T) {
+	mgr := NewManager()
+
+	// Two services sharing the same lease via annotation
+	sharedLeaseAnnotations := map[string]string{
+		serviceLeaseAnnotation: "shared-lease",
+	}
+	svc1 := createTestService("svc1", "default", sharedLeaseAnnotations)
+	svc2 := createTestService("svc2", "default", sharedLeaseAnnotations)
+
+	// First service gets a new lease
+	objectName1 := ServiceNamespacedName(svc1)
+
+	ctx1, leaseID1 := getSvcData(svc1)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	isNew1 := lease1.Add(objectName1)
+	if !isNew1 {
+		t.Error("expected first add to return isNew=true")
+	}
+
+	// Simulate first service starting leadership
+	close(lease1.Started)
+
+	objectName2 := ServiceNamespacedName(svc2)
+
+	ctx2, leaseID2 := getSvcData(svc2)
+	lease2 := mgr.Add(ctx2, leaseID2)
+	isNew2 := lease2.Add(objectName2)
+
+	// Second service should get the same lease
+	if !isNew2 {
+		t.Error("expected second add with same lease name to return isNew=true")
+	}
+	if lease1 != lease2 {
+		t.Error("expected same lease for services with same lease annotation")
+	}
+
+	// Delete first service - lease should still exist
+	mgr.Delete(leaseID1, objectName1)
+	if mgr.Get(getSvcID(svc1)) == nil {
+		t.Error("expected lease to still exist after first delete")
+	}
+
+	// Delete second service - lease should be removed
+	mgr.Delete(leaseID2, objectName2)
+	if mgr.Get(getSvcID(svc2)) != nil {
+		t.Error("expected lease to be removed after all services deleted")
+	}
+}
+
+// TestManager_RaceCondition_LeaseExistsBeforeDelete tests the scenario where
+// a second goroutine calls Add before the first goroutine's defer deletes the lease.
+// This simulates the race condition that could cause the gaps in the logs where there is no leader.
+func TestManager_RaceCondition_LeaseExistsBeforeDelete(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("traefik", "traefik", nil)
+
+	// Simulate first leader election start
+	objectName1 := ServiceNamespacedName(svc)
+
+	ctx1, leaseID1 := getSvcData(svc)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	isNew1 := lease1.Add(objectName1)
+	if !isNew1 {
+		t.Fatal("expected first add to return isNew=true")
+	}
+
+	// Simulate leadership acquired - close Started channel
+	close(lease1.Started)
+
+	// Simulate a second goroutine calling Add BEFORE the first goroutine's defer deletes the lease
+	// This is the race condition scenario
+	ctx2, leaseID2 := getSvcData(svc)
+	lease2 := mgr.Add(ctx2, leaseID2)
+	isNew2 := lease1.Add(objectName1)
+
+	if isNew2 {
+		t.Error("expected second add before delete to return isNew=false")
+	}
+	if lease1 != lease2 {
+		t.Error("expected same lease to be returned")
+	}
+
+	// The Started channel should be closed (from the first run)
+	select {
+	case <-lease2.Started:
+		// Expected - channel is closed
+	default:
+		t.Error("expected Started channel to be closed")
+	}
+
+	// Now the first goroutine's defer deletes the lease
+	mgr.Delete(leaseID1, objectName1)
+
+	// The lease should not still exist because same service was processed twice, so we do not increment the counter
+	if mgr.Get(getSvcID(svc)) != nil {
+		t.Error("expected lease tonot exist")
+	}
+
+	// Second delete does nothing
+	mgr.Delete(leaseID2, objectName1)
+	if mgr.Get(getSvcID(svc)) != nil {
+		t.Error("expected lease to not exist")
+	}
+}
+
+// TestManager_NonCommonLease_MultipleAdds tests that multiple Adds for a non-common
+// lease service increment the counter correctly.
+func TestManager_NonCommonLease_MultipleAdds(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("traefik", "traefik", nil) // No common lease annotation
+
+	// First Add
+	objectName1 := ServiceNamespacedName(svc)
+
+	ctx1, leaseID1 := getSvcData(svc)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	isNew1 := lease1.Add(objectName1)
+	if !isNew1 {
+		t.Error("expected first add to return isNew=true")
+	}
+
+	// Close Started to simulate leadership acquired
+	close(lease1.Started)
+
+	// Second Add (simulating another goroutine or restart attempt)
+	ctx2, leaseID2 := getSvcData(svc)
+	lease2 := mgr.Add(ctx2, leaseID2)
+	isNew2 := lease1.Add(objectName1)
+	if isNew2 {
+		t.Error("expected second add to return isNew=false")
+	}
+	if lease1 != lease2 {
+		t.Error("expected same lease")
+	}
+
+	// Third Add
+	ctx3, leaseID3 := getSvcData(svc)
+	lease3 := mgr.Add(ctx3, leaseID3)
+	isNew3 := lease1.Add(objectName1)
+	if isNew3 {
+		t.Error("expected third add to return isNew=false")
+	}
+	if lease1 != lease3 {
+		t.Error("expected same lease")
+	}
+
+	// Need one delete to remove the lease, another delete runs do nothing
+	mgr.Delete(leaseID1, objectName1)
+	if mgr.Get(getSvcID(svc)) != nil {
+		t.Error("expected lease to be deleted")
+	}
+
+	mgr.Delete(leaseID2, objectName1)
+	if mgr.Get(getSvcID(svc)) != nil {
+		t.Error("expected lease to be deleted")
+	}
+
+	mgr.Delete(leaseID3, objectName1)
+	if mgr.Get(getSvcID(svc)) != nil {
+		t.Error("expected lease to be deleted")
+	}
+}
+
+// TestManager_LeaseContextCancelledBeforeStarted tests the scenario where
+// the lease context is cancelled before the Started channel is closed.
+// This can happen if leadership is never acquired and the context times out.
+func TestManager_LeaseContextCancelledBeforeStarted(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("traefik", "traefik", nil)
+
+	// First Add
+	objectName1 := ServiceNamespacedName(svc)
+
+	ctx1, leaseID1 := getSvcData(svc)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	isNew1 := lease1.Add(objectName1)
+
+	if !isNew1 {
+		t.Fatal("expected first add to return isNew=true")
+	}
+
+	ctx2, leaseID2 := getSvcData(svc)
+	lease2 := mgr.Add(ctx2, leaseID2)
+	isNew2 := lease1.Add(objectName1)
+
+	if isNew2 {
+		t.Error("expected second add to return isNew=false")
+	}
+
+	// Verify Started is not closed yet
+	select {
+	case <-lease2.Started:
+		t.Error("expected Started channel to be open")
+	default:
+		// Expected
+	}
+
+	// Cancel the lease context (simulating timeout or leadership loss before acquiring)
+	lease1.Cancel()
+
+	// Verify context is cancelled
+	select {
+	case <-lease2.Ctx.Done():
+		// Expected
+	case <-time.After(100 * time.Millisecond):
+		t.Error("expected context to be cancelled")
+	}
+
+	// Delete should still work
+	mgr.Delete(leaseID1, objectName1)
+	mgr.Delete(leaseID2, objectName1)
+
+	if mgr.Get(getSvcID(svc)) != nil {
+		t.Error("expected lease to be removed")
+	}
+}
+
+// TestManager_RestartAfterLeaseContextCancelled tests that after the lease
+// context is cancelled and the lease is deleted, a new lease can be created.
+func TestManager_RestartAfterLeaseContextCancelled(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("traefik", "traefik", nil)
+
+	// First Add
+	objectName1 := ServiceNamespacedName(svc)
+
+	ctx1, leaseID1 := getSvcData(svc)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	_ = lease1.Add(objectName1)
+
+	// Cancel context before Started is closed
+	lease1.Cancel()
+
+	// Delete the lease
+	mgr.Delete(leaseID1, objectName1)
+
+	// Verify lease is gone
+	if mgr.Get(getSvcID(svc)) != nil {
+		t.Error("expected lease to be removed after delete")
+	}
+
+	// Add again - should create new lease
+	ctx2, leaseID2 := getSvcData(svc)
+	lease2 := mgr.Add(ctx2, leaseID2)
+	isNew2 := lease1.Add(objectName1)
+
+	if !isNew2 {
+		t.Error("expected new lease after delete")
+	}
+
+	// Verify new lease has fresh context and Started channel
+	select {
+	case <-lease2.Ctx.Done():
+		t.Error("expected new lease context to be active")
+	default:
+		// Expected
+	}
+
+	select {
+	case <-lease2.Started:
+		t.Error("expected new lease Started channel to be open")
+	default:
+		// Expected
+	}
+}
+
+// TestManager_NonCommonLease_WaitForLeaseContextDone tests the scenario where
+// a non-common lease service calls Add while another leader election is running.
+// The caller should wait for the lease context to be done before returning.
+// This test verifies the fix for the tight spin loop issue.
+func TestManager_NonCommonLease_WaitForLeaseContextDone(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("egress-service", "default", nil) // Non-common lease
+
+	// First Add - simulates the first leader election starting
+	objectName1 := ServiceNamespacedName(svc)
+
+	ctx1, leaseID1 := getSvcData(svc)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	isNew1 := lease1.Add(objectName1)
+
+	if !isNew1 {
+		t.Fatal("expected first add to return isNew=true")
+	}
+
+	// Simulate leadership acquired
+	close(lease1.Started)
+
+	// Second Add - simulates another goroutine trying to start leader election
+	// This should return isNew=false
+	ctx2, leaseID2 := getSvcData(svc)
+	lease2 := mgr.Add(ctx2, leaseID2)
+	isNew2 := lease1.Add(objectName1)
+
+	if isNew2 {
+		t.Error("expected second add to return isNew=false")
+	}
+
+	if lease1 != lease2 {
+		t.Error("expected same lease to be returned")
+	}
+
+	// Verify Started channel is closed (leadership was acquired by first)
+	select {
+	case <-lease2.Started:
+		// Expected - channel is closed
+	default:
+		t.Error("expected Started channel to be closed")
+	}
+
+	// In the actual code (leader.go), when isNew=false for non-common lease,
+	// the code waits on either svcCtx.Ctx.Done() or svcLease.Ctx.Done()
+	// Here we verify that the lease context gets cancelled when we delete the lease
+
+	// Start a goroutine that waits for the lease context to be done
+	// This simulates what the leader.go code does
+	waitDone := make(chan struct{})
+	go func() {
+		select {
+		case <-lease2.Ctx.Done():
+			close(waitDone)
+		case <-time.After(1 * time.Second):
+			// Timeout - test will fail
+		}
+	}()
+
+	// Verify the goroutine is still waiting (lease context not yet cancelled)
+	select {
+	case <-waitDone:
+		t.Fatal("goroutine should still be waiting")
+	case <-time.After(50 * time.Millisecond):
+		// Expected - still waiting
+	}
+
+	// Now simulate the first leader election ending (defer deletes the lease)
+	mgr.Delete(leaseID1, objectName1)
+
+	// The lease context should now be cancelled (because counter went to 0)
+	// But we added twice, so we need to delete twice
+	mgr.Delete(leaseID2, objectName1)
+
+	// Now the goroutine should have completed
+	select {
+	case <-waitDone:
+		// Expected - lease context was cancelled
+	case <-time.After(200 * time.Millisecond):
+		t.Error("expected goroutine to complete after lease context cancelled")
+	}
+
+	// Verify lease is removed
+	if mgr.Get(getSvcID(svc)) != nil {
+		t.Error("expected lease to be removed")
+	}
+}
+
+// TestManager_NonCommonLease_SpinLoopPrevention tests that the fix prevents
+// a tight spin loop when a non-common lease service repeatedly calls Add
+// while leader election is running. The key behavior is that when isNew=false,
+// the lease context should be used to block until the leader election ends.
+func TestManager_NonCommonLease_SpinLoopPrevention(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("egress-service", "default", nil) // Non-common lease
+
+	// First Add - leader election starts
+	objectName1 := ServiceNamespacedName(svc)
+
+	ctx1, leaseID1 := getSvcData(svc)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	isNew1 := lease1.Add(objectName1)
+
+	if !isNew1 {
+		t.Fatal("expected first add to return isNew=true")
+	}
+
+	close(lease1.Started)
+
+	// Track how many times Add is called in a tight loop
+	// In the buggy code, this would spin forever
+	// In the fixed code, Add returns isNew=false and the caller blocks on lease.Ctx.Done()
+	addCount := 0
+	done := make(chan struct{})
+
+	go func() {
+		for i := 0; i < 100; i++ {
+			objectName1 := ServiceNamespacedName(svc)
+
+			ctxTmp, leaseIDTmp := getSvcData(svc)
+			leaseTmp := mgr.Add(ctxTmp, leaseIDTmp)
+			isNewTmp := leaseTmp.Add(objectName1)
+			addCount++
+			if isNewTmp {
+				// This shouldn't happen while the first lease exists
+				t.Error("unexpected isNew=true")
+				break
+			}
+			// In the fixed code, we would block here on lease.Ctx.Done()
+			// For this test, we just verify that isNew=false is returned
+			// and the same lease is returned each time
+			if leaseTmp != lease1 {
+				t.Error("expected same lease")
+				break
+			}
+		}
+		close(done)
+	}()
+
+	// Wait for the loop to complete
+	select {
+	case <-done:
+		// Expected
+	case <-time.After(1 * time.Second):
+		t.Fatal("loop timed out")
+	}
+
+	// All 100 adds should have completed (returning isNew=false)
+	if addCount != 100 {
+		t.Errorf("expected 100 adds, got %d", addCount)
+	}
+
+	mgr.Delete(leaseID1, objectName1)
+	if mgr.Get(getSvcID(svc)) != nil {
+		t.Error("expected lease to be removed after first delete")
+	}
+}
+
+// TestManager_NonCommonLease_ServiceContextCancellation tests that when
+// a service is deleted (svcCtx.Ctx cancelled), the waiting goroutine
+// should also unblock. This is the other exit path from the wait.
+func TestManager_NonCommonLease_ServiceContextCancellation(t *testing.T) {
+	mgr := NewManager()
+	svc := createTestService("egress-service", "default", nil)
+
+	// First Add - leader election starts
+	objectName1 := ServiceNamespacedName(svc)
+
+	ctx1, leaseID1 := getSvcData(svc)
+	lease1 := mgr.Add(ctx1, leaseID1)
+	_ = lease1.Add(objectName1)
+	close(lease1.Started)
+
+	// Second Add - returns isNew=false
+	ctx2, leaseID2 := getSvcData(svc)
+	lease2 := mgr.Add(ctx2, leaseID2)
+	isNew2 := lease2.Add(objectName1)
+	if isNew2 {
+		t.Error("expected isNew=false")
+	}
+
+	// Create a simulated service context
+	svcCtx, svcCancel := context.WithCancel(context.Background())
+
+	// Start a goroutine that waits on either svcCtx or lease context
+	// This simulates the behavior in leader.go
+	waitDone := make(chan string)
+	go func() {
+		select {
+		case <-svcCtx.Done():
+			waitDone <- "svcCtx"
+		case <-lease2.Ctx.Done():
+			waitDone <- "leaseCtx"
+		case <-time.After(1 * time.Second):
+			waitDone <- "timeout"
+		}
+	}()
+
+	// Cancel the service context (simulates service deletion)
+	svcCancel()
+
+	// The goroutine should unblock via svcCtx.Done()
+	select {
+	case result := <-waitDone:
+		if result != "svcCtx" {
+			t.Errorf("expected to unblock via svcCtx, got %s", result)
+		}
+	case <-time.After(200 * time.Millisecond):
+		t.Error("goroutine should have unblocked")
+	}
+}

pkg/loadbalancer/ipvs.go

@@ -17,6 +17,7 @@ import (
 	"github.com/kube-vip/kube-vip/pkg/backend"
 	"github.com/kube-vip/kube-vip/pkg/sysctl"
 	"github.com/kube-vip/kube-vip/pkg/utils"
+	"github.com/kube-vip/kube-vip/pkg/vip"
 	"github.com/vishvananda/netlink"
 )
 
@@ -50,49 +51,53 @@ type IPVSLoadBalancer struct {
 	backendMap          backend.Map
 	interval            int
 	lock                sync.Mutex
-	stop                chan struct{}
-	networkInterface    string
-	leaderCancel        context.CancelFunc
-	signal              chan os.Signal
+	killFunc            func()
 	address             string
 	family              ipvs.AddressFamily
 }
 
-func NewIPVSLB(address string, port uint16, forwardingMethod string, backendHealthCheckInterval int, networkInterface string, leaderCancel context.CancelFunc, signal chan os.Signal) (*IPVSLoadBalancer, error) {
+func NewIPVSLB(ctx context.Context, address string, port uint16, forwardingMethod string, backendHealthCheckInterval int,
+	killFunc func(), wg *sync.WaitGroup) (*IPVSLoadBalancer, error) {
 	log.Info("Starting IPVS LoadBalancer", "address", address)
 
 	// Create IPVS client
 	c, err := ipvs.New()
 	if err != nil {
 		log.Error("ensure IPVS kernel modules are loaded")
-		log.Error("Error starting IPVS", "err", err)
-		panic("")
+		log.Error("error starting IPVS", "err", err)
+		return nil, fmt.Errorf("starting IPVS: %w", err)
 	}
 	i, err := c.Info()
 	if err != nil {
 		log.Error("ensure IPVS kernel modules are loaded")
-		log.Error("Error retrieving IPVS info", "err", err)
+		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")
 		}
-		panic("")
+		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)
 
 	if strings.ToLower(forwardingMethod) == "masquerade" {
-		enableProcSys("/proc/sys/net/ipv4/vs/conntrack", "net.ipv4.vs.conntrack")
+		if err := enableProcSys("/proc/sys/net/ipv4/vs/conntrack", "net.ipv4.vs.conntrack"); err != nil {
+			return nil, err
+		}
 		if family == ipvs.INET6 {
-			enableProcSys("/proc/sys/net/ipv6/conf/all/forwarding", "net.ipv6.conf.all.forwarding")
+			if err := enableProcSys("/proc/sys/net/ipv6/conf/all/forwarding", "net.ipv6.conf.all.forwarding"); err != nil {
+				return nil, err
+			}
 		} else {
-			enableProcSys("/proc/sys/net/ipv4/ip_forward", "net.ipv4.ip_forward")
+			if err := enableProcSys("/proc/sys/net/ipv4/ip_forward", "net.ipv4.ip_forward"); err != nil {
+				return nil, err
+			}
 		}
 	}
 
-	netMask := netmask.MaskFrom(31, 32) // For ipv4
+	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
 	}
 
 	// Generate out API Server LoadBalancer instance
@@ -133,34 +138,32 @@ func NewIPVSLB(address string, port uint16, forwardingMethod string, backendHeal
 		forwardingMethod:    m,
 		interval:            backendHealthCheckInterval,
 		backendMap:          make(backend.Map),
-		stop:                make(chan struct{}),
-		networkInterface:    networkInterface,
-		leaderCancel:        leaderCancel,
-		signal:              signal,
+		killFunc:            killFunc,
 		address:             address,
 		family:              family,
 	}
 
-	go lb.healthCheck()
+	wg.Go(func() {
+		lb.healthCheck(ctx)
+	})
 
 	// Return our created load-balancer
 	return lb, nil
 }
 
-func enableProcSys(path, name string) {
+func enableProcSys(path, name string) error {
 	isSet, err := sysctl.EnableProcSys(path)
 	if err != nil {
-		log.Error(fmt.Sprintf("ensuring %s enabled", name), "err", err)
-		panic("")
+		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 {
 	log.Info("Stopping IPVS LoadBalancer", "address", lb.address)
-	close(lb.stop)
 	err := lb.client.RemoveService(lb.loadBalancerService)
 	if err != nil {
 		return fmt.Errorf("error removing existing IPVS service: %v", err)
@@ -183,6 +186,7 @@ func (lb *IPVSLoadBalancer) AddBackend(address string, port uint16) error {
 		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}
@@ -227,7 +231,7 @@ func (lb *IPVSLoadBalancer) addBackend(address string, port uint16) error {
 			// Fatal error at this point as IPVS is probably not working
 			log.Error("Unable to create an IPVS service, ensure IPVS kernel modules are loaded")
 			log.Error("IPVS service", "err", err)
-			panic("")
+			return utils.NewPanicError(fmt.Sprintf("unable to create an IPVS service - %s", err))
 
 		}
 		log.Info("load-Balancer services created", "address", lb.addrString(), "port", lb.Port)
@@ -312,8 +316,8 @@ 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 {
@@ -338,43 +342,50 @@ func (lb *IPVSLoadBalancer) healthCheck() {
 					lb.backendMap[backend] = newStatus
 				}
 				if lb.forwardingMethod == ipvs.Local && !lb.localBackendExists() {
-					if lb.signal != nil {
-						close(lb.signal)
-					}
-
-					if lb.leaderCancel != nil {
-						lb.leaderCancel()
+					if lb.killFunc != nil {
+						log.Error("no local backends available, restarting kube-vip")
+						lb.killFunc()
 					}
 				}
 			}
 		}
-	}, lb.interval, lb.stop)
+	})
 }
 
 func (lb *IPVSLoadBalancer) isLocal(address string) (bool, error) {
-	link, err := netlink.LinkByName(lb.networkInterface)
-	if err != nil {
-		return false, fmt.Errorf("getting link '%s': %w", lb.networkInterface, err)
-	}
-
-	family := netlink.FAMILY_V6
-	if utils.IsIPv4(address) {
-		family = netlink.FAMILY_V4
-	}
-
 	target := net.ParseIP(address)
 	if target == nil {
-		return false, fmt.Errorf("address '%s' is not a valid IP address", address)
+		return false, fmt.Errorf("unable to parse IP address %s", address)
 	}
 
-	addrs, err := netlink.AddrList(link, family)
+	links, err := netlink.LinkList()
 	if err != nil {
-		return false, fmt.Errorf("listing addresses for link '%s': %w", lb.networkInterface, err)
+		return false, fmt.Errorf("listing links: %w", err)
 	}
 
-	for _, addr := range addrs {
-		if addr.IP.Equal(target) {
-			return true, nil
+	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
+			}
 		}
 	}
 

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/manager.go

@@ -2,6 +2,7 @@ package manager
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"net/http"
 	"os"
@@ -9,6 +10,7 @@ import (
 	"path/filepath"
 	"strings"
 	"sync"
+	"sync/atomic"
 	"syscall"
 	"time"
 
@@ -16,20 +18,26 @@ import (
 
 	"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/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"
 	"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
@@ -46,8 +54,7 @@ type Manager struct {
 	// 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
 
@@ -55,10 +62,6 @@ type Manager struct {
 	// 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
 
@@ -72,10 +75,22 @@ type Manager struct {
 	// 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`,
@@ -90,6 +105,7 @@ func New(configMap string, config *kubevip.Config) (*Manager, error) {
 		}
 		config.NodeName = hostname
 	}
+	config.NodeName = normalizeNodeName(config.NodeName)
 	log.Info("using node name", "name", config.NodeName)
 
 	adminConfigPath := "/etc/kubernetes/admin.conf"
@@ -187,8 +203,8 @@ func New(configMap string, config *kubevip.Config) (*Manager, error) {
 	// Add Notification for SIGTERM (sent from Kubernetes)
 	signal.Notify(signalChan, syscall.SIGTERM)
 
-	// All watchers and other goroutines should have an additional goroutine that blocks on this, to shut things down
-	shutdownChan := make(chan struct{})
+	// Add Notification for SIGUSR1 (for configuration dump)
+	signal.Notify(signalChan, syscall.SIGUSR1)
 
 	intfMgr := networkinterface.NewManager()
 	arpMgr := arp.NewManager(config)
@@ -198,14 +214,36 @@ func New(configMap string, config *kubevip.Config) (*Manager, error) {
 	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)
 		if err != nil {
 			return nil, fmt.Errorf("creating BGP server: %w", err)
 		}
 	}
 
-	svcProcessor := services.NewServicesProcessor(config, bgpServer, clientset, rwClientSet, shutdownChan, intfMgr, arpMgr, nodeLabelManager)
+	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,
@@ -218,39 +256,22 @@ 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,
-		shutdownChan:     shutdownChan,
 		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)
-
-	// Add Notification for SIGTERM (sent from Kubernetes)
-	signal.Notify(sm.signalChan, syscall.SIGTERM)
-
-	// Add Notification for SIGUSR1 (for configuration dump)
-	signal.Notify(sm.signalChan, syscall.SIGUSR1)
-
-	// All watchers and other goroutines should have an additional goroutine that blocks on this, to shut things down
-	sm.shutdownChan = make(chan struct{})
+func (sm *Manager) Start(ctx context.Context) error {
+	wg := sync.WaitGroup{}
+	defer wg.Wait()
 
 	// HealthCheck
 	if sm.config.HealthCheckPort != 0 {
@@ -260,7 +281,7 @@ func (sm *Manager) Start() error {
 		http.HandleFunc("/healthz", func(w http.ResponseWriter, _ *http.Request) {
 			fmt.Fprintf(w, "OK")
 		})
-		go func() {
+		wg.Go(func() {
 			server := &http.Server{
 				Addr:              fmt.Sprintf(":%d", sm.config.HealthCheckPort),
 				ReadHeaderTimeout: 3 * time.Second,
@@ -269,7 +290,7 @@ func (sm *Manager) Start() error {
 			if err != nil {
 				log.Error("healthcheck", "unable to start", err)
 			}
-		}()
+		})
 	}
 
 	// on exit, clean up the node labels
@@ -279,22 +300,9 @@ func (sm *Manager) Start() error {
 		}
 	}()
 
