Merge branch 'kubeovn:master' into bugfix

.github/workflows/build-x86-image.yaml

@@ -21,10 +21,10 @@ concurrency:
   cancel-in-progress: true
 
 env:
-  KIND_VERSION: v0.24.0
-  GOLANGCI_LINT_VERSION: 'v1.61.0'
+  KIND_VERSION: v0.25.0
+  GOLANGCI_LINT_VERSION: 'v1.62.0'
   HELM_VERSION: v3.16.2
-  SUBMARINER_VERSION: '0.18.1'
+  SUBMARINER_VERSION: '0.18.2'
 
 jobs:
   build-kube-ovn-base:
@@ -2844,6 +2844,118 @@ jobs:
         if: ${{ success() || (failure() && (steps.install.conclusion == 'failure' || steps.kube-ovn-ipsec-e2e.conclusion == 'failure')) }}
         run: make check-kube-ovn-pod-restarts
 
+  kube-ovn-connectivity-test:
+    name: Kube-OVN Connectivity E2E
+    needs:
+      - build-kube-ovn
+      - build-e2e-binaries
+    runs-on: ubuntu-24.04
+    timeout-minutes: 15
+    strategy:
+      fail-fast: false
+      matrix:
+        mode:
+          - overlay
+          - underlay
+    steps:
+      - uses: jlumbroso/[email protected]
+        with:
+          android: true
+          dotnet: true
+          haskell: true
+          docker-images: false
+          large-packages: false
+          tool-cache: false
+          swap-storage: false
+
+      - uses: actions/checkout@v4
+
+      - name: Create the default branch directory
+        if: (github.base_ref || github.ref_name) != github.event.repository.default_branch
+        run: mkdir -p test/e2e/source
+
+      - name: Check out the default branch
+        if: (github.base_ref || github.ref_name) != github.event.repository.default_branch
+        uses: actions/checkout@v4
+        with:
+          ref: ${{ github.event.repository.default_branch }}
+          fetch-depth: 1
+          path: test/e2e/source
+
+      - name: Export E2E directory
+        run: |
+          if [ '${{ github.base_ref || github.ref_name }}' = '${{ github.event.repository.default_branch }}' ]; then
+            echo "E2E_DIR=." >> "$GITHUB_ENV"
+          else
+            echo "E2E_DIR=test/e2e/source" >> "$GITHUB_ENV"
+          fi
+
+      - uses: actions/setup-go@v5
+        id: setup-go
+        with:
+          go-version-file: ${{ env.E2E_DIR }}/go.mod
+          check-latest: true
+          cache: false
+
+      - name: Export Go full version
+        run: echo "GO_VERSION=${{ steps.setup-go.outputs.go-version }}" >> "$GITHUB_ENV"
+
+      - name: Go cache
+        uses: actions/cache/restore@v4
+        with:
+          path: |
+            ~/.cache/go-build
+            ~/go/pkg/mod
+          key: ${{ runner.os }}-e2e-go-${{ env.GO_VERSION }}-x86-${{ hashFiles(format('{0}/**/go.sum', env.E2E_DIR)) }}
+          restore-keys: ${{ runner.os }}-e2e-go-${{ env.GO_VERSION }}-x86-
+
+      - name: Install kind
+        uses: helm/[email protected]
+        with:
+          version: ${{ env.KIND_VERSION }}
+          install_only: true
+
+      - name: Install ginkgo
+        working-directory: ${{ env.E2E_DIR }}
+        run: go install -v -mod=mod github.com/onsi/ginkgo/v2/ginkgo
+
+      - name: Download kube-ovn image
+        uses: actions/download-artifact@v4
+        with:
+          name: kube-ovn
+
+      - name: Load images
+        run: docker load -i kube-ovn.tar
+
+      - name: Create kind cluster
+        run: |
+          pipx install jinjanator
+          make kind-init
+
+      - name: Install Kube-OVN
+        id: install
+        run: make kind-install-${{ matrix.mode }}
+
+      - name: Run E2E
+        id: e2e
+        working-directory: ${{ env.E2E_DIR }}
+        env:
+          E2E_BRANCH: ${{ github.base_ref || github.ref_name }}
+        run: make kube-ovn-connectivity-e2e
+
+      - name: kubectl ko log
+        if: failure() && (steps.e2e.conclusion == 'failure')
+        run: |
+          make kubectl-ko-log
+          mv kubectl-ko-log.tar.gz kube-ovn-connectivity-e2e-${{ matrix.mode }}-ko-log.tar.gz
+
+      - name: upload kubectl ko log
+        uses: actions/upload-artifact@v4
+        if: failure() && (steps.kube-ovn-connectivity-e2e.conclusion == 'failure')
+        with:
+          name: kube-ovn-connectivity-e2e-${{ matrix.mode }}-ko-log
+          path: kube-ovn-connectivity-e2e-${{ matrix.mode }}-ko-log.tar.gz
+
   push:
     name: Push Images
     needs:

Makefile

@@ -30,7 +30,7 @@ CONTROL_PLANE_TAINTS = node-role.kubernetes.io/master node-role.kubernetes.io/co
 FRR_VERSION = 9.1.2
 FRR_IMAGE = quay.io/frrouting/frr:$(FRR_VERSION)
 
-CLAB_IMAGE = ghcr.io/srl-labs/clab:0.58.0
+CLAB_IMAGE = ghcr.io/srl-labs/clab:0.59.0
 
 MULTUS_VERSION = v4.1.3
 MULTUS_IMAGE = ghcr.io/k8snetworkplumbingwg/multus-cni:$(MULTUS_VERSION)-thick
@@ -59,7 +59,7 @@ CERT_MANAGER_CAINJECTOR = quay.io/jetstack/cert-manager-cainjector:$(CERT_MANAGE
 CERT_MANAGER_WEBHOOK = quay.io/jetstack/cert-manager-webhook:$(CERT_MANAGER_VERSION)
 CERT_MANAGER_YAML = https://github.com/cert-manager/cert-manager/releases/download/$(CERT_MANAGER_VERSION)/cert-manager.yaml
 
-SUBMARINER_VERSION = $(shell echo $${SUBMARINER_VERSION:-0.18.1})
+SUBMARINER_VERSION = $(shell echo $${SUBMARINER_VERSION:-0.18.2})
 SUBMARINER_OPERATOR = quay.io/submariner/submariner-operator:$(SUBMARINER_VERSION)
 SUBMARINER_GATEWAY = quay.io/submariner/submariner-gateway:$(SUBMARINER_VERSION)
 SUBMARINER_LIGHTHOUSE_AGENT = quay.io/submariner/lighthouse-agent:$(SUBMARINER_VERSION)
@@ -78,7 +78,7 @@ DEEPFLOW_GRAFANA_NODE_PORT = 30080
 DEEPFLOW_MAPPED_PORTS = $(DEEPFLOW_SERVER_NODE_PORT),$(DEEPFLOW_SERVER_GRPC_PORT),$(DEEPFLOW_SERVER_HTTP_PORT),$(DEEPFLOW_GRAFANA_NODE_PORT)
 DEEPFLOW_CTL_URL = https://deepflow-ce.oss-cn-beijing.aliyuncs.com/bin/ctl/$(DEEPFLOW_VERSION)/linux/$(shell arch | sed 's|x86_64|amd64|' | sed 's|aarch64|arm64|')/deepflow-ctl
 
-KWOK_VERSION = v0.6.0
+KWOK_VERSION = v0.6.1
 KWOK_IMAGE = registry.k8s.io/kwok/kwok:$(KWOK_VERSION)
 
 VPC_NAT_GW_IMG = $(REGISTRY)/vpc-nat-gateway:$(VERSION)
@@ -559,6 +559,9 @@ kind-install-ipv6:
 kind-install-dual:
 	@DUAL_STACK=true $(MAKE) kind-install
 
+.PHONY: kind-install-overlay
+kind-install-overlay: kind-install-overlay-ipv4
+
 .PHONY: kind-install-overlay-%
 kind-install-overlay-%:
 	@$(MAKE) kind-install-$*