-	// 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
-		}
-
-		log.Info("Starting Kube-vip Manager with the BGP engine")
-		return sm.startBGP()
-	}
-
 	if sm.config.EnableARP || sm.config.EnableWireguard {
 		if sm.config.EnableUPNP {
-			clients := upnp.GetConnectionClients(context.TODO())
+			clients := upnp.GetConnectionClients(ctx)
 			if len(clients) == 0 {
 				log.Error("Error Enabling UPNP. No Clients found")
 				// Set the struct to false so nothing should use it in future
@@ -309,49 +317,130 @@ func (sm *Manager) Start() error {
 				}
 			}
 			// TODO: It would be nice to run the UPNP refresh only on the leader.
-			go sm.svcProcessor.RefreshUPNPForwards()
+			wg.Go(func() {
+				sm.svcProcessor.RefreshUPNPForwards(ctx)
+			})
 		}
 	}
 
-	// If ARP is enabled then we start a LeaderElection that will use ARP to advertise VIPs
-	if sm.config.EnableARP {
-		log.Info("Starting Kube-vip Manager with the ARP engine")
-		return sm.startARP(sm.config.NodeName)
-	}
-
-	if sm.config.EnableWireguard {
-		log.Info("Starting Kube-vip Manager with the Wireguard engine")
-		return sm.startWireguard(sm.config.NodeName)
-	}
-
-	if sm.config.EnableRoutingTable {
-		log.Info("Starting Kube-vip Manager with the Routing Table engine")
-		return sm.startTableMode(sm.config.NodeName)
-	}
-
-	log.Error("prematurely exiting Load-balancer as no modes [ARP/BGP/Wireguard] are enabled")
-	return nil
+	return sm.startMode(ctx)
 }
 
-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
+// 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
 		}
-		return "", err
-	}
-	return "", fmt.Errorf("unable to find Namespace")
-}
-
-func (sm *Manager) parseAnnotations() error {
-	if sm.config.Annotations == "" {
-		log.Debug("No Node annotations to parse")
-		return nil
 	}
 
-	err := sm.annotationsWatcher()
-	if err != nil {
-		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 {
+			err := nftables.ClearTables()
+			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 errors.Is(err, &utils.PanicError{}) {
+						sm.Kill()
+						return fmt.Errorf("failed to reconcile services, non-recoverable error: %w", err)
+					} else {
+						log.Error("failed to reconcile services, restarting", "error", err)
+					}
+				}
+			}
+		}
+	}
+
 	return nil
 }
+
+func (sm *Manager) waitForShutdown(ctx context.Context, cancel context.CancelFunc, cpCluster *cluster.Cluster) {
+	for {
+		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

@@ -1,168 +0,0 @@
-package manager
-
-import (
-	"context"
-	"syscall"
-	"time"
-
-	log "log/slog"
-
-	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()
-
-	log.Info("Start ARP/NDP advertisement")
-	go sm.arpMgr.StartAdvertisement(ctx)
-
-	// Shutdown function that will wait on this signal, unless we call it ourselves
-	go func() {
-		for {
-			sig := <-sm.signalChan
-			switch sig {
-			case syscall.SIGUSR1:
-				log.Info("Received SIGUSR1, dumping configuration")
-				sm.dumpConfiguration()
-			case syscall.SIGINT, syscall.SIGTERM:
-				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()
-				return
-			}
-		}
-	}()
-
-	if sm.config.EnableControlPlane {
-		cpCluster, err = cluster.InitCluster(sm.config, false, sm.intfMgr, sm.arpMgr)
-		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.Error("starting control plane", "err", 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.shutdownChan
-			log.Info("Shutting down Kube-Vip")
-
-			return nil
-		}
-
-		ns = sm.config.Namespace
-	} else {
-
-		ns, err = returnNameSpace()
-		if err != nil {
-			log.Warn("unable to auto-detect namespace, dropping to config", "namespace", sm.config.Namespace)
-			ns = sm.config.Namespace
-		}
-	}
-
-	// This will tidy any dangling kube-vip iptables rules
-	if sm.config.EgressClean {
-		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.Info("beginning watching services, leaderelection will happen for every service")
-		err = sm.svcProcessor.StartServicesWatchForLeaderElection(ctx)
-		if err != nil {
-			return err
-		}
-	} else {
-
-		log.Info("beginning services leadership", "namespace", ns, "lock name", sm.config.ServicesLeaseName, "id", 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.svcProcessor.ServicesWatcher(ctx, sm.svcProcessor.SyncServices)
-					if err != nil {
-						log.Error("service watcher", "err", err)
-						panic("") // TODO: - emulating log.fatal here
-					}
-				},
-				OnStoppedLeading: func() {
-					// we can do cleanup here
-					sm.mutex.Lock()
-					defer sm.mutex.Unlock()
-					log.Info("leader lost", "new leader", id)
-					sm.svcProcessor.Stop()
-
-					log.Error("lost leadership, restarting kube-vip")
-					panic("") // TODO: - emulating log.fatal here
-				},
-				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.Info("new leader elected", "new leader", identity)
-				},
-			},
-		})
-	}
-	return nil
-}

pkg/manager/manager_bgp.go

@@ -1,134 +0,0 @@
-package manager
-
-import (
-	"context"
-	"fmt"
-	"syscall"
-
-	log "log/slog"
-
-	"github.com/kube-vip/kube-vip/pkg/bgp"
-	"github.com/kube-vip/kube-vip/pkg/cluster"
-	api "github.com/osrg/gobgp/v3/api"
-	"github.com/prometheus/client_golang/prometheus"
-)
-
-// 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 sm.bgpServer == nil {
-		sm.bgpServer, err = bgp.NewBGPServer(sm.config.BGPConfig)
-		if err != nil {
-			return fmt.Errorf("creating BGP server: %w", err)
-		}
-	}
-
-	log.Info("Starting the BGP server to advertise VIP routes to BGP peers")
-	if err := sm.bgpServer.Start(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)
-		}
-	}); err != nil {
-		return fmt.Errorf("starting BGP server: %w", 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() {
-		for {
-			sig := <-sm.signalChan
-			switch sig {
-			case syscall.SIGUSR1:
-				log.Info("Received SIGUSR1, dumping configuration")
-				sm.dumpConfiguration()
-			case syscall.SIGINT, syscall.SIGTERM:
-				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()
-				return
-			}
-		}
-	}()
-
-	if sm.config.EnableControlPlane {
-		cpCluster, err = cluster.InitCluster(sm.config, false, sm.intfMgr, sm.arpMgr)
-		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)
-			}
-			if err != nil {
-				log.Error("Control Plane", "err", 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.Info("Shutting down Kube-Vip")
-
-			return nil
-		}
-	}
-
-	if sm.config.EnableServicesElection {
-		log.Info("beginning watching services, leaderelection will happen for every service")
-		err = sm.svcProcessor.StartServicesWatchForLeaderElection(ctx)
-		if err != nil {
-			return err
-		}
-	} else {
-		log.Info("beginning watching services without leader election")
-		err = sm.svcProcessor.ServicesWatcher(ctx, sm.svcProcessor.SyncServices)
-		if err != nil {
-			return err
-		}
-	}
-
-	log.Info("Shutting down Kube-Vip")
-
-	return nil
-}

pkg/manager/manager_dump.go

@@ -12,7 +12,7 @@ import (
 )
 
 // dumpConfiguration prints the current configuration to stdout when SIGUSR1 is received
-func (sm *Manager) dumpConfiguration() {
+func (sm *Manager) dumpConfiguration(ctx context.Context) {
 	sm.mutex.Lock()
 	defer sm.mutex.Unlock()
 
@@ -29,7 +29,7 @@ func (sm *Manager) dumpConfiguration() {
 	sm.dumpConfigSection()
 	sm.dumpBGPSection()
 	sm.dumpARPSection()
-	sm.dumpServicesSection()
+	sm.dumpServicesSection(ctx)
 	sm.dumpNetworkInterfacesSection()
 	sm.dumpLeaderElectionSection()
 	sm.dumpRuntimeSection()
@@ -89,7 +89,7 @@ func (sm *Manager) dumpARPSection() {
 	fmt.Printf("\n")
 }
 
-func (sm *Manager) dumpServicesSection() {
+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 {
@@ -124,7 +124,7 @@ func (sm *Manager) dumpServicesSection() {
 			fmt.Println("--- KUBERNETES CONFIGURATION (SERVICES/ENDPOINTSLICES) ---")
 
 			fmt.Println("Service Configuration:")
-			svcList, err := sm.clientSet.CoreV1().Services(v1.NamespaceAll).List(context.TODO(), metav1.ListOptions{})
+			svcList, err := sm.clientSet.CoreV1().Services(v1.NamespaceAll).List(ctx, metav1.ListOptions{})
 			if err != nil {
 				fmt.Println("Unable to retrieve all Services")
 			} else {
@@ -144,7 +144,7 @@ func (sm *Manager) dumpServicesSection() {
 			}
 
 			fmt.Println("EndpointSlice Configuration (note endpoint names have -XXXXX prefixed):")
-			epList, err := sm.clientSet.DiscoveryV1().EndpointSlices(v1.NamespaceAll).List(context.TODO(), metav1.ListOptions{})
+			epList, err := sm.clientSet.DiscoveryV1().EndpointSlices(v1.NamespaceAll).List(ctx, metav1.ListOptions{})
 			if err != nil {
 				fmt.Println("Unable to retrieve all EndpointSlices")
 			} else {

pkg/manager/manager_dump_test.go

@@ -2,6 +2,7 @@ package manager
 
 import (
 	"bytes"
+	"context"
 	"io"
 	"os"
 	"testing"
@@ -36,7 +37,7 @@ func TestDumpConfiguration(t *testing.T) {
 	r, w, _ := os.Pipe()
 	os.Stdout = w
 
-	mgr.dumpConfiguration()
+	mgr.dumpConfiguration(context.TODO())
 
 	w.Close()
 	os.Stdout = old

pkg/manager/manager_table.go

@@ -1,203 +0,0 @@
-package manager
-
-import (
-	"context"
-	"fmt"
-	"syscall"
-	"time"
-
-	log "log/slog"
-
-	"github.com/kube-vip/kube-vip/pkg/cluster"
-	"github.com/kube-vip/kube-vip/pkg/endpoints"
-	"github.com/kube-vip/kube-vip/pkg/iptables"
-	"github.com/kube-vip/kube-vip/pkg/vip"
-	"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.Info("destination for routes", "table", sm.config.RoutingTableID, "protocol", 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.Error("error checking for old routes", "err", err)
-			}
-		}()
-	}
-
-	if sm.config.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() {
-		for {
-			sig := <-sm.signalChan
-			switch sig {
-			case syscall.SIGUSR1:
-				log.Info("Received SIGUSR1, dumping configuration")
-				sm.dumpConfiguration()
-			case syscall.SIGINT, syscall.SIGTERM:
-				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()
-				return
-			}
-		}
-	}()
-
-	if sm.config.EnableServices {
-		log.Debug("starting Services")
-		ns, err := returnNameSpace()
-		if err != nil {
-			log.Warn("unable to auto-detect namespace", "dropping to", 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.Info("beginning watching services, leaderelection will happen for every service")
-			err = sm.svcProcessor.StartServicesWatchForLeaderElection(ctx)
-			if err != nil {
-				return err
-			}
-		} else if sm.config.EnableLeaderElection {
-
-			log.Info("beginning services leadership", "namespace", ns, "lock name", plunderLock, "id", 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.svcProcessor.ServicesWatcher(ctx, sm.svcProcessor.SyncServices)
-						if err != nil {
-							log.Error(err.Error())
-							panic("")
-						}
-					},
-					OnStoppedLeading: func() {
-						// we can do cleanup here
-						sm.mutex.Lock()
-						defer sm.mutex.Unlock()
-						log.Info("leader lost", "id", id)
-						sm.svcProcessor.Stop()
-
-						log.Error("lost leadership, restarting kube-vip")
-						panic("")
-					},
-					OnNewLeader: func(identity string) {
-						// we're notified when new leader elected
-						if identity == id {
-							// I just got the lock
-							return
-						}
-						log.Info("new leader elected", "id", identity)
-					},
-				},
-			})
-		} else {
-			log.Info("beginning watching services without leader election")
-			err = sm.svcProcessor.ServicesWatcher(ctx, sm.svcProcessor.SyncServices)
-			if err != nil {
-				log.Error("Cannot watch services", "err", err)
-			} else {
-				log.Debug("watching services")
-			}
-		}
-	}
-	if sm.config.EnableControlPlane {
-		log.Debug("initCluster for ControlPlane")
-		cpCluster, err = cluster.InitCluster(sm.config, false, sm.intfMgr, sm.arpMgr)
-		if err != nil {
-			log.Debug("init of ControlPlane NOT successful")
-			return fmt.Errorf("cluster initialization error: %w", err)
-		}
-		log.Debug("init of ControlPlane successful")
-		log.Debug("init ClusterManager")
-		clusterManager, err := initClusterManager(sm)
-		if err != nil {
-			log.Debug("init cluster manager NOT successful")
-			return fmt.Errorf("cluster manager initialization error: %w", err)
-		}
-		log.Debug("init ClusterManager successful")
-		if err := cpCluster.StartVipService(sm.config, clusterManager, nil); err != nil {
-			log.Error("Control Plane", "err", err)
-			// Trigger the shutdown of this manager instance
-			sm.signalChan <- syscall.SIGINT
-		} else {
-			log.Debug("start VipServer for cluster manager successful")
-		}
-	}
-
-	return nil
-}
-
-func (sm *Manager) cleanRoutes() error {
-	sm.mutex.Lock()
-	defer sm.mutex.Unlock()
-	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 {
-			found = endpoints.CountRouteReferences(&routes[i], &sm.svcProcessor.ServiceInstances) > 0
-		}
-
-		if !found {
-			err = netlink.RouteDel(&(routes[i]))
-			if err != nil {
-				log.Error("[route] deletion", "route", routes[i], "err", err)
-			}
-			log.Debug("[route] deletion", "route", routes[i])
-		}
-
-	}
-	return nil
-}

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

@@ -1,131 +0,0 @@
-package manager
-
-import (
-	"context"
-	"syscall"
-	"time"
-
-	log "log/slog"
-
-	"github.com/kube-vip/kube-vip/pkg/wireguard"
-	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.Info("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() {
-		for {
-			sig := <-sm.signalChan
-			switch sig {
-			case syscall.SIGUSR1:
-				log.Info("Received SIGUSR1, dumping configuration")
-				sm.dumpConfiguration()
-			case syscall.SIGINT, syscall.SIGTERM:
-				log.Info("Received termination, signaling shutdown")
-				// Cancel the context, which will in turn cancel the leadership
-				cancel()
-				return
-			}
-		}
-	}()
-
-	ns, err = returnNameSpace()
-	if err != nil {
-		log.Warn("unable to auto-detect namespace", "dropping to", 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.Info("beginning watching services, leaderelection will happen for every service")
-		err = sm.svcProcessor.StartServicesWatchForLeaderElection(ctx)
-		if err != nil {
-			return err
-		}
-	} else {
-
-		log.Info("beginning services leadership", "namespace", ns, "lock name", plunderLock, "id", 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.svcProcessor.ServicesWatcher(ctx, sm.svcProcessor.SyncServices)
-					if err != nil {
-						log.Error(err.Error())
-						panic("")
-					}
-				},
-				OnStoppedLeading: func() {
-					// we can do cleanup here
-					sm.mutex.Lock()
-					defer sm.mutex.Unlock()
-					log.Info("leader lost", "id", id)
-					sm.svcProcessor.Stop()
-
-					log.Error("lost leadership, restarting kube-vip")
-					panic("")
-				},
-				OnNewLeader: func(identity string) {
-					// we're notified when new leader elected
-					if identity == id {
-						// I just got the lock
-						return
-					}
-					log.Info("new leader elected", "id", identity)
-				},
-			},
-		})
-	}
-	return nil
-}

pkg/manager/node_labeling.go

@@ -1,69 +0,0 @@
-package manager
-
-import (
-	"context"
-	"encoding/json"
-	"fmt"
-
-	log "log/slog"
-
-	"github.com/kube-vip/kube-vip/pkg/kubevip"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/apimachinery/pkg/types"
-	"k8s.io/client-go/kubernetes"
-)
-
-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.Error("can't query node labels", "node", id, "err", err)
-		return
-	}
-
-	log.Debug(fmt.Sprintf("node %s labels: %+v", id, node.Labels))
-
-	value, ok := node.Labels[kubevip.HasIP]
-	path := fmt.Sprintf("/metadata/labels/%s", kubevip.HasIPJSONPath)
-	log.Debug(fmt.Sprintf("Received identity: %s - id: %s", identity, id))
-	if ok && value == address {
-		log.Debug(fmt.Sprintf("removing node label `has-ip=%s` on %s", address, id))
-		// Remove label
-		applyPatchLabels(ctx, clientSet, id, "remove", path, address)
-	} else {
-		log.Debug(fmt.Sprintf("setting node label `has-ip=%s` on %s", address, id))
-		// Append label
-		applyPatchLabels(ctx, clientSet, id, "add", path, address)
-	}
-}
-
-// 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.Error("node patch marshaling failed", "err", err)
-		return
-	}
-	// patch node
-	node, err := clientSet.CoreV1().Nodes().Patch(ctx,
-		name, types.JSONPatchType, patchData, metav1.PatchOptions{})
-	if err != nil {
-		log.Error("node patching failed", "err", err)
-		return
-	}
-	log.Debug("updated", "node", name, "labels", node.Labels)
-}

pkg/manager/prom.go

@@ -1,15 +0,0 @@
-package manager
-
-import "github.com/prometheus/client_golang/prometheus"
-
-// PrometheusCollector defines a service watch event counter.
-func (sm *Manager) PrometheusCollector() []prometheus.Collector {
-	collectors := []prometheus.Collector{}
-	if sm.svcProcessor != nil {
-		collectors = append(collectors, sm.svcProcessor.CountServiceWatchEvent)
-	}
-	if sm.bgpServer != nil {
-		collectors = append(collectors, sm.bgpServer.BGPSessionInfoGauge)
-	}
-	return collectors
-}

pkg/manager/watch_annotations.go

@@ -17,6 +17,7 @@ import (
 	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"
@@ -25,18 +26,19 @@ 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.Info("Kube-Vip is waiting for annotation prefix to be present on this node", "prefix", 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
 	}
@@ -45,11 +47,11 @@ 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
 	}
 
@@ -57,30 +59,19 @@ func (sm *Manager) annotationsWatcher() error {
 	// they're as needed
 	log.Warn(err.Error())
 
-	// TODO, will need refactoring as part of rikatz work
-	rw, err := watchtools.NewRetryWatcherWithContext(context.TODO(), node.ResourceVersion, &cache.ListWatch{
+	rw, err := watchtools.NewRetryWatcherWithContext(ctx, node.ResourceVersion, &cache.ListWatch{
 		WatchFunc: func(_ metav1.ListOptions) (watch.Interface, error) {
-			return sm.rwClientSet.CoreV1().Nodes().Watch(context.Background(), listOptions)
+			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()
@@ -94,16 +85,17 @@ 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.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 {
@@ -130,10 +122,8 @@ func (sm *Manager) annotationsWatcher() error {
 		default:
 		}
 	}
-	close(exitFunction)
-	log.Info("Exiting Annotations watcher")
+	log.Info("[annotations] exiting annotations watcher")
 	return nil
-
 }
 
 // parseNodeAnnotations parses the annotations on the node and updates the configuration
@@ -154,9 +144,11 @@ func (sm *Manager) annotationsWatcher() error {
 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
 		}
@@ -173,8 +165,8 @@ func parseBgpAnnotations(bgpConfig kubevip.BGPConfig, node *v1.Node, prefix stri
 	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
 		}
@@ -188,8 +180,8 @@ func parseBgpAnnotations(bgpConfig kubevip.BGPConfig, node *v1.Node, prefix stri
 	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
 		}
@@ -206,8 +198,8 @@ func parseBgpAnnotations(bgpConfig kubevip.BGPConfig, node *v1.Node, prefix stri
 	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 + ","
 		}
@@ -215,8 +207,13 @@ func parseBgpAnnotations(bgpConfig kubevip.BGPConfig, node *v1.Node, prefix stri
 	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([]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)
 
@@ -225,8 +222,7 @@ func parseBgpAnnotations(bgpConfig kubevip.BGPConfig, node *v1.Node, prefix stri
 			// 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
 				}
 			}
@@ -239,6 +235,21 @@ func parseBgpAnnotations(bgpConfig kubevip.BGPConfig, node *v1.Node, prefix stri
 				// 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)
 		}
 	}

pkg/manager/watcher_test.go

@@ -25,9 +25,11 @@ func TestParseBgpAnnotations(t *testing.T) {
 	}
 
 	node.Annotations = map[string]string{
-		"bgp/node-asn": "65000",
-		"bgp/peer-asn": "64000",
-		"bgp/src-ip":   "10.0.0.254",
+		"bgp/node-asn":       "65000",
+		"bgp/peer-asn":       "64000",
+		"bgp/src-ip":         "10.0.0.254",
+		"bgp/peer-ip":        "10.0.0.1",
+		"bgp/peer-multi-hop": "true",
 	}
 
 	bgpConfig, bgpPeer, err := parseBgpAnnotations(bgpConfigBase, node, "bgp")
@@ -38,15 +40,17 @@ func TestParseBgpAnnotations(t *testing.T) {
 	assert.Equal(t, uint32(65000), bgpConfig.AS, "bgpConfig.AS parsed incorrectly")
 	assert.Equal(t, uint32(64000), bgpPeer.AS, "bgpPeer.AS parsed incorrectly")
 	assert.Equal(t, "10.0.0.254", bgpConfig.RouterID, "bgpConfig.RouterID parsed incorrectly")
+	assert.Equal(t, true, bgpPeer.MultiHop, "bgpPeer.MultiHop parsed incorrectly")
 	assert.EqualValues(t, 15, bgpConfig.HoldTime, "base bgpConfig.HoldTime should not be overwritten")
 	assert.EqualValues(t, 5, bgpConfig.KeepaliveInterval, "base bgpConfig.KeepaliveInterval should not be overwritten")
 
-	node.Annotations = map[string]string{
-		"bgp/node-asn": "65000",
-		"bgp/peer-asn": "64000",
-		"bgp/src-ip":   "10.0.0.254",
-		"bgp/peer-ip":  "10.0.0.1,10.0.0.2,10.0.0.3",
-		"bgp/bgp-pass": "cGFzc3dvcmQ=", // password
+	node.Annotations = map[string]string{ //nolint:gosec
+		"bgp/node-asn":       "65000",
+		"bgp/peer-asn":       "64000",
+		"bgp/src-ip":         "10.0.0.254",
+		"bgp/peer-ip":        "10.0.0.1,10.0.0.2,10.0.0.3",
+		"bgp/bgp-pass":       "cGFzc3dvcmQ=", // dummy password for the test, gosec linter disabled
+		"bgp/peer-multi-hop": "true",
 	}
 
 	bgpConfig, bgpPeer, err = parseBgpAnnotations(bgpConfigBase, node, "bgp")
@@ -55,9 +59,9 @@ func TestParseBgpAnnotations(t *testing.T) {
 	}
 
 	bgpPeers := []kubevip.BGPPeer{
-		{Address: "10.0.0.1", AS: uint32(64000), Password: "password"},
-		{Address: "10.0.0.2", AS: uint32(64000), Password: "password"},
-		{Address: "10.0.0.3", AS: uint32(64000), Password: "password"},
+		{Address: "10.0.0.1", AS: uint32(64000), Password: "password", MultiHop: true},
+		{Address: "10.0.0.2", AS: uint32(64000), Password: "password", MultiHop: true},
+		{Address: "10.0.0.3", AS: uint32(64000), Password: "password", MultiHop: true},
 	}
 	assert.Equal(t, bgpPeers, bgpConfig.Peers, "bgpConfig.Peers parsed incorrectly")
 	assert.Equal(t, "10.0.0.3", bgpPeer.Address, "bgpPeer.Address parsed incorrectly")
@@ -86,12 +90,12 @@ func TestParseNewBgpAnnotations(t *testing.T) {
 		t.Fatal("Parsing BGP annotations should return an error when no annotations exist")
 	}
 
-	node.Annotations = map[string]string{
+	node.Annotations = map[string]string{ //nolint:gosec
 		"bgp/bgp-peers-0-node-asn": "65000",
 		"bgp/bgp-peers-0-peer-asn": "64000",
 		"bgp/bgp-peers-0-peer-ip":  "10.0.0.1,10.0.0.2,10.0.0.3",
 		"bgp/bgp-peers-0-src-ip":   "10.0.0.254",
-		"bgp/bgp-peers-0-bgp-pass": "cGFzc3dvcmQ=", // password
+		"bgp/bgp-peers-0-bgp-pass": "cGFzc3dvcmQ=", // dummy password for the test, gosec linter disabled
 	}
 
 	bgpConfig, bgpPeer, err := parseBgpAnnotations(bgpConfigBase, node, "bgp")

pkg/manager/worker/arp.go

@@ -0,0 +1,73 @@
+package worker
+
+import (
+	"context"
+	log "log/slog"
+	"sync"
+	"sync/atomic"
+
+	"github.com/kube-vip/kube-vip/pkg/arp"
+	"github.com/kube-vip/kube-vip/pkg/election"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/lease"
+	"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/services"
+	"k8s.io/client-go/kubernetes"
+)
+
+type ARP struct {
+	Common
+}
+
+func NewARP(arpMgr *arp.Manager, intfMgr *networkinterface.Manager,
+	config *kubevip.Config, closing *atomic.Bool, killFunc func(),
+	svcProcessor *services.Processor, mutex *sync.Mutex, clientSet *kubernetes.Clientset,
+	electionMgr *election.Manager, leaseMgr *lease.Manager, routeMgr *route.Manager,
+	nodeLabelMgr node.Labeler) *ARP {
+	return &ARP{
+		Common: *newCommon(arpMgr, intfMgr, config, closing, killFunc,
+			svcProcessor, mutex, clientSet, electionMgr, leaseMgr, routeMgr,
+			nodeLabelMgr),
+	}
+}
+
+func (a *ARP) Configure(ctx context.Context, wg *sync.WaitGroup) error {
+	log.Info("Start ARP/NDP advertisement Global")
+	wg.Go(func() {
+		a.arpMgr.StartAdvertisement(ctx)
+	})
+	return nil
+}
+
+func (a *ARP) StartControlPlane(ctx context.Context, electionManager *election.Manager) {
+	err := a.cpCluster.StartCluster(ctx, a.config, electionManager, nil, a.leaseMgr, a.killFunc)
+	if err != nil {
+		log.Error("starting control plane", "err", err)
+	}
+
+	// Trigger the shutdown of this manager instance
+	a.killFunc()
+}
+
+func (a *ARP) ConfigureServices() {
+
+}
+
+func (a *ARP) StartServices(ctx context.Context) error {
+	// 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 a.config.EnableServicesElection {
+		if err := a.PerServiceLeader(ctx); err != nil {
+			return err
+		}
+	} else {
+		a.GlobalLeader(ctx, a.config.ServicesLeaseName)
+	}
+	return nil
+}
+
+func (a *ARP) Name() string {
+	return "ARP"
+}

pkg/manager/worker/bgp.go

@@ -0,0 +1,121 @@
+package worker
+
+import (
+	"context"
+	"fmt"
+	log "log/slog"
+	"sync"
+	"sync/atomic"
+
+	"github.com/kube-vip/kube-vip/pkg/arp"
+	"github.com/kube-vip/kube-vip/pkg/bgp"
+	"github.com/kube-vip/kube-vip/pkg/election"
+	"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/networkinterface"
+	"github.com/kube-vip/kube-vip/pkg/node"
+	"github.com/kube-vip/kube-vip/pkg/route"
+	"github.com/kube-vip/kube-vip/pkg/services"
+	api "github.com/osrg/gobgp/v3/api"
+
+	"github.com/prometheus/client_golang/prometheus"
+	"k8s.io/client-go/kubernetes"
+)
+
+type BGP struct {
+	Common
+	bgpServer *bgp.Server
+}
+
+func NewBGP(arpMgr *arp.Manager, intfMgr *networkinterface.Manager,
+	config *kubevip.Config, closing *atomic.Bool, killFunc func(),
+	svcProcessor *services.Processor, mutex *sync.Mutex, clientSet *kubernetes.Clientset,
+	bgpServer *bgp.Server, electionMgr *election.Manager, leaseMgr *lease.Manager,
+	routeMgr *route.Manager, nodeLabelMgr node.Labeler) *BGP {
+	return &BGP{
+		Common: *newCommon(arpMgr, intfMgr, config, closing, killFunc,
+			svcProcessor, mutex, clientSet, electionMgr, leaseMgr, routeMgr,
+			nodeLabelMgr),
+		bgpServer: bgpServer,
+	}
+}
+
+func (b *BGP) Configure(ctx context.Context, _ *sync.WaitGroup) error {
+	var err error
+	if b.bgpServer == nil {
+		b.bgpServer, err = bgp.NewBGPServer(b.config.BGPConfig)
+		if err != nil {
+			return fmt.Errorf("creating BGP server: %w", err)
+		}
+	}
+
+	log.Info("Starting the BGP server to advertise VIP routes to BGP peers")
+	if err := b.bgpServer.Start(ctx, 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
+			}
+
+			metrics.BGPSessionInfoGauge.With(prometheus.Labels{
+				"state": stateName,
+				"peer":  peerDescription,
+			}).Set(metricValue)
+		}
+	}); err != nil {
+		return fmt.Errorf("starting BGP server: %w", err)
+	}
+
+	return nil
+}
+
+func (b *BGP) Cleanup() {
+	// Defer a function to check if the bgpServer has been created and if so attempt to close it
+	if b.bgpServer != nil {
+		b.bgpServer.Close()
+	}
+}
+
+func (b *BGP) StartControlPlane(ctx context.Context, electionManager *election.Manager) {
+	var err error
+	if b.config.EnableLeaderElection {
+		err = b.cpCluster.StartCluster(ctx, b.config, electionManager, b.bgpServer, b.leaseMgr, b.killFunc)
+	} else {
+		err = b.cpCluster.StartVipService(ctx, b.config, electionManager, b.bgpServer, b.killFunc)
+	}
+	if err != nil {
+		log.Error("Control Plane", "err", err)
+		// Trigger the shutdown of this manager instance
+		b.killFunc()
+	}
+}
+
+func (b *BGP) ConfigureServices() {
+	// No configuration required
+}
+
+func (b *BGP) StartServices(ctx context.Context) error {
+	if b.config.EnableServicesElection {
+		if err := b.PerServiceLeader(ctx); err != nil {
+			return err
+		}
+	} else {
+		if err := b.ServicesNoLeader(ctx); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func (b *BGP) ServicesGlobalLeader(ctx context.Context, id string) {
+	// NOT IMPLEMENTED
+}
+
+func (b *BGP) Name() string {
+	return "BGP"
+}

pkg/manager/worker/common.go

@@ -0,0 +1,220 @@
+package worker
+
+import (
+	"context"
+	"fmt"
+	log "log/slog"
+	"sync"
+	"sync/atomic"
+
+	"github.com/kube-vip/kube-vip/pkg/arp"
+	"github.com/kube-vip/kube-vip/pkg/cluster"
+	"github.com/kube-vip/kube-vip/pkg/election"
+	"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/networkinterface"
+	"github.com/kube-vip/kube-vip/pkg/node"
+	"github.com/kube-vip/kube-vip/pkg/route"
+	"github.com/kube-vip/kube-vip/pkg/services"
+	"k8s.io/client-go/kubernetes"
+)
+
+type Common struct {
+	arpMgr       *arp.Manager
+	cpCluster    *cluster.Cluster
+	intfMgr      *networkinterface.Manager
+	config       *kubevip.Config
+	closing      *atomic.Bool
+	killFunc     func()
+	svcProcessor *services.Processor
+	mutex        *sync.Mutex
+	clientSet    *kubernetes.Clientset
+	electionMgr  *election.Manager
+	leaseMgr     *lease.Manager
+	routeMgr     *route.Manager
+	nodeLabelMgr node.Labeler
+}
+
+func newCommon(arpMgr *arp.Manager, intfMgr *networkinterface.Manager,
+	config *kubevip.Config, closing *atomic.Bool, killFunc func(),
+	svcProcessor *services.Processor, mutex *sync.Mutex, clientSet *kubernetes.Clientset,
+	electionMgr *election.Manager, leaseMgr *lease.Manager, routeMgr *route.Manager,
+	nodeLabelMgr node.Labeler) *Common {
+	return &Common{
+		arpMgr:       arpMgr,
+		intfMgr:      intfMgr,
+		config:       config,
+		closing:      closing,
+		killFunc:     killFunc,
+		svcProcessor: svcProcessor,
+		mutex:        mutex,
+		clientSet:    clientSet,
+		electionMgr:  electionMgr,
+		leaseMgr:     leaseMgr,
+		routeMgr:     routeMgr,
+		nodeLabelMgr: nodeLabelMgr,
+	}
+}
+
+func (c *Common) InitControlPlane() error {
+	var err error
+	c.cpCluster, err = cluster.InitCluster(c.config, false, c.intfMgr, c.arpMgr, c.routeMgr, c.nodeLabelMgr)
+	if err != nil {
+		return fmt.Errorf("cluster initialization error: %w", err)
+	}
+	return nil
+}
+
+func (c *Common) PerServiceLeader(ctx context.Context) error {
+	log.Info("beginning watching services, leaderelection will happen for every service")
+	err := c.svcProcessor.StartServicesWatchForLeaderElection(ctx)
+	if err != nil {
+		return err
+	}
+	return nil
+}
+
+func (c *Common) GlobalLeader(ctx context.Context, leaseName string) {
+	c.runGlobalElection(ctx, c, leaseName, c.config, c.electionMgr)
+}
+
+func (c *Common) ServicesNoLeader(ctx context.Context) error {
+	log.Info("beginning watching services without leader election")
+	err := c.svcProcessor.ServicesWatcher(ctx, services.NewCallback(c.svcProcessor.SyncServices, false))
+	if err != nil {
+		return fmt.Errorf("error while watching services: %w", err)
+	}
+	return nil
+}
+
+func (c *Common) Cleanup() {
+	// NOT IMPLEMENTED
+}
+
+func (c *Common) OnStartedLeading(ctx context.Context) {
+	err := c.svcProcessor.ServicesWatcher(ctx, services.NewCallback(c.svcProcessor.SyncServices, false))
+	if err != nil {
+		log.Error("service watcher", "err", err)
+		c.killFunc()
+	}
+}
+
+func (c *Common) OnStoppedLeading() {
+	// we can do cleanup here
+	c.mutex.Lock()
+	defer c.mutex.Unlock()
+	log.Info("leader lost", "former leader", c.config.NodeName)
+	c.svcProcessor.Stop()
+
+	log.Error("lost services leadership, restarting kube-vip")
+	c.killFunc()
+}
+
+func (c *Common) OnNewLeader(identity string) {
+	if identity == c.config.NodeName {
+		// I just got the lock
+		return
+	}
+	log.Info("new leader elected", "new leader", identity)
+}
+
+func (c *Common) runGlobalElection(ctx context.Context, a election.Actions, leaseName string,
+	config *kubevip.Config, electionManager *election.Manager) {
+
+	ns, leaseName := lease.NamespaceName(leaseName, config)
+
+	leaseID := lease.NewID(config.LeaderElectionType, ns, leaseName)
+	objectName := lease.ObjectName(leaseID, "svcs0")
+
+	// objLease, isNew, isSharedLease := c.leaseMgr.Add(leaseID, objectName)
+
+	objLease := c.leaseMgr.Add(ctx, leaseID)
+	isNew := objLease.Add(objectName)
+
+	// this service was already processed so we do not need to do anything
+	if !isNew {
+		log.Debug("this election was already done, waiting for it to finish", "lease", c.config.ServicesLeaseName)
+		// Wait for either the service context or lease context to be done
+		select {
+		case <-ctx.Done():
+			// Service was deleted
+			c.leaseMgr.Delete(leaseID, objectName)
+		case <-objLease.Ctx.Done():
+			// Leader election ended (leadership lost or context cancelled)
+		}
+		return
+	}
+
+	objLease.Lock()
+
+	defer func() {
+		objLease.Unlock()
+	}()
+
+	if objLease.Elected.Load() {
+		objLease.Unlock()
+		log.Debug("this election was already done, shared lease", "lease", leaseID.Name())
+
+		// 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", leaseID.Name())
+			return
+		}
+
+		a.OnStartedLeading(objLease.Ctx)
+
+		log.Debug("waiting for lease to finish", "lease", leaseID.Name())
+		// wait for leaderelection to be finished
+		<-objLease.Ctx.Done()
+
+		// we can do cleanup here
+		a.OnStoppedLeading()
+
+		log.Error("lost leadership, restarting kube-vip", "lease", leaseID.Name())
+		c.killFunc()
+
+		return
+	}
+
+	// For new leases (not shared), ensure cleanup when the leader election ends
+	// This is critical for the restartable service watcher to be able to restart
+	// the leader election after leadership loss
+	defer func() {
+		// Delete the lease from the manager so subsequent calls can create a fresh lease
+		// This handles the case where leader election ends due to:
+		// 1. Leadership loss (e.g., network timeout)
+		// 2. Context cancellation
+		// 3. Any other reason RunOrDie returns
+		c.leaseMgr.Delete(leaseID, objectName)
+	}()
+
+	run := &election.RunConfig{
+		Config:           config,
+		LeaseID:          leaseID,
+		LeaseAnnotations: map[string]string{},
+		Mgr:              electionManager,
+		OnStartedLeading: func(ctx context.Context) {
+			objLease.Elected.Store(true)
+			objLease.Unlock()
+			close(objLease.Started)
+			a.OnStartedLeading(ctx)
+			metrics.LeaderTransitionsTotal.WithLabelValues(leaseID.Name()).Inc()
+			metrics.IsLeader.WithLabelValues(config.NodeName, leaseID.Name()).Set(1)
+		},
+		OnStoppedLeading: func() {
+			objLease.Elected.Store(false)
+			a.OnStoppedLeading()
+			metrics.IsLeader.WithLabelValues(config.NodeName, leaseID.Name()).Set(0)
+		},
+		OnNewLeader: a.OnNewLeader,
+	}
+
+	if err := election.RunOrDie(ctx, run, config); err != nil {
+		log.Error("leaderelection failed", "err", err, "id", config.NodeName, "name", leaseID.Name())
+	}
+}

pkg/manager/worker/table.go

@@ -0,0 +1,116 @@
+package worker
+
+import (
+	"context"
+	"fmt"
+	log "log/slog"
+	"sync"
+	"sync/atomic"
+	"time"
+
+	"github.com/kube-vip/kube-vip/pkg/arp"
+	"github.com/kube-vip/kube-vip/pkg/election"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/lease"
+	"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/services"
+	"github.com/kube-vip/kube-vip/pkg/vip"
+	"github.com/vishvananda/netlink"
+	"k8s.io/client-go/kubernetes"
+)
+
+type Table struct {
+	Common
+}
+
+func NewTable(arpMgr *arp.Manager, intfMgr *networkinterface.Manager,
+	config *kubevip.Config, closing *atomic.Bool, killFUnc func(),
+	svcProcessor *services.Processor, mutex *sync.Mutex, clientSet *kubernetes.Clientset,
+	electionMgr *election.Manager, leaseMgr *lease.Manager, routeMgr *route.Manager,
+	nodeLabelMgr node.Labeler) *Table {
+	return &Table{
+		Common: *newCommon(arpMgr, intfMgr, config, closing, killFUnc,
+			svcProcessor, mutex, clientSet, electionMgr, leaseMgr, routeMgr,
+			nodeLabelMgr),
+	}
+}
+
+func (t *Table) Configure(ctx context.Context, wg *sync.WaitGroup) error {
+	log.Info("destination for routes", "table", t.config.RoutingTableID, "protocol", t.config.RoutingProtocol)
+
+	if t.config.CleanRoutingTable {
+		wg.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 := t.cleanRoutes(); err != nil {
+				log.Error("error checking for old routes", "err", err)
+			}
+		})
+	}
+	return nil
+}
+
+func (t *Table) StartControlPlane(ctx context.Context, electionManager *election.Manager) {
+	if err := t.cpCluster.StartVipService(ctx, t.config, electionManager, nil, t.killFunc); err != nil {
+		log.Error("Control Plane", "err", err)
+		// Trigger the shutdown of this manager instance
+		t.killFunc()
+	} else {
+		log.Debug("start VipServer for cluster manager successful")
+	}
+}
+
+func (t *Table) ConfigureServices() {
+	// No configuration required
+}
+
+func (t *Table) StartServices(ctx context.Context) error {
+	log.Debug("starting Services")
+
+	if t.config.EnableServicesElection {
+		if err := t.PerServiceLeader(ctx); err != nil {
+			return err
+		}
+	} else if t.config.EnableLeaderElection {
+		t.GlobalLeader(ctx, t.config.ServicesLeaseName)
+	} else {
+		if err := t.ServicesNoLeader(ctx); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func (t *Table) Name() string {
+	return "Routing Table"
+}
+
+func (t *Table) cleanRoutes() error {
+	t.mutex.Lock()
+	defer t.mutex.Unlock()
+	routes, err := vip.ListRoutes(t.config.RoutingTableID, t.config.RoutingProtocol)
+	if err != nil {
+		return fmt.Errorf("error getting routes: %w", err)
+	}
+
+	for i := range routes {
+		found := false
+		if t.config.EnableControlPlane {
+			found = (routes[i].Dst.IP.String() == t.config.Address)
+		} else {
+			found = t.routeMgr.Check(routes[i].String())
+		}
+
+		if !found {
+			err = netlink.RouteDel(&(routes[i]))
+			if err != nil {
+				log.Error("[route] deletion", "route", routes[i], "err", err)
+			}
+			log.Debug("[route] deletion", "route", routes[i])
+		}
+
+	}
+	return nil
+}

pkg/manager/worker/wireguard.go

@@ -0,0 +1,278 @@
+package worker
+
+import (
+	"context"
+	"fmt"
+	log "log/slog"
+	"os"
+	"sync"
+	"sync/atomic"
+
+	"github.com/kube-vip/kube-vip/pkg/arp"
+	"github.com/kube-vip/kube-vip/pkg/election"
+	"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/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/sysctl"
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	"github.com/kube-vip/kube-vip/pkg/wireguard"
+	v1 "k8s.io/api/core/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/watch"
+	"k8s.io/client-go/kubernetes"
+)
+
+type WireGuard struct {
+	Common
+	tunnelMgr           *wireguard.TunnelManager
+	kubeAPIHost         string
+	kubeAPIPort         string
+	endpointWatcherCtx  context.Context
+	endpointWatcherStop context.CancelFunc
+	endpointWatcherWg   sync.WaitGroup
+}
+
+func NewWireGuard(arpMgr *arp.Manager, intfMgr *networkinterface.Manager,
+	config *kubevip.Config, closing *atomic.Bool, killFUnc func(),
+	svcProcessor *services.Processor, mutex *sync.Mutex, clientSet *kubernetes.Clientset,
+	electionMgr *election.Manager, leaseMgr *lease.Manager, routeMgr *route.Manager,
+	nodeLabelMgr node.Labeler) *WireGuard {
+	return &WireGuard{
+		Common: *newCommon(arpMgr, intfMgr, config, closing, killFUnc,
+			svcProcessor, mutex, clientSet, electionMgr, leaseMgr, routeMgr,
+			nodeLabelMgr),
+	}
+}
+
+func (w *WireGuard) Configure(ctx context.Context, _ *sync.WaitGroup) error {
+	log.Info("reading wireguard tunnel configurations from Kubernetes secret")
+	tunnelMgr := wireguard.NewTunnelManager()
+
+	err := tunnelMgr.LoadConfigurationsFromSecret(ctx, w.clientSet, w.config.Namespace, "wireguard")
+	if err != nil {
+		return fmt.Errorf("failed to load WireGuard tunnel configurations: %w", err)
+	}
+
+	// Clean up any stale resources from previous runs (crash recovery for hostNetwork: true)
+	// Must be called AFTER loading configs so we know which interfaces/ports to clean
+	if err := tunnelMgr.CleanupStaleResources(); err != nil {
+		log.Warn("failed to cleanup stale resources", "err", err)
+		// Continue anyway - the cleanup is best-effort
+	}
+
+	if _, err := sysctl.EnableProcSys("/proc/sys/net/ipv4/conf/all/src_valid_mark"); err != nil {
+		return fmt.Errorf("net.ipv4.conf.all.src_valid_mark is disabled and could not be enabled %w", err)
+	}
+	if _, err := sysctl.EnableProcSys("/proc/sys/net/ipv4/conf/all/route_localnet"); err != nil {
+		return fmt.Errorf("net.ipv4.conf.all.route_localnet is disabled and could not be enabled %w", err)
+	}
+
+	w.tunnelMgr = tunnelMgr
+	configuredVIPs := tunnelMgr.ListConfiguredTunnels()
+	log.Info("loaded WireGuard tunnel configurations", "vips", configuredVIPs)
+
+	return nil
+}
+
+func (w *WireGuard) InitControlPlane() error {
+	// Get Kubernetes service IP and port from environment
+	w.kubeAPIHost = os.Getenv("KUBERNETES_SERVICE_HOST")
+	w.kubeAPIPort = os.Getenv("KUBERNETES_SERVICE_PORT_HTTPS")
+	if w.kubeAPIHost == "" || w.kubeAPIPort == "" {
+		return fmt.Errorf("KUBERNETES_SERVICE_HOST or KUBERNETES_SERVICE_PORT_HTTPS not set")
+	}
+	return nil
+}
+
+func (w *WireGuard) StartControlPlane(ctx context.Context, electionManager *election.Manager) {
+	if !w.tunnelMgr.HasConfigForVIP(w.config.VIP) {
+		log.Error("no WireGuard tunnel configuration found for control plane VIP", "vip", w.config.VIP)
+		return
+	}
+	w.runGlobalElection(ctx, w, w.config.LeaseName, w.config, electionManager)
+}
+
+func (w *WireGuard) ConfigureServices() {
+	w.svcProcessor.TunnelMgr = w.tunnelMgr
+}
+
+func (w *WireGuard) StartServices(ctx context.Context) error {
+	if w.config.EnableServicesElection {
+		log.Info("beginning watching services, leaderelection will happen for every service")
+		err := w.svcProcessor.StartServicesWatchForLeaderElection(ctx)
+		if err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func (w *WireGuard) Name() string {
+	return "WireGuard"
+}
+
+func (w *WireGuard) OnStartedLeading(ctx context.Context) {
+	// Bring up the WireGuard tunnel for control plane VIP
+	err := w.tunnelMgr.BringUpTunnelForVIP(w.config.VIP)
+	if err != nil {
+		log.Error("could not start wireguard tunnel for control plane", "vip", w.config.VIP, "err", err)
+		_ = w.tunnelMgr.TearDownTunnelForVIP(w.config.VIP)
+		w.killFunc()
+		return
+	}
+
+	// Get the tunnel to access its configuration
+	wg := w.tunnelMgr.GetTunnelForVIP(w.config.VIP)
+	if wg == nil {
+		log.Error("failed to get wireguard tunnel after bringing up", "vip", w.config.VIP)
+		w.killFunc()
+		return
+	}
+
+	tunnelConfig := w.tunnelMgr.GetConfigForVIP(w.config.VIP)
+	if tunnelConfig == nil {
+		log.Error("failed to get tunnel configuration", "vip", w.config.VIP)
+		_ = w.tunnelMgr.TearDownTunnelForVIP(w.config.VIP)
+		w.killFunc()
+		return
+	}
+
+	// Start endpoint watcher - DNAT rules will be applied when endpoints arrive
+	w.endpointWatcherCtx, w.endpointWatcherStop = context.WithCancel(ctx)
+	w.endpointWatcherWg.Go(func() {
+		w.watchKubernetesEndpoints(w.endpointWatcherCtx, tunnelConfig)
+	})
+
+	if w.config.EnableServices && !w.config.EnableServicesElection {
+		if err := w.svcProcessor.ServicesWatcher(ctx, services.NewCallback(w.svcProcessor.SyncServices, false)); err != nil {
+			log.Error("failed to start services watcher", "err", err)
+		}
+	}
+}
+
+// watchKubernetesEndpoints watches the kubernetes service EndpointSlices for changes
+// and updates the DNAT rules when API server endpoints change (e.g., when an API server goes down)
+func (w *WireGuard) watchKubernetesEndpoints(ctx context.Context, tunnelConfig *wireguard.TunnelConfig) {
+	log.Info("starting kubernetes endpoint watcher for control plane")
+
+	kubeSvc := &v1.Service{
+		ObjectMeta: metav1.ObjectMeta{
+			Name:      "kubernetes",
+			Namespace: "default",
+		},
+	}
+
+	provider := providers.NewEndpointslices()
+	rw, err := provider.CreateRetryWatcher(ctx, w.clientSet, kubeSvc)
+	if err != nil {
+		log.Error("failed to create kubernetes endpoint watcher", "err", err)
+		return
+	}
+	defer rw.Stop()
+
+	for event := range rw.ResultChan() {
+		select {
+		case <-ctx.Done():
+			log.Info("kubernetes endpoint watcher stopped")
+			return
+		default:
+		}
+
+		switch event.Type {
+		case watch.Added, watch.Modified, watch.Deleted:
+			if err := provider.LoadObject(event.Object, func() {}); err != nil {
+				log.Error("failed to load endpoint object", "err", err)
+				continue
+			}
+			endpoints, _ := provider.GetAllEndpoints()
+			log.Info("kubernetes endpoints changed, updating DNAT rules", "eventType", event.Type, "endpoints", endpoints)
+			if err := w.updateControlPlaneDNAT(tunnelConfig, endpoints); err != nil {
+				log.Error("failed to update control plane DNAT rules", "err", err)
+			}
+		case watch.Error:
+			log.Warn("kubernetes endpoint watch error", "event", event)
+		}
+	}
+}
+
+// updateControlPlaneDNAT updates the DNAT rules for the control plane with the given endpoints
+func (w *WireGuard) updateControlPlaneDNAT(tunnelConfig *wireguard.TunnelConfig, endpoints []string) error {
+	if len(endpoints) == 0 {
+		log.Warn("no kubernetes API server endpoints available")
+		// Don't delete rules - keep routing to last known endpoints
+		return nil
+	}
+
+	// Build targets with default port 6443
+	targets := make([]nftables.DNATTarget, len(endpoints))
+	for i, ep := range endpoints {
+		targets[i] = nftables.DNATTarget{IP: ep, Port: 6443}
+	}
+
+	vipIP := utils.StripCIDR(w.config.VIP)
+
+	err := nftables.ApplyDNAT(
+		tunnelConfig.InterfaceName,
+		vipIP,
+		6443,
+		targets,
+		"controlplane",
+		v1.ProtocolTCP,
+		false,
+		tunnelConfig.ListenPort,
+	)
+	if err != nil {
+		return fmt.Errorf("failed to apply updated DNAT rule: %w", err)
+	}
+
+	log.Info("control plane DNAT rules updated", "targetCount", len(targets))
+	return nil
+}
+
+func (w *WireGuard) OnStoppedLeading() {
+	// we can do cleanup here
+	w.mutex.Lock()
+	defer w.mutex.Unlock()
+	log.Info("leader lost", "id", w.config.NodeName)
+
+	// Stop the kubernetes endpoint watcher and wait for it to finish
+	if w.endpointWatcherStop != nil {
+		w.endpointWatcherStop()
+		w.endpointWatcherWg.Wait()
+	}
+
+	log.Info("deleting nftables DNAT chains")
+	err := nftables.DeleteIngressChains(false, "controlplane")
+	if err != nil {
+		log.Error("could not delete DNAT ingress chains", "err", err)
+	} else {
+		log.Info("nftables DNAT chains deleted successfully")
+	}
+
+	// Tear down all tunnels (control plane + services)
+	err = w.tunnelMgr.TearDownAllTunnels()
+	if err != nil {
+		log.Error("failed to tear down tunnels", "err", err)
+	}
+	if w.config.EnableServices && !w.config.EnableServicesElection {
+		w.svcProcessor.Stop()
+	}
+	log.Error("lost control plane leadership, restarting kube-vip")
+	w.killFunc()
+}
+
+func (w *WireGuard) OnNewLeader(identity string) {
+	// we're notified when new leader elected
+	if identity == w.config.NodeName {
+		// I just got the lock
+		return
+	}
+	// safety check - tear down tunnel if we're not the leader
+	_ = w.tunnelMgr.TearDownTunnelForVIP(w.config.VIP)
+	log.Info("new leader elected", "id", identity)
+}

pkg/manager/worker/worker.go

@@ -0,0 +1,60 @@
+package worker
+
+import (
+	"context"
+	"sync"
+	"sync/atomic"
+
+	"github.com/kube-vip/kube-vip/pkg/arp"
+	"github.com/kube-vip/kube-vip/pkg/bgp"
+	"github.com/kube-vip/kube-vip/pkg/election"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/lease"
+	"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/services"
+	"k8s.io/client-go/kubernetes"
+)
+
+type Worker interface {
+	Configure(context.Context, *sync.WaitGroup) error
+	InitControlPlane() error
+	StartControlPlane(context.Context, *election.Manager)
+	ConfigureServices()
+	StartServices(ctx context.Context) error
+	Name() string
+	Cleanup()
+}
+
+func New(arpMgr *arp.Manager, intfMgr *networkinterface.Manager,
+	config *kubevip.Config, closing *atomic.Bool, killFunc func(),
+	svcProcessor *services.Processor, mutex *sync.Mutex, clientSet *kubernetes.Clientset,
+	bgpServer *bgp.Server, electionMgr *election.Manager, leaseMgr *lease.Manager,
+	routeMgr *route.Manager, nodeLabelMgr node.Labeler) Worker {
+	if config.EnableARP {
+		return NewARP(arpMgr, intfMgr, config, closing, killFunc,
+			svcProcessor, mutex, clientSet, electionMgr, leaseMgr, routeMgr,
+			nodeLabelMgr)
+	}
+
+	if config.EnableBGP {
+		return NewBGP(arpMgr, intfMgr, config, closing, killFunc,
+			svcProcessor, mutex, clientSet, bgpServer,
+			electionMgr, leaseMgr, routeMgr, nodeLabelMgr)
+	}
+
+	if config.EnableRoutingTable {
+		return NewTable(arpMgr, intfMgr, config, closing, killFunc,
+			svcProcessor, mutex, clientSet, electionMgr, leaseMgr, routeMgr,
+			nodeLabelMgr)
+	}
+
+	if config.EnableWireguard {
+		return NewWireGuard(arpMgr, intfMgr, config, closing, killFunc,
+			svcProcessor, mutex, clientSet, electionMgr, leaseMgr, routeMgr,
+			nodeLabelMgr)
+	}
+
+	return nil
+}

pkg/metrics/prometheus.go

@@ -0,0 +1,68 @@
+package metrics
+
+import "github.com/prometheus/client_golang/prometheus"
+
+var (
+	// Service / VIP Lifecycle
+	ActiveServices = prometheus.NewGaugeVec(
+		prometheus.GaugeOpts{Name: "kube_vip_active_services", Help: "How many LB services are currently managed"},
+		[]string{"namespace"},
+	)
+	ServiceReconcileErrorsTotal = prometheus.NewCounterVec(
+		prometheus.CounterOpts{Name: "kube_vip_service_reconcile_errors_total", Help: "Count reconcile failures"},
+		[]string{"namespace", "name", "reason"},
+	)
+	ServiceReconcileDuration = prometheus.NewHistogramVec(
+		prometheus.HistogramOpts{Name: "kube_vip_service_reconcile_duration_seconds", Help: "How long AddOrModify takes end-to-end"},
+		[]string{"namespace"},
+	)
+
+	// This is a prometheus counter used to count the number of events received
+	// from the service watcher
+	CountServiceWatchEvent = prometheus.NewCounterVec(prometheus.CounterOpts{
+		Namespace: "kube_vip",
+		Subsystem: "manager",
+		Name:      "all_services_events",
+		Help:      "Count all events fired by the service watcher categorised by event type",
+	}, []string{"type"})
+
+	// Leader Election
+	LeaderTransitionsTotal = prometheus.NewCounterVec(
+		prometheus.CounterOpts{Name: "kube_vip_leader_election_transitions_total", Help: "Frequent transitions mean instability"},
+		[]string{"lease_name"},
+	)
+	IsLeader = prometheus.NewGaugeVec(
+		prometheus.GaugeOpts{Name: "kube_vip_is_leader", Help: "1 if this node currently holds the lease"},
+		[]string{"node", "lease_name"},
+	)
+
+	// This is a prometheus gauge indicating the state of the sessions.
+	// 1 means "ESTABLISHED", 0 means "NOT ESTABLISHED"
+	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"},
+	)
+
+	// General Health
+	BuildInfo = prometheus.NewGaugeVec(
+		prometheus.GaugeOpts{Name: "kube_vip_build_info", Help: "Constant 1; track version skew across nodes"},
+		[]string{"version", "build", "node"},
+	)
+)
+
+func RegisterPrometheusMetrics() {
+	// Register all metrics with Prometheus
+	prometheus.MustRegister(
+		ActiveServices,
+		ServiceReconcileErrorsTotal,
+		ServiceReconcileDuration,
+		LeaderTransitionsTotal,
+		IsLeader,
+		BGPSessionInfoGauge,
+		BuildInfo,
+		CountServiceWatchEvent,
+	)
+}

pkg/networkinterface/networkinterface.go

@@ -1,6 +1,7 @@
 package networkinterface
 
 import (
+	log "log/slog"
 	"sync"
 
 	"github.com/vishvananda/netlink"
@@ -23,6 +24,12 @@ func NewManager() *Manager {
 
 func (m *Manager) Get(intf netlink.Link) *Link {
 	if l, ok := m.interfaces[intf.Attrs().Name]; ok {
+		updated, err := netlink.LinkByName(l.Intf.Attrs().Name)
+		if err != nil {
+			log.Error("failed to get interface %q: %w", l.Intf.Attrs().Name, err)
+			return nil
+		}
+		l.Intf = updated
 		return l
 	}
 	result := &Link{

pkg/node/labeler/label_manager.go

@@ -2,29 +2,20 @@ package labeler
 
 import (
 	"context"
-	"encoding/hex"
 	"encoding/json"
 	"fmt"
-	"net"
 	"strings"
 	"time"
 
 	log "log/slog"
 
-	"github.com/kube-vip/kube-vip/pkg/instance"
+	"github.com/kube-vip/kube-vip/pkg/kubevip"
 	"github.com/pkg/errors"
-	corev1 "k8s.io/api/core/v1"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/types"
 	"k8s.io/client-go/kubernetes"
 )
 
-const (
-	// labelName is the name of the label that will be added to the node
-	// it is prefix for the label key before "/"
-	labelName = "service-provided.kube-vip.io"
-)
-
 // labelOperation is the operation to perform on the node labels
 type labelOperation string
 
@@ -52,23 +43,22 @@ type Manager struct {
 }
 
 // AddLabel a new label to the node
-func (m *Manager) AddLabel(ctx context.Context, svc *corev1.Service) error {
-	log.Debug("[service] add label to node", "namespace", svc.Namespace, "name", svc.Name)
-	labelKey, labelValue := generateNodeLabelKeyValue(svc)
-	return m.patchNode(ctx, labelOperationAdd, map[string]string{labelKey: labelValue})
+func (m *Manager) AddLabel(labels map[string]string) error {
+	log.Debug("add labels to node", "node", m.nodeName)
+	return m.patchNode(labelOperationAdd, labels)
 }
 
 // RemoveLabel a label from the node
-func (m *Manager) RemoveLabel(ctx context.Context, svc *corev1.Service) error {
-	log.Debug("[service] delete label from node", "namespace", svc.Namespace, "name", svc.Name)
-	labelKey, _ := generateNodeLabelKeyValue(svc)
-	return m.patchNode(ctx, labelOperationRemove, map[string]string{labelKey: ""})
+func (m *Manager) RemoveLabel(labels map[string]string) error {
+	log.Debug("delete label from node")
+	return m.patchNode(labelOperationRemove, labels)
 }
 
-// clean up the node labels
+// CleanUpLabels purges the node labels
 func (m *Manager) CleanUpLabels(timeout time.Duration) error {
 	log.Debug("cleaning up labels for node", "node", m.nodeName, "timeout", timeout)
 
+	// create new context for labels cleanup (independent)
 	ctx, cancel := context.WithTimeout(context.Background(), timeout)
 	defer cancel()
 
@@ -80,8 +70,8 @@ func (m *Manager) CleanUpLabels(timeout time.Duration) error {
 
 	// collect all labels with the prefix to remove
 	labels := map[string]string{}
-	for k := range node.Labels {
-		if strings.HasPrefix(k, labelName) {
+	for _, k := range kubevip.GetKeysForCleanup() {
+		if _, ok := node.Labels[k]; ok {
 			labels[k] = ""
 		}
 	}
@@ -92,38 +82,20 @@ func (m *Manager) CleanUpLabels(timeout time.Duration) error {
 	}
 
 	// patch the node with the labels to remove
-	return m.patchNode(ctx, labelOperationRemove, labels)
-}
-
-// generateNodeLabelKeyValue generates a label key and value for the given service
-func generateNodeLabelKeyValue(svc *corev1.Service) (string, string) {
-	addresses, _ := instance.FetchServiceAddresses(svc)
-
-	sanitized := make([]string, len(addresses))
-	for i, addr := range addresses {
-		sanitized[i] = sanitizeIPForLabel(addr)
-	}
-
-	return fmt.Sprintf("%s/%s.%s", labelName, svc.Name, svc.Namespace), strings.Join(sanitized, ",")
-}
-
-// Helper function to convert IPv6 hex address without colons
-func sanitizeIPForLabel(addr string) string {
-	ip := net.ParseIP(addr)
-	if ip == nil || ip.To4() != nil {
-		return addr
-	}
-	return hex.EncodeToString(ip.To16())
+	return m.patchNode(labelOperationRemove, labels)
 }
 
 // patchNode patches the node with the given labels
-func (m *Manager) patchNode(ctx context.Context, operation labelOperation, labels map[string]string) error {
+func (m *Manager) patchNode(operation labelOperation, labels map[string]string) error {
 	type patchStringLabel struct {
 		Op    string `json:"op"`
 		Path  string `json:"path"`
 		Value string `json:"value"`
 	}
 
+	ctx, cancel := context.WithTimeout(context.Background(), time.Second*30)
+	defer cancel()
+
 	patchLabels := []patchStringLabel{}
 	// generate the patch
 	for k, v := range labels {

pkg/node/noop/label_noop_manager.go

@@ -1,10 +1,7 @@
 package noop
 
 import (
-	"context"
 	"time"
-
-	corev1 "k8s.io/api/core/v1"
 )
 
 // NewManager creates a new NoOp label manager
@@ -14,11 +11,11 @@ func NewManager() *Manager {
 
 type Manager struct{}
 
-func (m *Manager) AddLabel(_ context.Context, _ *corev1.Service) error {
+func (m *Manager) AddLabel(labels map[string]string) error {
 	return nil
 }
 
-func (m *Manager) RemoveLabel(_ context.Context, _ *corev1.Service) error {
+func (m *Manager) RemoveLabel(labels map[string]string) error {
 	return nil
 }
 

pkg/node/types.go

@@ -1,7 +1,6 @@
 package node
 
 import (
-	"context"
 	"time"
 
 	log "log/slog"
@@ -9,35 +8,36 @@ import (
 	"github.com/kube-vip/kube-vip/pkg/kubevip"
 	"github.com/kube-vip/kube-vip/pkg/node/labeler"
 	"github.com/kube-vip/kube-vip/pkg/node/noop"
-	v1 "k8s.io/api/core/v1"
 	"k8s.io/client-go/kubernetes"
 )
 
 // LabelManager is the interface for the node label manager
 type LabelManager interface {
-	// AddLabel adds a label to the node for the given service
-	AddLabel(ctx context.Context, svc *v1.Service) error
+	Labeler
+	LabelCleaner
+}
 
-	// RemoveLabel removes the label from the node for the given service
-	RemoveLabel(ctx context.Context, svc *v1.Service) error
+type Labeler interface {
+	// AddLabel adds a label to the node
+	AddLabel(labels map[string]string) error
 
+	// RemoveLabel removes the label from the node
+	RemoveLabel(labels map[string]string) error
+}
+
+type LabelCleaner interface {
 	// CleanUpLabels removes all labels from the node
 	CleanUpLabels(timeout time.Duration) error
 }
 
 // NewManager creates a new Label Manager for the given node
-// NoOp implementation is returned if node labeling is disabled, or if
-// running in control plane mode, or if the client is not ready
+// NoOp implementation is returned if node labeling is disabled,
+// or if the client is not ready
 func NewManager(config *kubevip.Config, clientSet *kubernetes.Clientset) LabelManager {
 	if !config.EnableNodeLabeling {
 		return noop.NewManager()
 	}
 
-	if config.EnableControlPlane {
-		log.Debug("Skip node labeling, control plane mode enabled")
-		return noop.NewManager()
-	}
-
 	if clientSet == nil {
 		log.Debug("Skip node labeling, client is not ready")
 		return noop.NewManager()

pkg/route/manager.go

@@ -0,0 +1,126 @@
+package route
+
+import (
+	"errors"
+	"fmt"
+	log "log/slog"
+	"sync"
+	"syscall"
+)
+
+type Manager struct {
+	tracker map[string]*item
+	mtx     sync.Mutex
+}
+
+func NewManager() *Manager {
+	return &Manager{
+		tracker: make(map[string]*item),
+	}
+}
+
+type item struct {
+	objects map[string]bool
+	route   route
+}
+
+func newItem(r route) *item {
+	return &item{
+		objects: make(map[string]bool),
+		route:   r,
+	}
+}
+
+type route interface {
+	AddRoute(bool) (bool, error)
+	UpdateRoutes() (bool, error)
+	DeleteRoute() error
+	RouteHash() string
+	Interface() string
+}
+
+// Add will add route
+func (m *Manager) Add(object string, r route, precheck, update bool) error {
+	m.mtx.Lock()
+	defer m.mtx.Unlock()
+
+	key := r.RouteHash()
+
+	itm, exists := m.tracker[key]
+
+	if !exists {
+		m.tracker[key] = newItem(r)
+		itm = m.tracker[key]
+
+		added, err := r.AddRoute(precheck)
+		if err != nil {
+			if update && errors.Is(err, syscall.EEXIST) && update {
+				// If route exists, but protocol is not set (e.g. the route was created by the older version
+				// of kube-vip) try to update it if necessary
+				isUpdated, err := r.UpdateRoutes()
+				if err != nil {
+					return fmt.Errorf("error updating existing route: %w", err)
+				}
+				if isUpdated {
+					log.Debug("[RT] updated route", "path", key, "object", object, "interface", r.Interface())
+				}
+			} else {
+				// If other error occurs, return error
+				return fmt.Errorf("error adding route %q: %w", key, err)
+			}
+		}
+
+		if added {
+			log.Debug("[RT] added route", "path", key, "object", object)
+		}
+	}
+
+	itm.objects[object] = true
+
+	log.Debug("[RT] incremented route", "path", key, "object", object, "cnt", len(itm.objects))
+
+	return nil
+}
+
+// Delete will delete route
+func (m *Manager) Delete(object string, r route) error {
+	m.mtx.Lock()
+	defer m.mtx.Unlock()
+
+	key := r.RouteHash()
+
+	itm, exists := m.tracker[key]
+	if !exists {
+		log.Debug("[RT] deleting route - nothing to delete", "path", key, "object", object)
+		return nil
+	}
+
+	delete(itm.objects, object)
+
+	log.Debug("[RT] decremented route", "path", key, "object", object, "cnt", len(itm.objects))
+
+	if len(itm.objects) == 0 {
+		if err := r.DeleteRoute(); err != nil {
+			itm.objects[object] = true
+			return fmt.Errorf("failed to delete route: %w", err)
+		}
+		delete(m.tracker, key)
+		log.Debug("[RT] deleted route", "path", key, "object", object)
+	}
+
+	return nil
+}
+
+func (m *Manager) Clear() {
+	for _, itm := range m.tracker {
+		if err := itm.route.DeleteRoute(); err != nil {
+			log.Warn("[RT] failed to delete route", "err", err.Error())
+		}
+	}
+	m.tracker = make(map[string]*item)
+}
+
+func (m *Manager) Check(key string) bool {
+	_, exists := m.tracker[key]
+	return exists
+}

pkg/route/manager_test.go

@@ -0,0 +1,215 @@
+package route
+
+import (
+	"fmt"
+	"testing"
+)
+
+func Test_SingleAddDel(t *testing.T) {
+	t.Run("Add and delete single route", func(_ *testing.T) {
+		m := NewManager()
+		if m == nil { // nolint
+			t.Error("failed to create manager")
+		}
+
+		r := &mockRoute{hash: "123456"}
+
+		name := "test-service"
+
+		if err := m.Add(name, r, false, false); err != nil {
+			t.Errorf("add failed for route %q, object %q", r.RouteHash(), name)
+		}
+
+		expected := 1
+		if len(m.tracker) != expected { //nolint
+			t.Errorf("number of items in tracker should be %d, but is %d", expected, len(m.tracker))
+		}
+
+		item, exists := m.tracker[r.RouteHash()]
+		if !exists {
+			t.Errorf("object %q was not added to tracker for service %q", r.RouteHash(), name)
+		}
+
+		if len(item.objects) != 1 {
+			t.Errorf("number of objects for %q was %d instead of 1", name, len(item.objects))
+		}
+
+		if err := m.Delete(name, r); err != nil {
+			t.Errorf("deletion of object %q failed for service %q", r.RouteHash(), name)
+		}
+
+		expected = 0
+		if len(m.tracker) != expected {
+			t.Errorf("number of items in tracker should be %d, but is %d", expected, len(m.tracker))
+		}
+	})
+}
+
+func Test_MultipleAddDel(t *testing.T) {
+	t.Run("Add and delete same route multiple times", func(_ *testing.T) {
+		m := NewManager()
+		if m == nil {
+			t.Error("failed to create manager")
+		}
+
+		r := &mockRoute{hash: "123456"}
+
+		numOfServices := 50
+		expectedRoutes := 1
+
+		for i := range numOfServices {
+			name := fmt.Sprintf("test-service-%d", i)
+
+			if err := m.Add(name, r, false, false); err != nil {
+				t.Errorf("add failed for route %q, object %q", r.RouteHash(), name)
+			}
+
+			if len(m.tracker) != expectedRoutes {
+				t.Errorf("number of routes in tracker should be %d, but is %d", expectedRoutes, len(m.tracker))
+			}
+
+			item, exists := m.tracker[r.RouteHash()]
+			if !exists {
+				t.Errorf("route %q was not added to the tracker", r.RouteHash())
+			}
+
+			expected := i + 1
+			if len(item.objects) != expected {
+				t.Errorf("number of objects for %q was %d instead of 1", name, len(item.objects))
+			}
+		}
+
+		for i := range numOfServices {
+			name := fmt.Sprintf("test-service-%d", i)
+
+			if err := m.Delete(name, r); err != nil {
+				t.Errorf("delete failed for route %q, object %q", r.RouteHash(), name)
+			}
+
+			if i == numOfServices-1 {
+				expectedRoutes = 0
+			}
+
+			if len(m.tracker) != expectedRoutes {
+				t.Errorf("number of routes in tracker should be %d, but is %d", expectedRoutes, len(m.tracker))
+			}
+
+			exists := m.Check(r.RouteHash())
+			if i < numOfServices-1 {
+				if !exists {
+					t.Errorf("route %q was not added to the tracker", r.RouteHash())
+				}
+			} else {
+				if exists {
+					t.Errorf("route %q was not deleted from the tracker", r.RouteHash())
+				}
+			}
+		}
+	})
+}
+
+func Test_MultipleRoutesAddDel(t *testing.T) {
+	t.Run("Add and delete multiple routes multiple times", func(_ *testing.T) {
+		m := NewManager()
+		if m == nil {
+			t.Error("failed to create manager")
+		}
+
+		routes := []*mockRoute{}
+
+		numOfRoutes := 20
+		hashBase := 123456
+
+		for i := range numOfRoutes {
+			routes = append(routes, &mockRoute{hash: fmt.Sprintf("%d%d", hashBase, i)})
+		}
+
+		numOfServices := 50
+
+		for rtCnt, r := range routes {
+			for i := range numOfServices {
+
+				expectedRoutes := rtCnt + 1
+				name := fmt.Sprintf("test-service-%d", i)
+
+				if err := m.Add(name, r, false, false); err != nil {
+					t.Errorf("add failed for route %q, object %q", r.RouteHash(), name)
+				}
+
+				if len(m.tracker) != expectedRoutes {
+					t.Errorf("number of routes in tracker should be %d, but is %d", expectedRoutes, len(m.tracker))
+				}
+
+				item, exists := m.tracker[r.RouteHash()]
+				if !exists {
+					t.Errorf("route %q was not added to the tracker", r.RouteHash())
+				}
+
+				expected := i + 1
+				if len(item.objects) != expected {
+					t.Errorf("number of objects for %q was %d instead of 1", name, len(item.objects))
+				}
+			}
+		}
+
+		for rtCnt, r := range routes {
+			for i := range numOfServices {
+				name := fmt.Sprintf("test-service-%d", i)
+
+				if err := m.Delete(name, r); err != nil {
+					t.Errorf("delete failed for route %q, object %q", r.RouteHash(), name)
+				}
+
+				expectedRoutes := numOfRoutes - rtCnt
+				if i == numOfServices-1 {
+					expectedRoutes -= 1
+				}
+
+				if len(m.tracker) != expectedRoutes {
+					t.Errorf("number of routes in tracker should be %d, but is %d", expectedRoutes, len(m.tracker))
+				}
+
+				exists := m.Check(r.RouteHash())
+				if i < numOfServices-1 {
+					if !exists {
+						t.Errorf("route %q should not be deleted", r.RouteHash())
+					}
+				} else {
+					if exists {
+						t.Errorf("route %q was not deleted from the tracker", r.RouteHash())
+					}
+				}
+			}
+		}
+	})
+}
+
+type mockRoute struct {
+	added     bool
+	addErr    error
+	updated   bool
+	updateErr error
+	delErr    error
+	hash      string
+	intf      string
+}
+
+func (mr *mockRoute) AddRoute(_ bool) (bool, error) {
+	return mr.added, mr.addErr
+}
+
+func (mr *mockRoute) UpdateRoutes() (bool, error) {
+	return mr.updated, mr.updateErr
+}
+
+func (mr *mockRoute) DeleteRoute() error {
+	return mr.delErr
+}
+
+func (mr *mockRoute) RouteHash() string {
+	return mr.hash
+}
+
+func (mr *mockRoute) Interface() string {
+	return mr.intf
+}

pkg/servicecontext/servicecontext.go

@@ -3,21 +3,27 @@ package servicecontext
 import (
 	"context"
 	"sync"
+	"sync/atomic"
 )
 
 type Context struct {
 	Ctx                context.Context
 	Cancel             context.CancelFunc
-	IsActive           bool
 	IsWatched          bool
 	ConfiguredNetworks sync.Map
+	EndpointsReady     chan any
+	epReady            sync.Once
+	Signalled          atomic.Bool
+	LeaderCancel       context.CancelFunc
 }
 
 func New(ctx context.Context) *Context {
-	svcCtx, svcCancel := context.WithCancel(ctx)
+	// context and cancel stored for a future use, gosec linter disabled
+	svcCtx, svcCancel := context.WithCancel(ctx) //nolint:gosec
 	return &Context{
-		Ctx:    svcCtx,
-		Cancel: svcCancel,
+		Ctx:            svcCtx,
+		Cancel:         svcCancel,
+		EndpointsReady: make(chan any),
 	}
 }
 
@@ -34,3 +40,18 @@ func (ctx *Context) IsNetworkConfigured(ip string) bool {
 	_, exists := ctx.ConfiguredNetworks.Load(ip)
 	return exists
 }
+
+func (ctx *Context) SignalReadiness() {
+	ctx.epReady.Do(func() {
+		close(ctx.EndpointsReady)
+		ctx.Signalled.Store(true)
+	})
+}
+
+func (ctx *Context) ResetReadiness() {
+	if ctx.Signalled.Load() {
+		ctx.EndpointsReady = make(chan any)
+		ctx.epReady = sync.Once{}
+		ctx.Signalled.Store(false)
+	}
+}

pkg/services/callback.go

@@ -0,0 +1,24 @@
+package services
+
+import (
+	"sync"
+
+	"github.com/kube-vip/kube-vip/pkg/servicecontext"
+	v1 "k8s.io/api/core/v1"
+)
+
+type Callback struct {
+	Function           func(*servicecontext.Context, *v1.Service, *sync.WaitGroup, bool) error
+	UsesLeaderElection bool
+}
+
+func NewCallback(f func(*servicecontext.Context, *v1.Service, *sync.WaitGroup, bool) error, leaderElection bool) *Callback {
+	return &Callback{
+		Function:           f,
+		UsesLeaderElection: leaderElection,
+	}
+}
+
+func (c *Callback) Run(svcCtx *servicecontext.Context, svc *v1.Service, wg *sync.WaitGroup) error {
+	return c.Function(svcCtx, svc, wg, c.UsesLeaderElection)
+}

pkg/services/egress.go

@@ -135,7 +135,7 @@ func checkCIDR(ip, cidr string) (string, error) {
 	return "", nil
 }
 
-func (p *Processor) configureEgress(vipIP, podIP, namespace, serviceUUID string, annotations map[string]string) error {
+func (p *Processor) configureEgress(ctx context.Context, vipIP, podIP, namespace, serviceUUID string, annotations map[string]string) error {
 	var podCidr, serviceCidr string
 	var autoServiceCIDR, autoPodCIDR string
 	var discoverErr error
@@ -147,7 +147,7 @@ func (p *Processor) configureEgress(vipIP, podIP, namespace, serviceUUID string,
 	internalEgress := annotations[kubevip.EgressInternal]
 
 	if p.config.EgressPodCidr == "" || p.config.EgressServiceCidr == "" {
-		autoServiceCIDR, autoPodCIDR, discoverErr = p.AutoDiscoverCIDRs()
+		autoServiceCIDR, autoPodCIDR, discoverErr = p.AutoDiscoverCIDRs(ctx)
 	}
 
 	if discoverErr != nil {
@@ -224,8 +224,15 @@ func (p *Processor) configureEgress(vipIP, podIP, namespace, serviceUUID string,
 
 		}
 
+		// Add networks that we specifically should only SNAT for
+		allowCIDRs := []string{}
+		if allowedNetworks != "" {
+			networks := strings.Split(strings.TrimSpace(allowedNetworks), ",") //Remove whitespace characters and then create an array from the CIDRs
+			allowCIDRs = append(allowCIDRs, networks...)
+		}
+
 		// Apply the SNAT rules
-		err := nftables.ApplySNAT(podIP, vipIP, serviceUUID, destinationPorts, ignoreCIDRs, utils.IsIPv6(vipIP))
+		err := nftables.ApplySNAT(podIP, vipIP, serviceUUID, destinationPorts, ignoreCIDRs, allowCIDRs, utils.IsIPv6(vipIP))
 		if err != nil {
 			return fmt.Errorf("error performing netlink nftables [%s]", err)
 		}
@@ -269,9 +276,9 @@ func (p *Processor) configureEgress(vipIP, podIP, namespace, serviceUUID string,
 		}
 	}
 
-	mask := "/32"
+	mask := fmt.Sprintf("/%d", vip.DefaultMaskIPv4)
 	if !utils.IsIPv4(podIP) {
-		mask = "/128"
+		mask = fmt.Sprintf("/%d", vip.DefaultMaskIPv6)
 	}
 
 	if allowedNetworks != "" {
@@ -331,12 +338,12 @@ func (p *Processor) configureEgress(vipIP, podIP, namespace, serviceUUID string,
 	return nil
 }
 
-func (p *Processor) AutoDiscoverCIDRs() (serviceCIDR, podCIDR string, err error) {
+func (p *Processor) AutoDiscoverCIDRs(ctx context.Context) (serviceCIDR, podCIDR string, err error) {
 	log.Debug("Trying to automatically discover Service and Pod CIDRs")
 	options := v1.ListOptions{
 		LabelSelector: "component=kube-controller-manager",
 	}
-	podList, err := p.clientSet.CoreV1().Pods("kube-system").List(context.TODO(), options)
+	podList, err := p.clientSet.CoreV1().Pods("kube-system").List(ctx, options)
 	if err != nil {
 		return "", "", fmt.Errorf("[Egress] Unable to get kube-controller-manager pod: %w", err)
 	}

pkg/services/leader.go

@@ -3,33 +3,26 @@ package services
 import (
 	"context"
 	"fmt"
-	"time"
+	"sync"
 
 	log "log/slog"
 
+	"github.com/kube-vip/kube-vip/pkg/election"
 	"github.com/kube-vip/kube-vip/pkg/lease"
+	"github.com/kube-vip/kube-vip/pkg/metrics"
+	"github.com/kube-vip/kube-vip/pkg/servicecontext"
 	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 (p *Processor) StartServicesWatchForLeaderElection(ctx context.Context) error {
-	err := p.ServicesWatcher(ctx, p.StartServicesLeaderElection)
+	err := p.ServicesWatcher(ctx, NewCallback(p.StartServicesLeaderElection, true))
 	if err != nil {
 		return err
 	}
 
 	if p.config.EnableRoutingTable {
-		for _, instance := range p.ServiceInstances {
-			for _, cluster := range instance.Clusters {
-				for i := range cluster.Network {
-					_ = cluster.Network[i].DeleteRoute()
-				}
-				cluster.Stop()
-			}
-		}
+		p.routeMgr.Clear()
 	}
 
 	log.Info("Shutting down kube-Vip")
@@ -38,114 +31,159 @@ func (p *Processor) StartServicesWatchForLeaderElection(ctx context.Context) err
 }
 
 // The startServicesWatchForLeaderElection function will start a services watcher, the
-func (p *Processor) StartServicesLeaderElection(ctx context.Context, service *v1.Service) error {
-	serviceLease, _ := lease.GetName(service)
-	log.Info("new leader election", "service", service.Name, "namespace", service.Namespace, "lock_name", serviceLease, "host_id", p.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: p.clientSet.CoordinationV1(),
-		LockConfig: resourcelock.ResourceLockConfig{
-			Identity: p.config.NodeName,
-		},
+func (p *Processor) StartServicesLeaderElection(svcCtx *servicecontext.Context, service *v1.Service, _ *sync.WaitGroup, _ bool) error {
+	if svcCtx == nil {
+		return fmt.Errorf("no context context for service %q with UID %q: nil context", service.Name, service.UID)
 	}
 
-	go func() {
-		// wait for the service context to end and delete the lease then
-		<-ctx.Done()
-		p.leaseMgr.Delete(service)
+	leaseNamespace, serviceLease := lease.ServiceName(service)
+	id := lease.NewID(p.config.LeaderElectionType, leaseNamespace, serviceLease)
+	objectName := lease.ServiceNamespacedName(service)
+
+	svcLease := p.leaseMgr.Get(id)
+	if svcLease == nil {
+		return fmt.Errorf("no existing lease found for service %q with UID %q", service.Name, service.UID)
+	}
+
+	isNew := svcLease.Add(objectName)
+
+	svcLease.Lock()
+
+	defer func() {
+		svcLease.Unlock()
 	}()
 
-	svcCtx, err := p.getServiceContext(service.UID)
-	if err != nil {
-		return fmt.Errorf("failed to get context for service %q with UID %q: %w", service.Name, service.UID, err)
-	}
-	if svcCtx == nil {
-		return fmt.Errorf("failed to get context for service %q with UID %q: nil context", service.Name, service.UID)
+	// this service was already processed so we do not need to do anything
+	if !isNew && svcLease.Elected.Load() {
+		svcLease.Unlock()
+		log.Debug("this service was already handled, waiting for it to finish", "service", service.Name, "uid", service.UID)
+		// Wait for either the service context or lease context to be done
+		select {
+		case <-svcCtx.Ctx.Done():
+		case <-svcLease.Ctx.Done():
+		}
+		return nil
 	}
 
-	svcCtx.IsActive = true
+	wg := sync.WaitGroup{}
+	defer wg.Wait()
 
-	svcLease, isNew := p.leaseMgr.Add(service)
-	// this service is sharing lease
-	if !isNew {
+	// 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() {
+		<-svcCtx.Ctx.Done()
+		p.leaseMgr.Delete(id, objectName)
+	})
+
+	select {
+	case <-svcCtx.Ctx.Done():
+		return fmt.Errorf("service context cancelled before election start: %w", svcCtx.Ctx.Err())
+	case <-svcLease.Ctx.Done():
+		return fmt.Errorf("lease context cancelled before election start: %w", svcLease.Ctx.Err())
+	case <-svcCtx.EndpointsReady:
+	}
+
+	// this service is sharing lease with another service
+	if svcLease.Elected.Load() {
+		svcLease.Unlock()
 		// wait for leader election to start or context to be done
 		select {
 		case <-svcLease.Started:
 		case <-svcLease.Ctx.Done():
-			svcCtx.IsActive = false
+			// 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", "service", service.Name, "uid", service.UID)
 			return nil
 		}
-		<-svcLease.Started
 
-		if lease.UsesCommon(service) {
-			if err := p.SyncServices(ctx, service); err != nil {
-				log.Error("service sync", "err", err)
-				svcLease.Cancel()
-			}
-			// just block until context is cancelled
-			<-ctx.Done()
-			if svcCtx.IsActive {
-				if err := p.deleteService(service.UID); err != nil {
-					log.Error("service deletion", "err", err)
-				}
-			}
+		if err := p.onStartedLeading(svcCtx, service, &wg); err != nil {
+			log.Error("error on started leading", "error", err)
+		}
+
+		// Block until service context is cancelled
+		<-svcCtx.Ctx.Done()
+
+		if err := p.onStoppedLeading(svcLease, service); err != nil {
+			log.Error("error on stopped leading", "error", err)
 		}
 
 		// wait for leaderelection to be finished
 		<-svcLease.Ctx.Done()
-		// Mark this service is inactive
-		svcCtx.IsActive = false
+
 		return nil
 	}
-	// start the leader election code loop
-	leaderelection.RunOrDie(svcLease.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(p.config.LeaseDuration) * time.Second,
-		RenewDeadline:   time.Duration(p.config.RenewDeadline) * time.Second,
-		RetryPeriod:     time.Duration(p.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
-				if err := p.SyncServices(ctx, service); err != nil {
-					log.Error("service sync", "err", err)
-					svcLease.Cancel()
-				}
-				close(svcLease.Started)
-			},
-			OnStoppedLeading: func() {
-				// we can do cleanup here
-				log.Info("leadership lost", "service", service.Name, "leader", p.config.NodeName)
-				if svcCtx.IsActive {
-					if err := p.deleteService(service.UID); err != nil {
-						log.Error("service deletion", "err", err)
-					}
-				}
-				// Mark this service is inactive
-				svcCtx.IsActive = false
-			},
-			OnNewLeader: func(identity string) {
-				// we're notified when new leader elected
-				if identity == p.config.NodeName {
-					// I just got the lock
-					return
-				}
-				log.Info("new leader", "leader", identity)
-			},
+
+	log.Info("new leader election", "service", service.Name, "namespace", service.Namespace, "lock_name", serviceLease, "host_id", p.config.NodeName)
+
+	leaderCtx, leaderCancel := context.WithCancel(svcLease.Ctx)
+	svcCtx.LeaderCancel = leaderCancel
+
+	run := election.RunConfig{
+		Config:           p.config,
+		LeaseID:          id,
+		Mgr:              p.electionMgr,
+		LeaseAnnotations: map[string]string{},
+
+		OnStartedLeading: func(_ context.Context) {
+			svcLease.Elected.Store(true)
+			svcLease.Unlock()
+			close(svcLease.Started)
+			// Mark this service as active (as we've started leading)
+			// we run this in background as it's blocking
+			if err := p.onStartedLeading(svcCtx, service, &wg); err != nil {
+				leaderCancel()
+			}
+			metrics.LeaderTransitionsTotal.WithLabelValues(id.Name()).Inc()
+			metrics.IsLeader.WithLabelValues(p.config.NodeName, id.Name()).Set(1)
 		},
-	})
-	log.Info("stopping leader election", "service", service.Name)
+		OnStoppedLeading: func() {
+			// we can do cleanup here
+			svcLease.Elected.Store(false)
+			log.Info("leadership lost", "service", service.Name, "uid", service.UID, "leader", p.config.NodeName)
+			if err := p.onStoppedLeading(svcLease, service); err != nil {
+				metrics.ServiceReconcileErrorsTotal.WithLabelValues(service.Namespace, service.Name, "delete_service").Inc()
+				leaderCancel()
+			}
+			metrics.IsLeader.WithLabelValues(p.config.NodeName, id.Name()).Set(0)
+			svcLease.Started = make(chan any)
+		},
+		OnNewLeader: func(identity string) {
+			// we're notified when new leader elected
+			if identity == p.config.NodeName {
+				// I just got the lock
+				return
+			}
+			log.Info("new leader", "leader", identity, "service", service.Name, "uid", service.UID)
+		},
+	}
+
+	if err := election.RunOrDie(leaderCtx, &run, p.config); err != nil {
+		return fmt.Errorf("services election failed: %w", err)
+	}
+
+	log.Info("stopping leader election", "service", service.Name, "uid", service.UID)
+	return nil
+}
+
+func (p *Processor) onStartedLeading(svcCtx *servicecontext.Context, service *v1.Service, wg *sync.WaitGroup) error {
+	// Mark this service as active (as we've started leading)
+	// we run this in background as it's blocking
+	err := p.SyncServices(svcCtx, service, wg, true)
+	if err != nil {
+		log.Error("service sync", "uid", service.UID, "err", err)
+		return err
+	}
+	return nil
+}
+
+func (p *Processor) onStoppedLeading(svcLease *lease.Lease, service *v1.Service) error {
+	log.Debug("deleting service due to lost leadership", "uid", service.UID)
+	err := p.deleteService(svcLease.Ctx, service.UID)
+	if err != nil {
+		log.Error("service deletion", "err", err)
+		return err
+	}
 	return nil
 }

pkg/services/processor.go

@@ -2,24 +2,32 @@ package services
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	log "log/slog"
 	"reflect"
 	"sync"
+	"time"
 
 	"github.com/kube-vip/kube-vip/pkg/arp"
 	"github.com/kube-vip/kube-vip/pkg/bgp"
+	"github.com/kube-vip/kube-vip/pkg/election"
 	"github.com/kube-vip/kube-vip/pkg/endpoints"
 	"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/networkinterface"
+	"github.com/kube-vip/kube-vip/pkg/node"
+	"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/vip"
+	"github.com/kube-vip/kube-vip/pkg/wireguard"
 	"github.com/prometheus/client_golang/prometheus"
-	"github.com/vishvananda/netlink"
 	v1 "k8s.io/api/core/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/types"
 	"k8s.io/apimachinery/pkg/watch"
 	"k8s.io/client-go/kubernetes"
@@ -39,30 +47,28 @@ type Processor struct {
 	clientSet   *kubernetes.Clientset
 	rwClientSet *kubernetes.Clientset
 
-	shutdownChan chan struct{}
-
-	// This is a prometheus counter used to count the number of events received
-	// from the service watcher
-	CountServiceWatchEvent *prometheus.CounterVec
-
 	intfMgr *networkinterface.Manager
 	arpMgr  *arp.Manager
 
 	leaseMgr *lease.Manager
 
 	// nodeLabelManager is the manager for the node labels
-	nodeLabelManager labelManager
+	nodeLabelManager node.Labeler
+
+	electionMgr *election.Manager
+
+	// TunnelMgr manages multiple WireGuard tunnels (one per service VIP)
+	TunnelMgr *wireguard.TunnelManager
+
+	routeMgr *route.Manager
 }
 
 // labelManager is the interface for the node label manager to add/remove labels
-type labelManager interface {
-	AddLabel(ctx context.Context, svc *v1.Service) error
-	RemoveLabel(ctx context.Context, svc *v1.Service) error
-}
 
 func NewServicesProcessor(config *kubevip.Config, bgpServer *bgp.Server,
-	clientSet *kubernetes.Clientset, rwClientSet *kubernetes.Clientset, shutdownChan chan struct{},
-	intfMgr *networkinterface.Manager, arpMgr *arp.Manager, nodeLabelManager labelManager) *Processor {
+	clientSet *kubernetes.Clientset, rwClientSet *kubernetes.Clientset,
+	intfMgr *networkinterface.Manager, arpMgr *arp.Manager, nodeLabelManager node.Labeler,
+	electionMgr *election.Manager, leaseMgr *lease.Manager, routeMgr *route.Manager) *Processor {
 	lbClassFilterFunc := lbClassFilter
 	if config.LoadBalancerClassLegacyHandling {
 		lbClassFilterFunc = lbClassFilterLegacy
@@ -75,75 +81,82 @@ func NewServicesProcessor(config *kubevip.Config, bgpServer *bgp.Server,
 		bgpServer:        bgpServer,
 		clientSet:        clientSet,
 		rwClientSet:      rwClientSet,
-		shutdownChan:     shutdownChan,
-		CountServiceWatchEvent: prometheus.NewCounterVec(prometheus.CounterOpts{
-			Namespace: "kube_vip",
-			Subsystem: "manager",
-			Name:      "all_services_events",
-			Help:      "Count all events fired by the service watcher categorised by event type",
-		}, []string{"type"}),
-
 		intfMgr:          intfMgr,
 		arpMgr:           arpMgr,
-		leaseMgr:         lease.NewManager(),
+		leaseMgr:         leaseMgr,
 		nodeLabelManager: nodeLabelManager,
+		electionMgr:      electionMgr,
+		TunnelMgr:        wireguard.NewTunnelManager(),
+		routeMgr:         routeMgr,
 	}
 }
 
-func (p *Processor) AddOrModify(ctx context.Context, event watch.Event, serviceFunc func(context.Context, *v1.Service) error) (bool, error) {
-	// log.Debugf("Endpoints for service [%s] have been Created or modified", s.service.ServiceName)
+func (p *Processor) AddOrModify(ctx context.Context, event watch.Event, serviceFunc *Callback, wg *sync.WaitGroup) error {
 	svc, ok := event.Object.(*v1.Service)
 	if !ok {
-		return false, fmt.Errorf("unable to parse Kubernetes services from API watcher")
+		return fmt.Errorf("unable to parse Kubernetes services from API watcher")
 	}
 
+	timer := prometheus.NewTimer(metrics.ServiceReconcileDuration.WithLabelValues(svc.Namespace))
+	defer timer.ObserveDuration()
+
 	// We only care about LoadBalancer services
 	if svc.Spec.Type != v1.ServiceTypeLoadBalancer {
-		return true, nil
+		return nil
 	}
 
 	// Check if we ignore this service
 	if svc.Annotations[kubevip.LoadbalancerIgnore] == "true" {
 		log.Info("ignore annotation for kube-vip", "service name", svc.Name)
-		return true, nil
+		return nil
 	}
 
 	// Check the loadBalancer class
 	if p.lbClassFilter(svc, p.config) {
-		return true, nil
+		return nil
 	}
 
 	svcAddresses, svcHostnames := instance.FetchServiceAddresses(svc)
 
 	// We only care about LoadBalancer services that have been allocated an address
-	if len(svcAddresses) <= 0 {
-		return true, nil
+	if len(svcAddresses) <= 0 && len(svcHostnames) <= 0 {
+		s, err := p.waitForAddress(ctx, svc)
+		if err != nil {
+			return fmt.Errorf("failed to get updated LB addresses for service %s/%s: %w", svc.Namespace, svc.Name, err)
+		}
+		svc = s
+	}
+
+	svcInstance := instance.FindServiceInstance(svc, p.ServiceInstances)
+	var err error
+	if svcInstance == nil {
+		svcInstance, err = instance.NewInstance(ctx, svc, p.config, p.intfMgr, p.arpMgr, p.routeMgr, p.nodeLabelManager, wg)
+		if err != nil {
+			metrics.ServiceReconcileErrorsTotal.WithLabelValues(svc.Namespace, svc.Name, "new_instance").Inc()
+			return fmt.Errorf("unable to create instance for service %s/%s", svc.Namespace, svc.Name)
+		}
+		p.ServiceInstances = append(p.ServiceInstances, svcInstance)
+		p.updateActiveServicesMetric()
 	}
 
 	_, usesCommonLease := svc.Annotations[kubevip.ServiceLease]
 	if usesCommonLease && svc.Spec.ExternalTrafficPolicy != v1.ServiceExternalTrafficPolicyTypeCluster {
-		return false, fmt.Errorf("annotation %q cannot be used with service traffic policy other than %q",
-			kubevip.ServiceLease, v1.ServiceExternalTrafficPolicyTypeCluster)
+		metrics.ServiceReconcileErrorsTotal.WithLabelValues(svc.Namespace, svc.Name, "invalid_config").Inc()
+		return fmt.Errorf("annotation %q cannot be used with service traffic policy other than %q, service %s/%s",
+			kubevip.ServiceLease, v1.ServiceExternalTrafficPolicyTypeCluster, svc.Namespace, svc.Name)
 	}
 
 	svcCtx, err := p.getServiceContext(svc.UID)
 	if err != nil {
-		return false, fmt.Errorf("failed to get service context: %w", err)
+		metrics.ServiceReconcileErrorsTotal.WithLabelValues(svc.Namespace, svc.Name, "service_context").Inc()
+		return fmt.Errorf("failed to get service context: %w", err)
 	}
 
 	// The modified event should only be triggered if the service has been modified (i.e. moved somewhere else)
 	if event.Type == watch.Modified {
-		i := instance.FindServiceInstance(svc, p.ServiceInstances)
-		shouldGarbageCollect := true
-		if i != nil {
-			originalServiceAddresses, originalServiceHostnames := instance.FetchServiceAddresses(i.ServiceSnapshot)
-			shouldGarbageCollect =
-				// Service addresses changed
-				!reflect.DeepEqual(originalServiceAddresses, svcAddresses) ||
-					// Service hostnames changed
-					!reflect.DeepEqual(originalServiceHostnames, svcHostnames) ||
-					// ExternalTrafficPolicy changed
-					svc.Spec.ExternalTrafficPolicy != i.ServiceSnapshot.Spec.ExternalTrafficPolicy
+		shouldGarbageCollect := false
+		if svcInstance != nil {
+			shouldGarbageCollect = serviceChanged(svcInstance, svc)
 		}
 		if shouldGarbageCollect {
 			for _, addr := range svcAddresses {
@@ -157,15 +170,13 @@ func (p *Processor) AddOrModify(ctx context.Context, event watch.Event, serviceF
 				}
 			}
 			// This service has been modified, but it was also active.
-			if svcCtx != nil && svcCtx.IsActive {
+			if svcCtx != nil {
 				log.Warn("(svcs) The load balancer has changed, cancelling original load balancer")
 				//Set it to inactive
-				svcCtx.IsActive = false
 				svcCtx.Cancel()
-				log.Warn("(svcs) waiting for load balancer to finish")
-				<-svcCtx.Ctx.Done()
 
-				if err := p.deleteService(svc.UID); err != nil {
+				if err := p.deleteService(ctx, svc.UID); err != nil {
+					metrics.ServiceReconcileErrorsTotal.WithLabelValues(svc.Namespace, svc.Name, "delete_service").Inc()
 					log.Error("(svc) unable to remove", "service", svc.UID)
 				}
 				// in theory this should never fail
@@ -173,163 +184,146 @@ func (p *Processor) AddOrModify(ctx context.Context, event watch.Event, serviceF
 				// Reset the the svcCtx when it was garbage collected
 				// As the next function will create a new context when nil
 				svcCtx = nil
+				p.updateActiveServicesMetric()
 			}
 		}
 	}
 
-	// Architecture walkthrough: (Had to do this as this code path is making my head hurt)
+	ips, hostnames := instance.FetchServiceAddresses(svc)
+	log.Debug("(svcs) has been added/modified with addresses", "service name", svc.Name, "ips", ips, "hostnames", hostnames)
 
-	// Is the service active (bool), if not then process this new service
-	// Does this service use an election per service?
-	//
+	if svcCtx == nil {
+		ns, name := lease.ServiceName(svc)
+		leaseID := lease.NewID(p.config.LeaderElectionType, ns, name)
+		lease := p.leaseMgr.Add(ctx, leaseID)
+		svcCtx = servicecontext.New(lease.Ctx)
+		p.svcMap.Store(svc.UID, svcCtx)
+	}
 
-	if svcCtx == nil || svcCtx != nil && !svcCtx.IsActive {
-		ips, hostnames := instance.FetchServiceAddresses(svc)
-		log.Debug("(svcs) has been added/modified with addresses", "service name", svc.Name, "ips", ips, "hostnames", hostnames)
+	// this goroutine starts service handling function (with or without leaderelection)
+	if !svcCtx.IsWatched {
+		wg.Go(func() {
+			watchWg := sync.WaitGroup{}
+			defer func() {
+				// wait for the sub-goroutines and tag service as not watched
+				watchWg.Wait()
+				svcCtx.IsWatched = false
+			}()
 
-		if svcCtx == nil {
-			svcCtx = servicecontext.New(ctx)
-			p.svcMap.Store(svc.UID, svcCtx)
-		}
-
-		if p.config.EnableServicesElection || // Service Election
-			((p.config.EnableRoutingTable || p.config.EnableBGP) && // Routing table mode or BGP
-				(!p.config.EnableLeaderElection && !p.config.EnableServicesElection)) { // No leaderelection or services election
-
-			// If this load balancer Traffic Policy is "local"
-			if svc.Spec.ExternalTrafficPolicy == v1.ServiceExternalTrafficPolicyTypeLocal {
-
-				// Start an endpoint watcher if we're not watching it already
-				if !svcCtx.IsWatched {
-					// background the endpoint watcher
-					if (p.config.EnableRoutingTable || p.config.EnableBGP) && (!p.config.EnableLeaderElection && !p.config.EnableServicesElection) {
-						err = serviceFunc(svcCtx.Ctx, svc)
-						if err != nil {
-							log.Error(err.Error())
-						}
-					}
-
-					go func() {
-						if svc.Spec.ExternalTrafficPolicy == v1.ServiceExternalTrafficPolicyTypeLocal {
-							// Add Endpoint or EndpointSlices watcher
-							var provider providers.Provider
-							if p.config.EnableEndpoints {
-								provider = providers.NewEndpoints()
-							} else {
-								provider = providers.NewEndpointslices()
-							}
-							if err = p.watchEndpoint(svcCtx, p.config.NodeName, svc, provider); err != nil {
-								log.Error(err.Error())
-							}
-						}
-					}()
-
-					// We're now watching this service
-					svcCtx.IsWatched = true
-				}
-			} else if (p.config.EnableBGP || p.config.EnableRoutingTable) && (!p.config.EnableLeaderElection && !p.config.EnableServicesElection) {
-				err = serviceFunc(svcCtx.Ctx, svc)
+			watchWg.Go(func() {
+				// start if service is not already watched/handled
+				// signal endpoints goroutine we are ready to start and run service handling function
+				log.Info("(svcs) service function starting", "uid", svc.UID)
+				err = serviceFunc.Run(svcCtx, svc, wg)
 				if err != nil {
 					log.Error(err.Error())
+					if errors.Is(err, &utils.PanicError{}) {
+						// cancel service context on panic error
+						// TODO:  should we quit kube-vip altogether here?
+						svcCtx.Cancel()
+					}
 				}
+				log.Info("(svcs) service function done", "uid", svc.UID)
+			})
 
-				go func() {
-					if svc.Spec.ExternalTrafficPolicy == v1.ServiceExternalTrafficPolicyTypeCluster {
-						// Add Endpoint watcher
-						var provider providers.Provider
-						if p.config.EnableEndpoints {
-							provider = providers.NewEndpoints()
-						} else {
-							provider = providers.NewEndpointslices()
-						}
-						if err = p.watchEndpoint(svcCtx, p.config.NodeName, svc, provider); err != nil {
-							log.Error(err.Error())
-						}
-					}
-				}()
-				// We're now watching this service
-				svcCtx.IsWatched = true
-			} else {
+			// this goroutine will watch endpoints for the service
+			watchWg.Go(func() {
+				// create provider and start watching the endpoints
+				var provider providers.Provider
+				if p.config.EnableEndpoints {
+					provider = providers.NewEndpoints()
+				} else {
+					provider = providers.NewEndpointslices()
+				}
+				if err = p.watchEndpoint(svcCtx, p.config.NodeName, svc, provider); err != nil {
+					log.Error(err.Error())
+				}
+			})
 
-				go func() {
-					for {
-						select {
-						case <-svcCtx.Ctx.Done():
-							log.Warn("(svcs) restartable service watcher ending", "uid", svc.UID)
-							return
-						default:
-							log.Info("(svcs) restartable service watcher starting", "uid", svc.UID)
-							err = serviceFunc(svcCtx.Ctx, svc)
+		})
 
-							if err != nil {
-								log.Error(err.Error())
-							}
-						}
-					}
-
-				}()
-			}
-		} else {
-			// Increment the waitGroup before the service Func is called (Done is completed in there)
-			err = serviceFunc(svcCtx.Ctx, svc)
-			if err != nil {
-				log.Error(err.Error())
-			}
-		}
-		svcCtx.IsActive = true
+		// tag service as watched
+		svcCtx.IsWatched = true
 	}
 
-	return false, nil
+	if !p.config.EnableServicesElection {
+		log.Debug("Service now active", "name", svc.Name, "uid", svc.UID)
+	}
+
+	return nil
 }
 
-func (p *Processor) Delete(event watch.Event) (bool, error) {
+func (p *Processor) waitForAddress(ctx context.Context, svc *v1.Service) (*v1.Service, error) {
+	addressCtx, cancel := context.WithTimeout(ctx, 30*time.Second)
+	defer cancel()
+	ticker := time.NewTicker(time.Second)
+
+	for {
+		select {
+		case <-addressCtx.Done():
+			return nil, fmt.Errorf("failed to wait for the service LB address: %w", ctx.Err())
+		case <-ticker.C:
+			s, err := p.clientSet.CoreV1().Services(svc.Namespace).Get(addressCtx, svc.Name, metav1.GetOptions{})
+			if err != nil {
+				return nil, fmt.Errorf("failed to get updated service data: %w", err)
+			}
+			addrs, hostnames := instance.FetchServiceAddresses(s)
+			if len(addrs) > 0 || len(hostnames) > 0 {
+				return s, nil
+			}
+		}
+	}
+}
+
+func (p *Processor) Delete(event watch.Event) error {
 	svc, ok := event.Object.(*v1.Service)
 	if !ok {
-		return false, fmt.Errorf("(svcs) unable to parse Kubernetes services from API watcher")
+		return fmt.Errorf("(svcs) unable to parse Kubernetes services from API watcher")
 	}
+
 	svcCtx, err := p.getServiceContext(svc.UID)
 	if err != nil {
-		return false, fmt.Errorf("(svcs) unable to get context: %w", err)
+		return fmt.Errorf("(svcs) unable to get context: %w", err)
 	}
 
 	if svcCtx != nil {
 		// We only care about LoadBalancer services
 		if svc.Spec.Type != v1.ServiceTypeLoadBalancer {
-			return true, nil
+			return nil
 		}
 
 		// We can ignore this service
 		if svc.Annotations[kubevip.LoadbalancerIgnore] == "true" {
 			log.Info("(svcs) ignore annotation for kube-vip", "service name", svc.Name)
-			return true, nil
+			return nil
 		}
 
 		// If no leader election is enabled, delete routes here
 		if !p.config.EnableLeaderElection && !p.config.EnableServicesElection &&
 			p.config.EnableRoutingTable && svcCtx.HasConfiguredNetworks() {
-			if errs := endpoints.ClearRoutes(svc, &p.ServiceInstances); len(errs) == 0 {
+			if errs := endpoints.ClearRoutes(svc, &p.ServiceInstances, p.routeMgr); len(errs) == 0 {
 				svcCtx.ConfiguredNetworks.Clear()
 			}
 		}
 
-		// If this is an active service then and additional leaderElection will handle stopping
-		err = p.deleteService(svc.UID)
-		if err != nil {
-			log.Error(err.Error())
+		if !p.config.EnableServicesElection {
+			// If this is an active service then and additional leaderElection will handle stopping
+			err = p.deleteService(svcCtx.Ctx, svc.UID)
+			if err != nil {
+				log.Error(err.Error())
+			}
 		}
 
 		// Calls the cancel function of the context
-		log.Warn("(svcs) The load balancer was deleted, cancelling context")
-		svcCtx.IsActive = false
+		log.Warn("(svcs) The load balancer was deleted, cancelling context", "namespace", svc.Namespace, "name", svc.Name, "uid", svc.UID)
 		svcCtx.Cancel()
-		log.Warn("(svcs) waiting for load balancer to finish")
-		<-svcCtx.Ctx.Done()
 		p.svcMap.Delete(svc.UID)
+		p.updateActiveServicesMetric()
 	}
 
 	log.Info("(svcs) deleted", "service name", svc.Name, "namespace", svc.Namespace)
 
-	return true, nil
+	return nil
 }
 
 func (p *Processor) Stop() {
@@ -352,6 +346,32 @@ func (p *Processor) getServiceContext(uid types.UID) (*servicecontext.Context, e
 	return ctx, nil
 }
 
-func (p *Processor) CountRouteReferences(route *netlink.Route) int {
-	return endpoints.CountRouteReferences(route, &p.ServiceInstances)
+func serviceChanged(i *instance.Instance, svc *v1.Service) bool {
+	svcAddresses, svcHostnames := instance.FetchServiceAddresses(svc)
+	originalServiceAddresses, originalServiceHostnames := instance.FetchServiceAddresses(i.ServiceSnapshot)
+
+	// Service addresses changed
+	return !reflect.DeepEqual(originalServiceAddresses, svcAddresses) ||
+		// Service hostnames changed
+		!reflect.DeepEqual(originalServiceHostnames, svcHostnames) ||
+		// ExternalTrafficPolicy changed
+		svc.Spec.ExternalTrafficPolicy != i.ServiceSnapshot.Spec.ExternalTrafficPolicy ||
+		// IP stack configuration changed
+		!reflect.DeepEqual(svc.Spec.IPFamilies, i.ServiceSnapshot.Spec.IPFamilies) ||
+		*svc.Spec.IPFamilyPolicy != *i.ServiceSnapshot.Spec.IPFamilyPolicy ||
+		// DDNS was disabled/enabled
+		svc.Annotations[kubevip.ServiceDDNS] != i.ServiceSnapshot.Annotations[kubevip.ServiceDDNS]
+}
+
+func (p *Processor) updateActiveServicesMetric() {
+	counts := map[string]int{}
+	for _, inst := range p.ServiceInstances {
+		if inst.ServiceSnapshot != nil {
+			counts[inst.ServiceSnapshot.Namespace]++
+		}
+	}
+	metrics.ActiveServices.Reset()
+	for ns, count := range counts {
+		metrics.ActiveServices.WithLabelValues(ns).Set(float64(count))
+	}
 }

pkg/services/services.go

@@ -2,9 +2,12 @@ package services
 
 import (
 	"context"
+	"errors"
 	"fmt"
+	"net"
 	"slices"
 	"strings"
+	"sync"
 	"time"
 
 	log "log/slog"
@@ -23,6 +26,8 @@ import (
 	"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/servicecontext"
 	"github.com/kube-vip/kube-vip/pkg/upnp"
 	"github.com/kube-vip/kube-vip/pkg/utils"
 	"github.com/kube-vip/kube-vip/pkg/vip"
@@ -34,35 +39,40 @@ const (
 	ActionDelete ServiceInstanceAction = "delete"
 	ActionAdd    ServiceInstanceAction = "add"
 	ActionNone   ServiceInstanceAction = "none"
+
+	// Default UPNP lease requested is 3600 seconds.
+	defaultUPNPLeaseDuration = 1 * time.Hour
 )
 
-func (p *Processor) SyncServices(ctx context.Context, svc *v1.Service) error {
-	log.Debug("[STARTING] Service Sync", "namespace", svc.Namespace, "name", svc.Name)
+func (p *Processor) SyncServices(ctx *servicecontext.Context, svc *v1.Service, wg *sync.WaitGroup, usesLeaderElection bool) error {
+	log.Debug("[STARTING] Service Sync", "namespace", svc.Namespace, "name", svc.Name, "uid", svc.UID)
 
 	// Iterate through the synchronising services
-
-	action := p.getServiceInstanceAction(svc)
+	action, instance := p.getServiceInstanceAction(svc)
 	switch action {
 	case ActionDelete:
-		// remove the label from the node before deleting the service
-		if err := p.nodeLabelManager.RemoveLabel(ctx, svc); err != nil {
-			return fmt.Errorf("error removing label from node: %w", err)
-		}
-
 		log.Debug("[service] delete", "namespace", svc.Namespace, "name", svc.Name, "uid", svc.UID)
-		if err := p.deleteService(svc.UID); err != nil {
+		if err := p.deleteService(ctx.Ctx, svc.UID); err != nil {
 			return fmt.Errorf("error deleting service %s/%s: %w", svc.Namespace, svc.Name, err)
 		}
 	case ActionAdd:
 		log.Debug("[service] add", "namespace", svc.Namespace, "name", svc.Name, "uid", svc.UID)
-		if err := p.addService(ctx, svc); err != nil {
+		if instance != nil {
+			instance.AddCalled = true
+		}
+
+		if !usesLeaderElection {
+			select {
+			case <-ctx.Ctx.Done():
+				return nil
+			case <-ctx.EndpointsReady:
+			}
+		}
+
+		if err := p.addService(ctx.Ctx, instance, svc, wg); err != nil {
 			return fmt.Errorf("error adding service %s/%s: %w", svc.Namespace, svc.Name, err)
 		}
 
-		// add the label to the node after adding the service
-		if err := p.nodeLabelManager.AddLabel(ctx, svc); err != nil {
-			return fmt.Errorf("error adding label to node: %w", err)
-		}
 	case ActionNone:
 		log.Debug("[service] no action", "namespace", svc.Namespace, "name", svc.Name, "uid", svc.UID)
 	}
@@ -70,7 +80,7 @@ func (p *Processor) SyncServices(ctx context.Context, svc *v1.Service) error {
 	return nil
 }
 
-func (p *Processor) getServiceInstanceAction(svc *v1.Service) ServiceInstanceAction {
+func (p *Processor) getServiceInstanceAction(svc *v1.Service) (ServiceInstanceAction, *instance.Instance) {
 	// protect against multiple calls
 	// get the annotations or legacy values from manual configuration
 	addresses, hostnames := instance.FetchServiceAddresses(svc)
@@ -81,37 +91,54 @@ func (p *Processor) getServiceInstanceAction(svc *v1.Service) ServiceInstanceAct
 
 	for _, instance := range p.ServiceInstances {
 		if instance != nil && instance.ServiceSnapshot.UID == svc.UID {
+			if !instance.AddCalled {
+				return ActionAdd, instance
+			}
 			for _, address := range addresses {
 				// handle the case where the service instance needs to be deleted
-				if instance.IsDHCP {
-					if address != "0.0.0.0" && address != "::" {
-						return ActionDelete
+				if instance.IsDHCPv4 {
+					if address != "0.0.0.0" {
+						return ActionDelete, instance
 					}
-					if len(svc.Status.LoadBalancer.Ingress) > 0 && !slices.Contains(statusAddresses, instance.DHCPInterfaceIP) {
-						return ActionDelete
+					if len(svc.Status.LoadBalancer.Ingress) > 0 && !slices.Contains(statusAddresses, instance.DHCPInterfaceIPv4) {
+						return ActionDelete, instance
 					}
-				}
-				if !instance.IsDHCP {
-					if address == "0.0.0.0" || address == "::" {
-						return ActionDelete
+				} else {
+					if address == "0.0.0.0" {
+						return ActionDelete, instance
 					}
 					if len(svc.Status.LoadBalancer.Ingress) > 0 && !slices.Contains(statusAddresses, address) {
-						return ActionDelete
+						return ActionDelete, instance
+					}
+				}
+				if instance.IsDHCPv6 {
+					if address != "::" {
+						return ActionDelete, instance
+					}
+					if len(svc.Status.LoadBalancer.Ingress) > 0 && !slices.Contains(statusAddresses, instance.DHCPInterfaceIPv6) {
+						return ActionDelete, instance
+					}
+				} else {
+					if address == "::" {
+						return ActionDelete, instance
+					}
+					if len(svc.Status.LoadBalancer.Ingress) > 0 && !slices.Contains(statusAddresses, address) {
+						return ActionDelete, instance
 					}
 				}
 				if len(svc.Status.LoadBalancer.Ingress) > 0 && !comparePortsAndPortStatuses(svc) {
-					return ActionDelete
+					return ActionDelete, instance
 				}
 			}
 			// If we reach here, it means the service instance matches the service UID and is not a DHCP service, so we can return "no action"
-			return ActionNone
+			return ActionNone, instance
 		}
 	}
 	if len(addresses) > 0 || len(hostnames) > 0 {
-		log.Debug("no matching service instance found", "service", svc.Name, "namespace", svc.Namespace, "addresses", addresses, "hostnames", hostnames)
-		return ActionAdd // If no matching instance is found, we need to add a new service instance
+		log.Debug("no matching service instance found", "service", svc.Name, "namespace", svc.Namespace, "uid", svc.UID, "addresses", addresses, "hostnames", hostnames)
+		return ActionAdd, nil // If no matching instance is found, we need to add a new service instance
 	}
-	return ActionNone
+	return ActionNone, nil
 }
 
 func comparePortsAndPortStatuses(svc *v1.Service) bool {
@@ -130,47 +157,115 @@ func comparePortsAndPortStatuses(svc *v1.Service) bool {
 	return true
 }
 
-func (p *Processor) addService(ctx context.Context, svc *v1.Service) error {
+func (p *Processor) addService(ctx context.Context, inst *instance.Instance, svc *v1.Service, wg *sync.WaitGroup) error {
 	// protect against addService while reading
 	p.mutex.Lock()
 	defer p.mutex.Unlock()
 
 	startTime := time.Now()
 
-	newService, err := instance.NewInstance(svc, p.config, p.intfMgr, p.arpMgr)
-	if err != nil {
-		return err
+	var err error
+	if inst == nil {
+		inst, err = instance.NewInstance(ctx, svc, p.config, p.intfMgr, p.arpMgr, p.routeMgr, p.nodeLabelManager, wg)
+		if err != nil {
+			return err
+		}
+		inst.AddCalled = true
+
+		p.ServiceInstances = append(p.ServiceInstances, inst)
 	}
 
-	for x := range newService.VIPConfigs {
-		log.Debug("starting loadbalancer for service", "name", svc.Name, "namespace", svc.Namespace)
-		newService.Clusters[x].StartLoadBalancerService(ctx, newService.VIPConfigs[x], p.bgpServer, svc.Name, p.CountRouteReferences)
+	if err := p.configureService(ctx, inst, svc, wg); err != nil {
+		return fmt.Errorf("failed to configure service: %w", err)
 	}
 
-	p.upnpMap(ctx, newService)
+	// add the label to the node after adding the service
+	labels := generateLabelsFromService(svc, kubevip.ServiceProvided)
+	if err := p.nodeLabelManager.AddLabel(labels); err != nil {
+		return fmt.Errorf("error adding label to node: %w", err)
+	}
+	inst.LabelAdded = true
 
-	if newService.IsDHCP && len(newService.VIPConfigs) == 1 {
-		go func() {
-			for ip := range newService.DHCPClient.IPChannel() {
-				log.Debug("IP changed", "ip", ip)
-				newService.VIPConfigs[0].VIP = ip
-				newService.DHCPInterfaceIP = ip
-				if !p.config.DisableServiceUpdates {
-					if err := p.updateStatus(newService); err != nil {
-						log.Warn("updating svc", "err", err)
-					}
+	finishTime := time.Since(startTime)
+	log.Info("[service]", "service", svc.Name, "namespace", svc.Namespace, "synchronised in", fmt.Sprintf("%dms", finishTime.Milliseconds()))
+
+	return nil
+}
+
+func (p *Processor) configureService(ctx context.Context, inst *instance.Instance, svc *v1.Service, wg *sync.WaitGroup) error {
+	// is not a global leader election mode
+	if p.config.EnableServicesElection || (!p.config.EnableARP && !p.config.EnableLeaderElection) || (!p.config.EnableARP && !p.config.EnableRoutingTable) {
+		for x := range inst.VIPConfigs {
+			log.Debug("starting loadbalancer for service", "name", svc.Name, "namespace", svc.Namespace, "uid", svc.UID)
+			if err := inst.Clusters[x].StartLoadBalancerService(ctx, inst.VIPConfigs[x], p.bgpServer, lease.ServiceNamespacedName(svc), wg); err != nil {
+				return fmt.Errorf("failed to start lb: %w", err)
+			}
+		}
+	}
+
+	p.upnpMap(ctx, inst)
+
+	if inst.IsDHCPv4 {
+		wg.Go(func() {
+			index := -1
+			for i := range inst.VIPConfigs {
+				ip := net.ParseIP(inst.VIPConfigs[i].VIP)
+				if ip.To4() != nil {
+					index = i
+					break
 				}
 			}
-			log.Debug("IP update channel closed, stopping")
-		}()
+			if index == -1 {
+				log.Error("unable to find proper VIPConfig for the DHCPv4")
+			} else {
+				for ip := range inst.DHCPv4Client.IPChannel() {
+					log.Debug("IP changed", "ip", ip)
+					inst.VIPConfigs[index].VIP = ip
+					inst.DHCPInterfaceIPv4 = ip
+					if !p.config.DisableServiceUpdates {
+						if err := p.updateStatus(ctx, inst); err != nil {
+							log.Warn("updating svc", "err", err)
+						}
+					}
+				}
+				log.Debug("IPv4 update channel closed, stopping")
+			}
+
+		})
 	}
 
-	p.ServiceInstances = append(p.ServiceInstances, newService)
+	if inst.IsDHCPv6 {
+		wg.Go(func() {
+			index := -1
+			for i := range inst.VIPConfigs {
+				ip := net.ParseIP(inst.VIPConfigs[i].VIP)
+				if ip.To4() == nil {
+					index = i
+					break
+				}
+			}
+			if index == -1 {
+				log.Error("unable to find proper VIPConfig for the DHCPv6")
+			} else {
+				for ip := range inst.DHCPv4Client.IPChannel() {
+					log.Debug("IP changed", "ip", ip)
+					inst.VIPConfigs[index].VIP = ip
+					inst.DHCPInterfaceIPv6 = ip
+					if !p.config.DisableServiceUpdates {
+						if err := p.updateStatus(ctx, inst); err != nil {
+							log.Warn("updating svc", "err", err)
+						}
+					}
+				}
+				log.Debug("IPv6 update channel closed, stopping")
+			}
+		})
+	}
 
 	if !p.config.DisableServiceUpdates {
-		log.Debug("[service] update", "namespace", newService.ServiceSnapshot.Namespace, "name", newService.ServiceSnapshot.Name)
-		if err := p.updateStatus(newService); err != nil {
-			log.Error("[service] updating status", "namespace", newService.ServiceSnapshot.Namespace, "name", newService.ServiceSnapshot.Name, "err", err)
+		log.Debug("[service] update", "namespace", inst.ServiceSnapshot.Namespace, "name", inst.ServiceSnapshot.Name)
+		if err := p.updateStatus(ctx, inst); err != nil {
+			log.Error("[service] updating status", "namespace", inst.ServiceSnapshot.Namespace, "name", inst.ServiceSnapshot.Name, "err", err)
 		}
 	}
 
@@ -180,7 +275,7 @@ func (p *Processor) addService(ctx context.Context, svc *v1.Service) error {
 
 		log.Debug("[service] Flushing conntrack rules", "service", svc.Name, "namespace", svc.Namespace)
 		for _, serviceIP := range serviceIPs {
-			err = vip.DeleteExistingSessions(serviceIP, false, svc.Annotations[kubevip.EgressDestinationPorts], svc.Annotations[kubevip.EgressSourcePorts])
+			err := vip.DeleteExistingSessions(serviceIP, false, svc.Annotations[kubevip.EgressDestinationPorts], svc.Annotations[kubevip.EgressSourcePorts])
 			if err != nil {
 				log.Error("[service] flushing any remaining egress connections", "service", svc.Name, "namespace", svc.Namespace, "err", err)
 			}
@@ -197,13 +292,13 @@ func (p *Processor) addService(ctx context.Context, svc *v1.Service) error {
 		// If we'er not using NFtables, then ensure that the correct iptables modules are loaded
 		if p.config.EgressWithNftables {
 			// Ensure that kernel modules are loaded and report back missing modules.
-			err = p.nftablesCheck()
+			err := p.nftablesCheck()
 			if err != nil {
 				log.Warn("[service] configuring nft egress", "service", svc.Name, "namespace", svc.Namespace, "err", err)
 			}
 		} else {
 			// Ensure that kernel modules are loaded and report back missing modules.
-			err = p.iptablesCheck()
+			err := p.iptablesCheck()
 			if err != nil {
 				log.Warn("[service] configuring egress", "service", svc.Name, "namespace", svc.Namespace, "err", err)
 			}
@@ -220,7 +315,7 @@ func (p *Processor) addService(ctx context.Context, svc *v1.Service) error {
 
 						podIP = svc.Annotations[kubevip.ActiveEndpointIPv6]
 
-						err = p.configureEgress(serviceIP, podIP, svc.Namespace, string(svc.UID), svc.Annotations)
+						err := p.configureEgress(ctx, serviceIP, podIP, svc.Namespace, string(svc.UID), svc.Annotations)
 						if err != nil {
 							errList = append(errList, err)
 							log.Warn("[service] configuring egress IPv6", "service", svc.Name, "namespace", svc.Namespace, "err", err)
@@ -234,7 +329,7 @@ func (p *Processor) addService(ctx context.Context, svc *v1.Service) error {
 				if !p.config.EnableEndpoints && utils.IsIPv6(serviceIP) {
 					podIPs = svc.Annotations[kubevip.ActiveEndpointIPv6]
 				}
-				err = p.configureEgress(serviceIP, podIPs, svc.Namespace, string(svc.UID), svc.Annotations)
+				err := p.configureEgress(ctx, serviceIP, podIPs, svc.Namespace, string(svc.UID), svc.Annotations)
 				if err != nil {
 					errList = append(errList, err)
 					log.Warn("[service] configuring egress IPv4", "service", svc.Name, "namespace", svc.Namespace, "err", err)
@@ -248,20 +343,26 @@ func (p *Processor) addService(ctx context.Context, svc *v1.Service) error {
 			} else {
 				provider = providers.NewEndpointslices()
 			}
-			err = provider.UpdateServiceAnnotation(svc.Annotations[kubevip.ActiveEndpoint], svc.Annotations[kubevip.ActiveEndpointIPv6], svc, p.clientSet)
+			err := provider.UpdateServiceAnnotation(ctx, svc.Annotations[kubevip.ActiveEndpoint], svc.Annotations[kubevip.ActiveEndpointIPv6], svc, p.clientSet)
 			if err != nil {
 				log.Warn("[service] configuring egress", "service", svc.Name, "namespace", svc.Namespace, "err", err)
 			}
 		}
 	}
 
-	finishTime := time.Since(startTime)
-	log.Info("[service]", "service", svc.Name, "namespace", svc.Namespace, "synchronised in", fmt.Sprintf("%dms", finishTime.Milliseconds()))
+	// Configure WireGuard DNAT rules if WireGuard is enabled
+	if p.config.EnableWireguard {
+		log.Debug("[service] configuring WireGuard DNAT rules", "service", svc.Name, "namespace", svc.Namespace)
+		if err := p.addServiceWireguard(ctx, svc); err != nil {
+			log.Warn("[service] failed to configure WireGuard DNAT", "service", svc.Name, "namespace", svc.Namespace, "err", err)
+			// Don't fail the entire service if WireGuard config fails
+		}
+	}
 
 	return nil
 }
 
-func (p *Processor) deleteService(uid types.UID) error {
+func (p *Processor) deleteService(ctx context.Context, uid types.UID) error {
 	// protect multiple calls
 	p.mutex.Lock()
 	defer p.mutex.Unlock()
@@ -270,7 +371,7 @@ func (p *Processor) deleteService(uid types.UID) error {
 	var serviceInstance *instance.Instance
 	found := false
 	for x := range p.ServiceInstances {
-		log.Debug("[service] lookup", "target UID", uid, "found UID ", p.ServiceInstances[x].ServiceSnapshot.UID, "name", p.ServiceInstances[x].ServiceSnapshot.Name, "namespace", p.ServiceInstances[x].ServiceSnapshot.Namespace)
+		log.Debug("[service] lookup", "target UID", uid, "found UID", p.ServiceInstances[x].ServiceSnapshot.UID, "name", p.ServiceInstances[x].ServiceSnapshot.Name, "namespace", p.ServiceInstances[x].ServiceSnapshot.Namespace)
 		// Add the running services to the new array
 		if p.ServiceInstances[x].ServiceSnapshot.UID != uid {
 			updatedInstances = append(updatedInstances, p.ServiceInstances[x])
@@ -281,6 +382,13 @@ func (p *Processor) deleteService(uid types.UID) error {
 		}
 	}
 
+	if serviceInstance.LabelAdded {
+		labels := generateLabelsFromService(serviceInstance.ServiceSnapshot, kubevip.ServiceProvided)
+		if err := p.nodeLabelManager.RemoveLabel(labels); err != nil {
+			return fmt.Errorf("error removing label from node: %w", err)
+		}
+	}
+
 	// If we've been through all services and not found the correct one then error
 	if !found {
 		// TODO: - fix UX
@@ -295,7 +403,7 @@ func (p *Processor) deleteService(uid types.UID) error {
 		}
 	}
 
-	// Determine if this this VIP is shared with other loadbalancers
+	// Determine if this VIP is shared with other loadbalancers
 	shared := false
 	vipSet := make(map[string]interface{})
 	for x := range updatedInstances {
@@ -310,12 +418,45 @@ func (p *Processor) deleteService(uid types.UID) error {
 			shared = true
 		}
 	}
+
+	if p.config.EnableBGP {
+		endpoints.ClearBGPHostsByInstance(ctx, serviceInstance, p.bgpServer)
+	}
+
+	if p.config.EnableRoutingTable && (p.config.EnableLeaderElection || p.config.EnableServicesElection) {
+		if errs := endpoints.ClearRoutesByInstance(serviceInstance.ServiceSnapshot, serviceInstance, &p.ServiceInstances, p.routeMgr); len(errs) > 0 {
+			for _, err := range errs {
+				log.Error("unable to clear routes", "err", err)
+			}
+		}
+	}
+
 	if !shared {
 		for x := range serviceInstance.Clusters {
 			serviceInstance.Clusters[x].Stop()
 		}
-		if serviceInstance.IsDHCP {
-			serviceInstance.DHCPClient.Stop()
+
+		if serviceInstance.IsVLAN {
+			vlan, err := netlink.LinkByName(serviceInstance.VLANInterface)
+			if err != nil {
+				return fmt.Errorf("[service] error finding VLAN Interface: %v", err)
+			}
+
+			err = netlink.LinkDel(vlan)
+			if err != nil {
+				return fmt.Errorf("[service] error deleting VLAN interface : %v", err)
+			}
+		}
+
+		if serviceInstance.IsDHCPv4 || serviceInstance.IsDHCPv6 {
+			if serviceInstance.IsDHCPv4 {
+				serviceInstance.DHCPv4Client.Stop()
+			}
+
+			if serviceInstance.IsDHCPv6 {
+				serviceInstance.DHCPv6Client.Stop()
+			}
+
 			macvlan, err := netlink.LinkByName(serviceInstance.DHCPInterface)
 			if err != nil {
 				return fmt.Errorf("[service] error finding VIP Interface: %v", err)
@@ -327,17 +468,6 @@ func (p *Processor) deleteService(uid types.UID) error {
 			}
 		}
 
-		if p.config.EnableBGP {
-			endpoints.ClearBGPHostsByInstance(serviceInstance, p.bgpServer)
-		}
-		if p.config.EnableRoutingTable && (p.config.EnableLeaderElection || p.config.EnableServicesElection) {
-			if errs := endpoints.ClearRoutesByInstance(serviceInstance.ServiceSnapshot, serviceInstance, &p.ServiceInstances); len(errs) > 0 {
-				for _, err := range errs {
-					log.Error("unable to clear routes", "err", err)
-				}
-			}
-		}
-
 		// We will need to tear down the egress
 		if serviceInstance.ServiceSnapshot.Annotations[kubevip.Egress] == "true" {
 			if serviceInstance.ServiceSnapshot.Annotations[kubevip.ActiveEndpoint] != "" {
@@ -353,11 +483,183 @@ func (p *Processor) deleteService(uid types.UID) error {
 	// Update the service array
 	p.ServiceInstances = updatedInstances
 
+	// Clean up WireGuard DNAT rules if WireGuard is enabled
+	if p.config.EnableWireguard {
+		log.Debug("[service] cleaning up WireGuard DNAT rules", "uid", uid, "name", serviceInstance.ServiceSnapshot.Name)
+		p.deleteServiceWireguard(ctx, serviceInstance.ServiceSnapshot)
+	}
+
 	log.Info("Removed instance from manager", "uid", uid, "name", serviceInstance.ServiceSnapshot.Name, "remaining advertised services", len(p.ServiceInstances))
 
 	return nil
 }
 
+func (p *Processor) updateEgressConfiguration(ctx context.Context, svc *v1.Service) error {
+	p.mutex.Lock()
+	defer p.mutex.Unlock()
+
+	i := instance.FindServiceInstance(svc, p.ServiceInstances)
+	if i == nil {
+		return fmt.Errorf("service instance not found for %s/%s", svc.Namespace, svc.Name)
+	}
+
+	oldIPv4 := i.ServiceSnapshot.Annotations[kubevip.ActiveEndpoint]
+	newIPv4 := svc.Annotations[kubevip.ActiveEndpoint]
+	oldIPv6 := i.ServiceSnapshot.Annotations[kubevip.ActiveEndpointIPv6]
+	newIPv6 := svc.Annotations[kubevip.ActiveEndpointIPv6]
+
+	// Skip update if endpoints haven't changed
+	if oldIPv4 == newIPv4 && oldIPv6 == newIPv6 {
+		return nil
+	}
+
+	log.Info("[service] updating egress configuration",
+		"service", svc.Name,
+		"namespace", svc.Namespace,
+		"old_ipv4", oldIPv4,
+		"new_ipv4", newIPv4,
+		"old_ipv6", oldIPv6,
+		"new_ipv6", newIPv6)
+
+	// Remove old egress rules if they exist
+	if oldIPv4 != "" || oldIPv6 != "" {
+		oldEndpoint := oldIPv4
+		if oldEndpoint == "" {
+			oldEndpoint = oldIPv6
+		}
+		serviceIPs, _ := instance.FetchServiceAddresses(i.ServiceSnapshot)
+		for _, serviceIP := range serviceIPs {
+			if err := egress.Teardown(
+				oldEndpoint,
+				serviceIP,
+				i.ServiceSnapshot.Namespace,
+				string(i.ServiceSnapshot.UID),
+				i.ServiceSnapshot.Annotations,
+				p.config.EgressWithNftables,
+			); err != nil {
+				log.Warn("[service] removing old egress rules", "service", svc.Name, "namespace", svc.Namespace, "err", err)
+			}
+		}
+	}
+
+	// Apply new egress rules with updated endpoint
+	serviceIPs, _ := instance.FetchServiceAddresses(svc)
+	errList := []error{}
+
+	// Check if egress should be IPv6
+	if svc.Annotations[kubevip.EgressIPv6] == "true" {
+		// Does the service have an active IPv6 endpoint
+		if newIPv6 != "" {
+			for _, serviceIP := range serviceIPs {
+				if !p.config.EnableEndpoints && utils.IsIPv6(serviceIP) {
+					podIP := newIPv6
+					err := p.configureEgress(ctx, serviceIP, podIP, svc.Namespace, string(svc.UID), svc.Annotations)
+					if err != nil {
+						errList = append(errList, err)
+						log.Warn("[service] configuring egress IPv6", "service", svc.Name, "namespace", svc.Namespace, "err", err)
+					}
+				}
+			}
+		}
+	} else if newIPv4 != "" { // Not expected to be IPv6, so should be an IPv4 address
+		for _, serviceIP := range serviceIPs {
+			podIPs := newIPv4
+			if !p.config.EnableEndpoints && utils.IsIPv6(serviceIP) {
+				podIPs = newIPv6
+			}
+			err := p.configureEgress(ctx, serviceIP, podIPs, svc.Namespace, string(svc.UID), svc.Annotations)
+			if err != nil {
+				errList = append(errList, err)
+				log.Warn("[service] configuring egress IPv4", "service", svc.Name, "namespace", svc.Namespace, "err", err)
+			}
+		}
+	}
+
+	if len(errList) > 0 {
+		return fmt.Errorf("errors configuring egress: %v", errList)
+	}
+
+	// Update the service snapshot to reflect the new state
+	// NOTE: Do NOT call UpdateServiceAnnotation here - the annotation was already updated
+	// by the endpoint processor, which is why we're being called in the first place.
+	// Calling it again would create an infinite loop of Modified events.
+	// svc is already a DeepCopy from the endpoint processor, so no need to copy again.
+	i.ServiceSnapshot = svc
+
+	log.Info("[service] egress configuration updated successfully", "service", svc.Name, "namespace", svc.Namespace)
+	return nil
+}
+
+// upnpLeaseDurationForService determines the UPNP lease duration for a given service, based on its annotations.
+//
+// The default lease duration is set to 1 hour, maintaining the default of 3600 seconds that was previously passed. If
+// the service has an annotation of [kubevip.UpnpLeaseDuration], the function attempts to parse its value as a
+// [time.Duration] using [time.ParseDuration].
+//
+// If parsing is successful, the lease duration is updated accordingly; otherwise, a warning is logged and the default
+// duration is retained.
+//
+// Overriding the default lease duration can be useful for services that require longer or shorter UPNP port mappings,
+// or for buggy UPNP implementations that may not handle renewals correctly. At least one router's implementation
+// completely times out the mapping very shortly after creation if it is set to 3600 or 7200, but works fine if 0 is
+// used.
+//
+// This function must therefore explicitly permit duration of 0, and callers and the underlying library must pass that
+// value in XML correctly. A duration of 0 indicates to the UPNP gateway that the mapping should be permanent.
+//
+// It may be useful to update this function to read a global configuration option as well. This helper could also take
+// in v1.Service instead of [instance.Instance], but the latter is more convenient for callers.
+//
+// Example where 0 was observed to stay on the problematic router: miniupnpc's test client, upnpc v2.2.4.
+func upnpLeaseDurationForService(s *instance.Instance) time.Duration {
+	if s == nil || s.ServiceSnapshot == nil || s.ServiceSnapshot.Annotations == nil {
+		// No warning output. No annotation is unusual but perfectly ok.
+		return defaultUPNPLeaseDuration
+	}
+
+	// Constant is named `UpnpLeaseDuration` for consistency with `UpnpEnabled`. According to Go naming conventions
+	// regarding use of acronyms, `UPNPLeaseDuration` would be preferred. Cleanup of both of these is left for a future
+	// refactor as these are public symbols and might be used elsewhere in the ecosystem.
+	val, ok := s.ServiceSnapshot.Annotations[kubevip.UpnpLeaseDuration]
+	if !ok {
+		// No warning output. No annotation is common and perfectly ok.
+		return defaultUPNPLeaseDuration
+	}
+
+	if val == "" {
+		log.Warn("[UPNP] Lease duration annotation is empty, using default of 1 hour", "service", s.ServiceSnapshot.Name)
+		return defaultUPNPLeaseDuration
+	}
+
+	parsed, err := time.ParseDuration(val)
+	if err != nil {
+		log.Warn("[UPNP] Unable to parse lease duration from annotation, using default of 1 hour", "service", s.ServiceSnapshot.Name, "err", err)
+		return defaultUPNPLeaseDuration
+	}
+
+	if parsed < 0 {
+		log.Warn("[UPNP] Lease duration from annotation is negative, using default of 1 hour", "service", s.ServiceSnapshot.Name)
+		return defaultUPNPLeaseDuration
+	}
+
+	return parsed
+}
+
+// upnpLeaseDurationForServiceSec returns the UPNP lease duration for a service in uint32 seconds, as expected by the
+// helper library. This is a convenience wrapper around [upnpLeaseDurationForService], and in case
+// upnpLeaseDurationForService returns a duration that maps to a negative value of seconds or invalid float of seconds,
+// it will return the default lease duration in seconds instead. (Technically, it will check for a reasonable range of
+// seconds, e.g. ~10 years-ish.)
+func upnpLeaseDurationForServiceSec(s *instance.Instance) uint32 {
+	duration := upnpLeaseDurationForService(s)
+	seconds := duration.Seconds()
+	// Check if within range.
+	if seconds >= 0 && seconds <= float64(10*365*24*60*60) {
+		return uint32(seconds)
+	}
+	return uint32(defaultUPNPLeaseDuration.Seconds())
+}
+
 // 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 (p *Processor) upnpMap(ctx context.Context, s *instance.Instance) {
@@ -374,6 +676,10 @@ func (p *Processor) upnpMap(ctx context.Context, s *instance.Instance) {
 
 	gateways := upnp.GetGatewayClients(ctx)
 
+	// Determine desired UPNP TTL / "lease duration". Passed into the library as integer seconds from now, as the
+	// underlying XML API wants integer seconds.
+	leaseDurationSec := upnpLeaseDurationForServiceSec(s)
+
 	// Reset Gateway IPs to remove stale addresses
 	s.UPNPGatewayIPs = make([]string, 0)
 
@@ -384,9 +690,9 @@ func (p *Processor) upnpMap(ctx context.Context, s *instance.Instance) {
 
 				forwardSucessful := false
 				if gw.WANIPv6FirewallControlClient != nil {
-					log.Info("[UPNP] Adding map", "vip", vip, "port", port.Port, "service", s.ServiceSnapshot.Name, "gateway", gw.WANIPv6FirewallControlClient.Location)
+					log.Info("[UPNP] Adding map", "vip", vip, "port", port.Port, "service", s.ServiceSnapshot.Name, "gateway", gw.WANIPv6FirewallControlClient.Location, "leaseDurationSec", leaseDurationSec)
 
-					pinholeID, pinholeErr := gw.WANIPv6FirewallControlClient.AddPinholeCtx(ctx, "0.0.0.0", uint16(port.Port), vip, uint16(port.Port), upnp.MapProtocolToIANA(string(port.Protocol)), 3600) //nolint  TODO
+					pinholeID, pinholeErr := gw.WANIPv6FirewallControlClient.AddPinholeCtx(ctx, "0.0.0.0", uint16(port.Port), vip, uint16(port.Port), upnp.MapProtocolToIANA(string(port.Protocol)), leaseDurationSec) //nolint  TODO
 					if pinholeErr == nil {
 						forwardSucessful = true
 						log.Info("[UPNP] Service should be accessible externally", "port", port.Port, "pinhold ID", pinholeID)
@@ -397,16 +703,16 @@ func (p *Processor) upnpMap(ctx context.Context, s *instance.Instance) {
 				}
 				// Fallback to PortForward
 				if !forwardSucessful {
-					log.Info("[UPNP] Adding map", "vip", vip, "port", port.Port, "service", s.ServiceSnapshot.Name)
+					log.Info("[UPNP] Adding map", "vip", vip, "port", port.Port, "service", s.ServiceSnapshot.Name, "leaseDurationSec", leaseDurationSec)
 
-					portMappingErr := gw.ConnectionClient.AddPortMapping("0.0.0.0", uint16(port.Port), strings.ToUpper(string(port.Protocol)), uint16(port.Port), vip, true, s.ServiceSnapshot.Name, 3600) //nolint  TODO
+					portMappingErr := gw.ConnectionClient.AddPortMapping("0.0.0.0", uint16(port.Port), strings.ToUpper(string(port.Protocol)), uint16(port.Port), vip, true, s.ServiceSnapshot.Name, leaseDurationSec) //nolint  TODO
 					if portMappingErr == nil {
 						ip, err := gw.ConnectionClient.GetExternalIPAddress()
 						if err != nil {
 							// Log the error but continue on the off chance the mapping was successful
-							log.Error("[UPNP] Unable to get external IP address from gateway", "port", port.Port, "err", err)
+							log.Error("[UPNP] Unable to get external IP address from gateway", "service", s.ServiceSnapshot.Name, "port", port.Port, "err", err)
 						} else {
-							log.Info("[UPNP] Service should be accessible externally", "port", port.Port, "externalip", ip)
+							log.Info("[UPNP] Service should be accessible externally", "service", s.ServiceSnapshot.Name, "port", port.Port, "externalip", ip)
 						}
 						forwardSucessful = true
 					} else {
@@ -430,7 +736,7 @@ func (p *Processor) upnpMap(ctx context.Context, s *instance.Instance) {
 	s.UPNPGatewayIPs = slices.Compact(s.UPNPGatewayIPs)
 }
 
-func (p *Processor) updateStatus(i *instance.Instance) error {
+func (p *Processor) updateStatus(ctx context.Context, i *instance.Instance) error {
 	// let's retry status update every 10ms for 30s
 	retryConfig := wait.Backoff{
 		Steps:    3000,
@@ -439,10 +745,12 @@ func (p *Processor) updateStatus(i *instance.Instance) error {
 		Jitter:   0.1,
 	}
 	// will retry for every error encountered, TODO: should a list of errors that will trigger retry be specified?
-	err := retry.OnError(retryConfig, func(error) bool { return true }, func() error {
+	err := retry.OnError(retryConfig, func(err error) bool {
+		return !errors.Is(err, context.Canceled)
+	}, func() error {
 		// Retrieve the latest version of Deployment before attempting update
 		// RetryOnConflict uses exponential backoff to avoid exhausting the apiserver
-		currentService, err := p.clientSet.CoreV1().Services(i.ServiceSnapshot.Namespace).Get(context.TODO(), i.ServiceSnapshot.Name, metav1.GetOptions{})
+		currentService, err := p.clientSet.CoreV1().Services(i.ServiceSnapshot.Namespace).Get(ctx, i.ServiceSnapshot.Name, metav1.GetOptions{})
 		if err != nil {
 			return err
 		}
@@ -457,9 +765,19 @@ func (p *Processor) updateStatus(i *instance.Instance) error {
 			// Add the current host
 			currentServiceCopy.Annotations[kubevip.VipHost] = p.config.NodeName
 		}
-		if i.DHCPInterfaceHwaddr != "" || i.DHCPInterfaceIP != "" {
+		if i.DHCPInterfaceHwaddr != "" || i.DHCPInterfaceIPv4 != "" || i.DHCPInterfaceIPv6 != "" {
 			currentServiceCopy.Annotations[kubevip.HwAddrKey] = i.DHCPInterfaceHwaddr
-			currentServiceCopy.Annotations[kubevip.RequestedIP] = i.DHCPInterfaceIP
+			dhcpInterfaceIP := ""
+			if i.DHCPInterfaceIPv4 != "" {
+				dhcpInterfaceIP = i.DHCPInterfaceIPv4
+				if i.DHCPInterfaceIPv6 != "" {
+					dhcpInterfaceIP += ","
+				}
+			}
+			if i.DHCPInterfaceIPv6 != "" {
+				dhcpInterfaceIP += i.DHCPInterfaceIPv6
+			}
+			currentServiceCopy.Annotations[kubevip.RequestedIP] = dhcpInterfaceIP
 		}
 
 		if currentService.Annotations["development.kube-vip.io/synthetic-api-server-error-on-update"] == "true" {
@@ -468,7 +786,7 @@ func (p *Processor) updateStatus(i *instance.Instance) error {
 		}
 
 		if !cmp.Equal(currentService, currentServiceCopy) {
-			currentService, err = p.clientSet.CoreV1().Services(currentServiceCopy.Namespace).Update(context.TODO(), currentServiceCopy, metav1.UpdateOptions{})
+			currentService, err = p.clientSet.CoreV1().Services(currentServiceCopy.Namespace).Update(ctx, currentServiceCopy, metav1.UpdateOptions{})
 			if err != nil {
 				log.Error("updating Spec", "service", i.ServiceSnapshot.Name, "err", err)
 				return err
@@ -487,7 +805,7 @@ func (p *Processor) updateStatus(i *instance.Instance) error {
 
 		for _, c := range i.VIPConfigs {
 			if !utils.IsIP(c.VIP) {
-				ips, err := utils.LookupHost(c.VIP, p.config.DNSMode)
+				ips, err := utils.LookupHost(c.VIP, c.DNSMode, *i.ServiceSnapshot.Spec.IPFamilyPolicy == v1.IPFamilyPolicyRequireDualStack)
 				if err != nil {
 					return err
 				}
@@ -515,9 +833,11 @@ func (p *Processor) updateStatus(i *instance.Instance) error {
 				}
 			}
 		}
-		if !cmp.Equal(currentService.Status.LoadBalancer.Ingress, ingresses) {
+		log.Debug("LB status", "current", currentService.Status.LoadBalancer.Ingress, "new", ingresses)
+		if !ingressEqual(currentService.Status.LoadBalancer.Ingress, ingresses) {
 			currentService.Status.LoadBalancer.Ingress = ingresses
-			_, err = p.clientSet.CoreV1().Services(currentService.Namespace).UpdateStatus(context.TODO(), currentService, metav1.UpdateOptions{})
+			log.Debug("updating service status", "namespace", currentService.Namespace, "name", currentService.Name, "uid", currentService.UID)
+			_, err = p.clientSet.CoreV1().Services(currentService.Namespace).UpdateStatus(ctx, currentService, metav1.UpdateOptions{})
 			if err != nil && !apierrors.IsInvalid(err) {
 				log.Error("updating Service", "namespace", i.ServiceSnapshot.Namespace, "name", i.ServiceSnapshot.Name, "err", err)
 				return err
@@ -529,22 +849,61 @@ func (p *Processor) updateStatus(i *instance.Instance) error {
 	return err
 }
 
+func ingressEqual(a, b []v1.LoadBalancerIngress) bool {
+	if len(a) != len(b) {
+		return false
+	}
+	for i := range len(a) {
+		if a[i].IP != b[i].IP || a[i].Hostname != b[i].Hostname ||
+			!cmp.Equal(a[i].Ports, b[i].Ports) {
+			return false
+		}
+	}
+	return true
+}
+
 func isUPNPEnabled(s *v1.Service) bool {
 	return metav1.HasAnnotation(s.ObjectMeta, kubevip.UpnpEnabled) && s.Annotations[kubevip.UpnpEnabled] == "true"
 }
 
 // Refresh UPNP Port Forwards for all Service Instances registered in the processor
-func (p *Processor) RefreshUPNPForwards() {
+func (p *Processor) RefreshUPNPForwards(ctx context.Context) {
 	log.Info("Starting UPNP Port Refresher")
-	for {
-		time.Sleep(300 * time.Second)
 
-		log.Info("[UPNP] Refreshing Instances", "number of instances", len(p.ServiceInstances))
-		for i := range p.ServiceInstances {
-			p.upnpMap(context.TODO(), p.ServiceInstances[i])
-			if err := p.updateStatus(p.ServiceInstances[i]); err != nil {
-				log.Warn("[UPNP] Error updating service", "ip", p.ServiceInstances[i].ServiceSnapshot.Name, "err", err)
+	ticker := time.NewTicker(300 * time.Second)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case <-ticker.C:
+			// Skip logging if no service instances
+			if len(p.ServiceInstances) == 0 {
+				continue
+			}
+
+			log.Info("[UPNP] Refreshing Instances", "number of instances", len(p.ServiceInstances))
+			for i := range p.ServiceInstances {
+				p.upnpMap(ctx, p.ServiceInstances[i])
+				if err := p.updateStatus(ctx, p.ServiceInstances[i]); err != nil {
+					log.Warn("[UPNP] Error updating service", "ip", p.ServiceInstances[i].ServiceSnapshot.Name, "err", err)
+				}
 			}
 		}
 	}
 }
+
+// GenerateLabelFromService generates a label key and value for the given service
+func generateLabelsFromService(svc *v1.Service, labelKey string) map[string]string {
+	addresses, _ := instance.FetchServiceAddresses(svc)
+
+	sanitized := make([]string, len(addresses))
+	for i, addr := range addresses {
+		sanitized[i] = utils.SanitizeIPForLabel(addr)
+	}
+
+	return map[string]string{
+		fmt.Sprintf("%s/%s.%s", labelKey, svc.Name, svc.Namespace): strings.Join(sanitized, ","),
+	}
+}

pkg/services/services_test.go

@@ -0,0 +1,68 @@
+package services
+
+import (
+	"testing"
+
+	v1 "k8s.io/api/core/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+
+	"github.com/kube-vip/kube-vip/pkg/instance"
+)
+
+// Test_upnpLeaseDurationForService tests whether the default lease duration is used, and whether the annotation
+// overrides it correctly.
+//
+// For simplicity, this table driven test does not cover all edge cases (passing in nil instance, nil service snapshot,
+// nil annotations, etc).
+func Test_upnpLeaseDurationForService(t *testing.T) {
+	const annotation = "kube-vip.io/upnp-lease-duration" // Validating value of kubevip.UpnpLeaseDuration.
+	tcs := []struct {
+		name        string
+		annotations map[string]string
+		want        int // in seconds
+	}{
+		{
+			name:        "No annotation uses default",
+			annotations: map[string]string{},
+			want:        3600,
+		},
+		{
+			name: "Valid annotation overrides default",
+			annotations: map[string]string{
+				annotation: "2h",
+			},
+			want: 7200,
+		},
+		{
+			name: "Valid short annotation overrides default",
+			annotations: map[string]string{
+				annotation: "30m",
+			},
+			want: 1800,
+		},
+		{
+			name: "Invalid annotation uses default",
+			annotations: map[string]string{
+				annotation: "invalid-duration",
+			},
+			want: 3600,
+		},
+	}
+
+	for _, tc := range tcs {
+		t.Run(tc.name, func(t *testing.T) {
+			i := &instance.Instance{
+				ServiceSnapshot: &v1.Service{
+					ObjectMeta: metav1.ObjectMeta{
+						Annotations: tc.annotations,
+					},
+				},
+			}
+			gotDuration := upnpLeaseDurationForService(i)
+			got := int(gotDuration.Seconds())
+			if got != tc.want {
+				t.Errorf("upnpLeaseDurationForService(%+v) = %v, want %v", tc.annotations, got, tc.want)
+			}
+		})
+	}
+}

pkg/services/services_wireguard.go

@@ -0,0 +1,170 @@
+package services
+
+import (
+	"context"
+	"fmt"
+
+	log "log/slog"
+
+	"github.com/kube-vip/kube-vip/pkg/nftables"
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	v1 "k8s.io/api/core/v1"
+)
+
+// addServiceWireguard configures a WireGuard tunnel for a service
+// The tunnel is brought up here, but DNAT rules are configured by the endpoint watcher
+// via wireguardWorker.processInstance() when endpoints become available
+func (p *Processor) addServiceWireguard(_ context.Context, svc *v1.Service) error {
+	if !p.config.EnableWireguard {
+		return nil
+	}
+
+	// Get service VIPs
+	serviceIPs, err := utils.FetchServiceIPs(svc)
+	if err != nil {
+		return fmt.Errorf("failed to get service IPs for %s/%s: %w", svc.Namespace, svc.Name, err)
+	}
+
+	if len(serviceIPs) == 0 {
+		return fmt.Errorf("no service IPs found for service %s/%s", svc.Namespace, svc.Name)
+	}
+
+	// For each VIP, bring up the WireGuard tunnel
+	// DNAT rules will be configured by the endpoint watcher when endpoints are available
+	var successCount int
+	var lastErr error
+	for _, vip := range serviceIPs {
+		if err := p.setupServiceWireguardTunnel(svc, vip); err != nil {
+			log.Error("[wireguard] failed to setup tunnel for VIP",
+				"service", svc.Name,
+				"namespace", svc.Namespace,
+				"vip", vip,
+				"err", err)
+			lastErr = err
+			// Continue with other VIPs even if one fails
+			continue
+		}
+		successCount++
+	}
+
+	if successCount == 0 {
+		return fmt.Errorf("failed to setup WireGuard tunnel for any VIP in service %s/%s: %w", svc.Namespace, svc.Name, lastErr)
+	}
+
+	return nil
+}
+
+// setupServiceWireguardTunnel brings up the WireGuard tunnel for a single VIP
+// DNAT rules are NOT configured here - they are handled by the endpoint watcher
+func (p *Processor) setupServiceWireguardTunnel(svc *v1.Service, vip string) error {
+	// Check if we have a tunnel configuration for this VIP
+	if !p.TunnelMgr.HasConfigForVIP(vip) {
+		return fmt.Errorf("no WireGuard tunnel configuration found for VIP %s", vip)
+	}
+
+	// Get the tunnel configuration to determine the interface name
+	tunnelConfig := p.TunnelMgr.GetConfigForVIP(vip)
+	if tunnelConfig == nil {
+		return fmt.Errorf("failed to get tunnel configuration for VIP %s", vip)
+	}
+
+	// Bring up the WireGuard tunnel for this VIP
+	if err := p.TunnelMgr.BringUpTunnelForVIP(vip); err != nil {
+		return fmt.Errorf("failed to bring up WireGuard tunnel for VIP %s: %w", vip, err)
+	}
+
+	log.Info("[wireguard] brought up tunnel for service",
+		"namespace", svc.Namespace,
+		"name", svc.Name,
+		"vip", vip,
+		"interface", tunnelConfig.InterfaceName)
+
+	// DNAT rules will be configured by wireguardWorker.processInstance()
+	// when the endpoint watcher detects available endpoints
+
+	return nil
+}
+
+// deleteServiceWireguard removes nftables DNAT rules and tears down WireGuard tunnel for a service
+func (p *Processor) deleteServiceWireguard(_ context.Context, svc *v1.Service) {
+	if !p.config.EnableWireguard {
+		return
+	}
+
+	serviceID := fmt.Sprintf("%s_%s", svc.Namespace, svc.Name)
+	serviceID = utils.SanitizeServiceID(serviceID)
+
+	log.Info("[wireguard] deleting DNAT rules and tunnel for service",
+		"namespace", svc.Namespace,
+		"name", svc.Name,
+		"serviceID", serviceID)
+
+	// Get service IPs
+	serviceIPs, _ := utils.FetchServiceIPs(svc)
+
+	// Delete DNAT chains for each port
+	for _, port := range svc.Spec.Ports {
+		if port.Protocol != v1.ProtocolTCP && port.Protocol != v1.ProtocolUDP {
+			continue
+		}
+
+		portServiceID := fmt.Sprintf("%s_p%d", serviceID, port.Port)
+
+		// Try to delete for both IPv4 and IPv6 if we have mixed IPs
+		hasIPv4 := false
+		hasIPv6 := false
+		for _, vip := range serviceIPs {
+			// Strip CIDR notation before checking IP version
+			addr := utils.StripCIDR(vip)
+			if utils.IsIPv6(addr) {
+				hasIPv6 = true
+			} else {
+				hasIPv4 = true
+			}
+		}
+
+		if hasIPv4 {
+			if err := nftables.DeleteIngressChains(false, portServiceID); err != nil {
+				log.Error("[wireguard] failed to delete IPv4 DNAT chains",
+					"service", svc.Name,
+					"port", port.Port,
+					"err", err)
+			} else {
+				log.Debug("[wireguard] deleted IPv4 DNAT chains",
+					"service", svc.Name,
+					"port", port.Port)
+			}
+		}
+
+		if hasIPv6 {
+			if err := nftables.DeleteIngressChains(true, portServiceID); err != nil {
+				log.Error("[wireguard] failed to delete IPv6 DNAT chains",
+					"service", svc.Name,
+					"port", port.Port,
+					"err", err)
+			} else {
+				log.Debug("[wireguard] deleted IPv6 DNAT chains",
+					"service", svc.Name,
+					"port", port.Port)
+			}
+		}
+	}
+
+	// Tear down the WireGuard tunnel for each VIP
+	for _, vip := range serviceIPs {
+		if err := p.TunnelMgr.TearDownTunnelForVIP(vip); err != nil {
+			log.Error("[wireguard] failed to tear down tunnel",
+				"service", svc.Name,
+				"vip", vip,
+				"err", err)
+		} else {
+			log.Info("[wireguard] tore down tunnel",
+				"service", svc.Name,
+				"vip", vip)
+		}
+	}
+
+	log.Info("[wireguard] DNAT rules deleted and tunnels torn down for service",
+		"namespace", svc.Namespace,
+		"name", svc.Name)
+}

pkg/services/watch_endpoints.go

@@ -1,8 +1,8 @@
 package services
 
 import (
-	"context"
 	"fmt"
+	"sync"
 
 	log "log/slog"
 
@@ -18,46 +18,35 @@ func (p *Processor) watchEndpoint(svcCtx *servicecontext.Context, id string, ser
 	log.Info("watching", "provider", provider.GetLabel(), "service_name", service.Name, "namespace", service.Namespace)
 	// Use a restartable watcher, as this should help in the event of etcd or timeout issues
 
-	leaderCtx, cancel := context.WithCancel(svcCtx.Ctx)
-	defer cancel()
-
-	var leaderElectionActive bool
-
-	rw, err := provider.CreateRetryWatcher(leaderCtx, p.rwClientSet, service)
+	rw, err := provider.CreateRetryWatcher(svcCtx.Ctx, p.rwClientSet, service)
 	if err != nil {
 		return fmt.Errorf("[%s] error watching endpoints: %w", provider.GetLabel(), err)
 	}
 
-	exitFunction := make(chan struct{})
-	go func() {
+	wg := sync.WaitGroup{}
+
+	stopChan := make(chan any)
+
+	defer func() {
+		close(stopChan)
+		rw.Stop()
+		wg.Wait()
+	}()
+
+	wg.Go(func() {
 		select {
 		case <-svcCtx.Ctx.Done():
-			log.Debug("context cancelled", "provider", provider.GetLabel())
-			// Stop the retry watcher
+			log.Debug("[endpoint watcher] context cancelled", "provider", provider.GetLabel())
 			rw.Stop()
-			// Cancel the context, which will in turn cancel the leadership
-			cancel()
-			return
-		case <-p.shutdownChan:
-			log.Debug("shutdown called", "provider", provider.GetLabel())
-			// Stop the retry watcher
-			rw.Stop()
-			// Cancel the context, which will in turn cancel the leadership
-			cancel()
-			return
-		case <-exitFunction:
-			log.Debug("function ending", "provider", provider.GetLabel())
-			// Stop the retry watcher
-			rw.Stop()
-			// Cancel the context, which will in turn cancel the leadership
-			cancel()
-			return
+		case <-stopChan:
+			svcCtx.Cancel()
+			log.Debug("[endpoint watcher] exiting endpoint watcher", "namespace", service.Namespace, "service", service.Name, "provider", provider.GetLabel())
 		}
-	}()
+	})
 
 	ch := rw.ResultChan()
 
-	epProcessor := endpoints.NewEndpointProcessor(p.config, provider, p.bgpServer, &p.ServiceInstances)
+	epProcessor := endpoints.NewEndpointProcessor(p.config, provider, p.bgpServer, &p.ServiceInstances, p.leaseMgr, p.TunnelMgr, p.routeMgr)
 
 	var lastKnownGoodEndpoint string
 	for event := range ch {
@@ -65,7 +54,8 @@ func (p *Processor) watchEndpoint(svcCtx *servicecontext.Context, id string, ser
 		switch event.Type {
 
 		case watch.Added, watch.Modified:
-			restart, err := epProcessor.AddOrModify(svcCtx, event, &lastKnownGoodEndpoint, service, id, &leaderElectionActive, p.StartServicesLeaderElection, &leaderCtx, &cancel)
+			restart, err := epProcessor.AddOrModify(svcCtx, event, &lastKnownGoodEndpoint, service, id,
+				p.StartServicesLeaderElection, &wg, p.clientSet, p.updateEgressConfiguration)
 			if restart {
 				continue
 			} else if err != nil {
@@ -73,14 +63,11 @@ func (p *Processor) watchEndpoint(svcCtx *servicecontext.Context, id string, ser
 			}
 
 		case watch.Deleted:
-			if err := epProcessor.Delete(service, id); err != nil {
+			if err := epProcessor.Delete(svcCtx.Ctx, service, id); err != nil {
 				return fmt.Errorf("[%s] error while processing delete event: %w", provider.GetLabel(), err)
 			}
 
-			// Close the goroutine that will end the retry watcher, then exit the endpoint watcher function
-			close(exitFunction)
-			log.Info("stopping watching", "provider", provider.GetLabel(), "service name", service.Name, "namespace", service.Namespace)
-
+			log.Info("[endpoint watcher] stopping watching - endpoint object deleted", "provider", provider.GetLabel(), "service name", service.Name, "namespace", service.Namespace)
 			return nil
 		case watch.Error:
 			errObject := apierrors.FromObject(event.Object)
@@ -88,7 +75,6 @@ func (p *Processor) watchEndpoint(svcCtx *servicecontext.Context, id string, ser
 			log.Error("watch error", "provider", provider.GetLabel(), "err", statusErr)
 		}
 	}
-	close(exitFunction)
-	log.Info("stopping watching", "provider", provider.GetLabel(), "service name", service.Name, "namespace", service.Namespace)
+	log.Info("[endpoint watcher] stopping watching", "provider", provider.GetLabel(), "service name", service.Name, "namespace", service.Namespace)
 	return nil //nolint:govet
 }

pkg/services/watch_services.go

@@ -2,13 +2,17 @@ package services
 
 import (
 	"context"
+	"errors"
 	"fmt"
+	"sync"
 
 	log "log/slog"
 
 	"github.com/davecgh/go-spew/spew"
 	"github.com/kube-vip/kube-vip/pkg/kubevip"
+	"github.com/kube-vip/kube-vip/pkg/metrics"
 	"github.com/kube-vip/kube-vip/pkg/trafficmirror"
+	"github.com/kube-vip/kube-vip/pkg/utils"
 	"github.com/prometheus/client_golang/prometheus"
 	v1 "k8s.io/api/core/v1"
 	apierrors "k8s.io/apimachinery/pkg/api/errors"
@@ -19,7 +23,7 @@ import (
 )
 
 // This function handles the watching of a services endpoints and updates a load balancers endpoint configurations accordingly
-func (p *Processor) ServicesWatcher(ctx context.Context, serviceFunc func(context.Context, *v1.Service) error) error {
+func (p *Processor) ServicesWatcher(ctx context.Context, serviceFunc *Callback) error {
 	// first start port mirroring if enabled
 	if err := p.startTrafficMirroringIfEnabled(); err != nil {
 		return err
@@ -43,50 +47,50 @@ func (p *Processor) ServicesWatcher(ctx context.Context, serviceFunc func(contex
 	// Use a restartable watcher, as this should help in the event of etcd or timeout issues
 	rw, err := watchtools.NewRetryWatcherWithContext(ctx, "1", &cache.ListWatch{
 		WatchFunc: func(_ metav1.ListOptions) (watch.Interface, error) {
-			return p.rwClientSet.CoreV1().Services(p.config.ServiceNamespace).Watch(ctx, metav1.ListOptions{})
+			return utils.WatchWithAuthRetry(ctx, func(ctx context.Context) (watch.Interface, error) {
+				return p.rwClientSet.CoreV1().Services(p.config.ServiceNamespace).Watch(ctx, metav1.ListOptions{})
+			})
 		},
 	})
 	if err != nil {
 		return fmt.Errorf("error creating services watcher: %s", err.Error())
 	}
-	exitFunction := make(chan struct{})
-	go func() {
-		select {
-		case <-p.shutdownChan:
-			log.Debug("(svcs) shutdown called")
-			// Stop the retry watcher
-			rw.Stop()
-			return
-		case <-exitFunction:
-			log.Debug("(svcs) function ending")
-			// Stop the retry watcher
-			rw.Stop()
-			return
-		}
-	}()
+
+	wg := sync.WaitGroup{}
+	defer wg.Wait()
+
+	watcherCtx, watcherCancel := context.WithCancel(ctx)
+	defer watcherCancel()
+
+	wg.Go(func() {
+		<-watcherCtx.Done()
+		log.Debug("(svcs) watcher context cancelled")
+		rw.Stop()
+		p.Stop()
+	})
 	ch := rw.ResultChan()
 
 	// Used for tracking an active endpoint / pod
 	for event := range ch {
-		p.CountServiceWatchEvent.With(prometheus.Labels{"type": string(event.Type)}).Add(1)
+		metrics.CountServiceWatchEvent.With(prometheus.Labels{"type": string(event.Type)}).Add(1)
 
 		// We need to inspect the event and get ResourceVersion out of it
 		switch event.Type {
 		case watch.Added, watch.Modified:
-			restart, err := p.AddOrModify(ctx, event, serviceFunc)
-			if restart {
-				break
-			}
-			if err != nil {
-				return fmt.Errorf("add/modify service error: %w", err)
+			if err := p.AddOrModify(watcherCtx, event, serviceFunc, &wg); err != nil {
+				if errors.Is(err, &utils.PanicError{}) {
+					return fmt.Errorf("add/modify service error: %w", err)
+				} else {
+					log.Error("service watcher event failed", "type", event.Type, "error", err)
+				}
 			}
 		case watch.Deleted:
-			restart, err := p.Delete(event)
-			if restart {
-				break
-			}
-			if err != nil {
-				return fmt.Errorf("delete service error: %w", err)
+			if err := p.Delete(event); err != nil {
+				if errors.Is(err, &utils.PanicError{}) {
+					return fmt.Errorf("delete service error: %w", err)
+				} else {
+					log.Error("service watcher event failed", "type", event.Type, "error", err)
+				}
 			}
 		case watch.Bookmark:
 			// Un-used
@@ -105,7 +109,7 @@ func (p *Processor) ServicesWatcher(ctx context.Context, serviceFunc func(contex
 		default:
 		}
 	}
-	close(exitFunction)
+
 	log.Warn("Stopping watching services for type: LoadBalancer in all namespaces")
 	return nil
 }

pkg/services/watch_services_test.go

@@ -0,0 +1,99 @@
+package services
+
+import (
+	"context"
+	"fmt"
+	"testing"
+	"time"
+
+	"github.com/kube-vip/kube-vip/pkg/utils"
+	apierrors "k8s.io/apimachinery/pkg/api/errors"
+	"k8s.io/apimachinery/pkg/runtime/schema"
+	"k8s.io/apimachinery/pkg/watch"
+)
+
+// fakeWatchInterface is a minimal watch.Interface for testing.
+type fakeWatchInterface struct {
+	ch chan watch.Event
+}
+
+func newFakeWatchInterface() *fakeWatchInterface {
+	return &fakeWatchInterface{ch: make(chan watch.Event)}
+}
+
+func (f *fakeWatchInterface) Stop()                          { close(f.ch) }
+func (f *fakeWatchInterface) ResultChan() <-chan watch.Event { return f.ch }
+
+func TestWatchWithAuthRetry(t *testing.T) {
+	svcResource := schema.GroupResource{Resource: "services"}
+	fw := newFakeWatchInterface()
+
+	tcs := []struct {
+		name         string
+		watchFn      func(int) (watch.Interface, error)
+		wantErr      bool
+		wantAttempts int
+	}{
+		{
+			name: "403 Forbidden retried then succeeds",
+			watchFn: func(attempt int) (watch.Interface, error) {
+				if attempt <= 2 {
+					return nil, apierrors.NewForbidden(svcResource, "", nil)
+				}
+				return fw, nil
+			},
+			wantAttempts: 3,
+		},
+		{
+			name: "401 Unauthorized retried then succeeds",
+			watchFn: func(attempt int) (watch.Interface, error) {
+				if attempt <= 2 {
+					return nil, apierrors.NewUnauthorized("not authorized yet")
+				}
+				return fw, nil
+			},
+			wantAttempts: 3,
+		},
+		{
+			name: "non-auth error fails immediately",
+			watchFn: func(_ int) (watch.Interface, error) {
+				return nil, fmt.Errorf("connection refused")
+			},
+			wantErr:      true,
+			wantAttempts: 1,
+		},
+		{
+			name: "immediate success no retry",
+			watchFn: func(_ int) (watch.Interface, error) {
+				return fw, nil
+			},
+			wantAttempts: 1,
+		},
+	}
+
+	for _, tc := range tcs {
+		t.Run(tc.name, func(t *testing.T) {
+			attempts := 0
+			ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+			defer cancel()
+
+			w, err := utils.WatchWithAuthRetry(ctx, func(_ context.Context) (watch.Interface, error) {
+				attempts++
+				return tc.watchFn(attempts)
+			})
+
+			if tc.wantErr && err == nil {
+				t.Fatal("expected error, got nil")
+			}
+			if !tc.wantErr && err != nil {
+				t.Fatalf("expected success, got: %v", err)
+			}
+			if !tc.wantErr && w != fw {
+				t.Fatal("returned watcher does not match expected")
+			}
+			if attempts != tc.wantAttempts {
+				t.Errorf("expected %d attempts, got %d", tc.wantAttempts, attempts)
+			}
+		})
+	}
+}

pkg/utils/dns.go

@@ -2,6 +2,7 @@ package utils
 
 import (
 	"fmt"
+	"log/slog"
 	"net"
 	"strings"
 
@@ -15,7 +16,7 @@ const (
 )
 
 // LookupHost resolves dnsName and return an IP or an error
-func LookupHost(dnsName, dnsMode string) ([]string, error) {
+func LookupHost(dnsName, dnsMode string, requireDualStack bool) ([]string, error) {
 	result, err := net.LookupHost(dnsName)
 	if err != nil {
 		return nil, err
@@ -26,7 +27,7 @@ func LookupHost(dnsName, dnsMode string) ([]string, error) {
 	addrs := []string{}
 	switch dnsMode {
 	case strings.ToLower(IPv4Family), strings.ToLower(IPv6Family), DualFamily:
-		a, err := getIPbyFamily(result, dnsMode)
+		a, err := getIPbyFamily(result, dnsMode, requireDualStack)
 		if err != nil {
 			return nil, err
 		}
@@ -38,7 +39,7 @@ func LookupHost(dnsName, dnsMode string) ([]string, error) {
 	return addrs, nil
 }
 
-func getIPbyFamily(addresses []string, family string) ([]string, error) {
+func getIPbyFamily(addresses []string, family string, requireDualStack bool) ([]string, error) {
 	var checkers []func(string) bool
 	families := []string{}
 	if family == DualFamily || family == strings.ToLower(IPv4Family) {
@@ -54,11 +55,19 @@ func getIPbyFamily(addresses []string, family string) ([]string, error) {
 	for i, c := range checkers {
 		addr, err := getIPbyChecker(addresses, c)
 		if err != nil {
+			if len(checkers) > 1 && !requireDualStack {
+				slog.Warn("no address found", "family", families[i])
+				continue
+			}
 			return nil, fmt.Errorf("error getting %s address: %w", families[i], err)
 		}
 		addrs = append(addrs, addr)
 	}
 
+	if len(addrs) == 0 {
+		return nil, fmt.Errorf("no addresses found")
+	}
+
 	return addrs, nil
 }
 

pkg/utils/ip.go

@@ -1,8 +1,12 @@
 package utils
 
 import (
+	"encoding/hex"
 	"fmt"
 	"net"
+	"strings"
+
+	v1 "k8s.io/api/core/v1"
 )
 
 // FormatIPWithSubnetMask takes a raw IP address and a subnet mask, and returns a formatted string in CIDR notation.
@@ -56,3 +60,83 @@ func IsIPv6CIDR(cidr string) bool {
 	}
 	return ip.To4() == nil
 }
+
+// StripCIDR removes the CIDR notation (e.g., "/24") from an IP address string.
+// If no CIDR notation is present, the original string is returned unchanged.
+func StripCIDR(ip string) string {
+	if idx := strings.Index(ip, "/"); idx >= 0 {
+		return ip[:idx]
+	}
+	return ip
+}
+
+// SanitizeServiceID sanitizes a service ID to be valid for nftables chain names.
+// Only alphanumeric characters and underscores are allowed; other characters are replaced with underscores.
+// The result is truncated to 50 characters to respect nftables name length limits.
+func SanitizeServiceID(id string) string {
+	var result strings.Builder
+	for _, r := range id {
+		if (r >= 'a' && r <= 'z') || (r >= 'A' && r <= 'Z') || (r >= '0' && r <= '9') || r == '_' {
+			result.WriteRune(r)
+		} else {
+			result.WriteRune('_')
+		}
+	}
+	sanitized := result.String()
+
+	// Ensure it doesn't exceed nftables name length limit
+	if len(sanitized) > 50 {
+		sanitized = sanitized[:50]
+	}
+
+	return sanitized
+}
+
+// FetchServiceIPs extracts IP addresses from a Kubernetes Service.
+// It checks the following sources in order:
+// 1. kube-vip.io/loadbalancerIPs annotation (comma-separated list)
+// 2. spec.LoadBalancerIP (deprecated but still used)
+// 3. status.loadBalancer.ingress
+// Returns an error if no IPs are found.
+func FetchServiceIPs(service *v1.Service) ([]string, error) {
+	var ips []string
+
+	// Check for loadBalancerIPs annotation first (new style)
+	if loadBalancerIPs, ok := service.Annotations["kube-vip.io/loadbalancerIPs"]; ok {
+		for _, ip := range strings.Split(loadBalancerIPs, ",") {
+			ip = strings.TrimSpace(ip)
+			if ip != "" {
+				ips = append(ips, ip)
+			}
+		}
+	}
+
+	// Fallback to spec.LoadBalancerIP (deprecated but still used)
+	if len(ips) == 0 && service.Spec.LoadBalancerIP != "" {
+		ips = append(ips, service.Spec.LoadBalancerIP)
+	}
+
+	// Check status.loadBalancer.ingress as well
+	if len(ips) == 0 {
+		for _, ingress := range service.Status.LoadBalancer.Ingress {
+			if ingress.IP != "" {
+				ips = append(ips, ingress.IP)
+			}
+		}
+	}
+
+	if len(ips) == 0 {
+		return nil, fmt.Errorf("no IPs found for service")
+	}
+
+	return ips, nil
+}
+
+// Helper function to convert IPv6 hex address without colons
+func SanitizeIPForLabel(addr string) string {
+	ip := net.ParseIP(addr)
+	if ip == nil || ip.To4() != nil {
+		return addr
+	}
+	return hex.EncodeToString(ip.To16())
+}

pkg/utils/panic.go

@@ -0,0 +1,15 @@
+package utils
+
+import "fmt"
+
+type PanicError struct {
+	cause string
+}
+
+func (e *PanicError) Error() string {
+	return fmt.Sprintf("%s - unrecoverable error", e.cause)
+}
+
+func NewPanicError(cause string) error {
+	return &PanicError{cause: cause}
+}