dist/images/Dockerfile.base

@@ -37,29 +37,29 @@ RUN cd /usr/src/ && \
 RUN cd /usr/src/ && git clone -b branch-24.03 --depth=1 https://github.com/ovn-org/ovn.git && \
     cd ovn && \
     # change hash type from dp_hash to hash with field src_ip
-    curl -s https://github.com/kubeovn/ovn/commit/4b2d0bffda64fcfca876fe62a02486d7927c1162.patch | git apply && \
+    curl -s https://github.com/kubeovn/ovn/commit/61ece99b63e88bb12c030c34b46716b01e242c92.patch | git apply && \
     # modify src route priority
-    curl -s https://github.com/kubeovn/ovn/commit/3dbceb75c5e23d8d5b555e5d87238d40cb557424.patch | git apply && \
+    curl -s https://github.com/kubeovn/ovn/commit/cecf1515de299b3ec6b30ed8a167748e879188a6.patch | git apply && \
     # fix reaching resubmit limit in underlay
-    curl -s https://github.com/kubeovn/ovn/commit/37d093c99c4980696cf917a1ad47492881908c90.patch | git apply && \
+    curl -s https://github.com/kubeovn/ovn/commit/eb935585bdeecdaa3fba8ff2c46ef394ce8d2424.patch | git apply && \
     # ovn-controller: do not send GARP on localnet for Kube-OVN ports
-    curl -s https://github.com/kubeovn/ovn/commit/35efb84d45d2598fc83fabaee8100987dcbc6795.patch | git apply && \
+    curl -s https://github.com/kubeovn/ovn/commit/8236ec78aedd114b3ce22158a9e0282337787259.patch | git apply && \
     # northd: add nb option version_compatibility
-    curl -s https://github.com/kubeovn/ovn/commit/ef92cdbc10ee247ea7db7803e86e1d51c9cdaddf.patch | git apply && \
+    curl -s https://github.com/kubeovn/ovn/commit/19bdf15ec5542005e3a1ef073dfe688c023d144c.patch | git apply && \
     # add support for conditionally skipping conntrack
-    curl -s https://github.com/kubeovn/ovn/commit/41b419e54ef5671214eb3b5dfb53301152ce6dd1.patch | git apply && \
+    curl -s https://github.com/kubeovn/ovn/commit/c6ea5e214a4a20eccdf8cd18c232284edc5ce951.patch | git apply && \
     # northd: skip conntrack when access node local dns ip
-    curl -s https://github.com/kubeovn/ovn/commit/b01a339c3b56eff45b22d414ed9d6e4a37527cd5.patch | git apply && \
+    curl -s https://github.com/kubeovn/ovn/commit/d8ce6dc8e512b703903c53e344b1dce1a0c61d2e.patch | git apply && \
     # lflow: do not send direct traffic between lports to conntrack
-    curl -s https://github.com/kubeovn/ovn/commit/5687a692ef4c10c6ec9bfeca7b920e7736a4016c.patch | git apply && \
+    curl -s https://github.com/kubeovn/ovn/commit/baf2d74f0f2d746c01c22002ea72426d75cfc79a.patch | git apply && \
     # set ether dst addr for dnat on logical switch
-    curl -s https://github.com/kubeovn/ovn/commit/08cfd7a4c56c610d6bf7dcf8be016f5ac7928f28.patch | git apply && \
+    curl -s https://github.com/kubeovn/ovn/commit/63762db7a4c37185f4c87962c50b414d925b5d33.patch | git apply && \
     # fix lr-lb dnat with multiple distributed gateway ports
-    curl -s https://github.com/kubeovn/ovn/commit/d326dd12b2ecc3b65f7360aac8e7c35c4bed2026.patch | git apply && \
+    curl -s https://github.com/kubeovn/ovn/commit/ca5ccf3c8ac67bcc005110c8c62214d2684e7e41.patch | git apply && \
     # northd: skip arp/nd request for lrp addresses from localnet ports
-    curl -s https://github.com/kubeovn/ovn/commit/cf6ba220851ae13de76473b6569097bb70ec76d8.patch | git apply && \
+    curl -s https://github.com/kubeovn/ovn/commit/283930b627ffa843ebf0e7c3fa0cc70edacfdd12.patch | git apply && \
     # ovn-controller: make activation strategy work for single chassis
-    curl -s https://github.com/kubeovn/ovn/commit/487cd1c6c0cc9def7cedaaaf3dd4bc99c23974c4.patch | git apply
+    curl -s https://github.com/kubeovn/ovn/commit/1160d956e49e8f3f1b19535dbf1b9a624a090717.patch | git apply
 
 RUN apt install -y build-essential fakeroot \
     autoconf automake bzip2 debhelper-compat dh-exec dh-python dh-sequence-python3 dh-sequence-sphinxdoc \

test/e2e/connectivity/e2e_test.go

@@ -72,7 +72,7 @@ func TestE2E(t *testing.T) {
 
 type suiteContext struct {
 	Node     string
-	HostIP   string
+	NodeIP   string
 	NodePort int32
 }
 
@@ -108,9 +108,28 @@ var _ = ginkgo.SynchronizedBeforeSuite(func() []byte {
 	framework.ExpectNoError(err)
 	framework.ExpectNotNil(pods)
 	framework.ExpectNotEmpty(pods.Items, "no pod found in deployment "+deploymentName)
-	suiteCtx.HostIP = pods.Items[0].Status.HostIP
+	suiteCtx.NodeIP = pods.Items[0].Status.HostIP
 	suiteCtx.Node = pods.Items[0].Spec.NodeName
 
+	ginkgo.By("Getting all nodes")
+	nodes, err := cs.CoreV1().Nodes().List(context.TODO(), metav1.ListOptions{})
+	framework.ExpectNoError(err)
+	framework.ExpectNotNil(nodes)
+	framework.ExpectNotEmpty(nodes.Items)
+
+	// use the internal IP of the node that is not the same as the pod node
+	for _, node := range nodes.Items {
+		if node.Name != suiteCtx.Node {
+			ipv4, ipv6 := util.GetNodeInternalIP(node)
+			if ipv4 != "" {
+				suiteCtx.NodeIP = ipv4
+			} else {
+				suiteCtx.NodeIP = ipv6
+			}
+			break
+		}
+	}
+
 	ginkgo.By("Creating service " + serviceName)
 	ports := []corev1.ServicePort{{
 		Name:       "tcp",
@@ -157,13 +176,13 @@ var _ = framework.Describe("[group:connectivity]", func() {
 	})
 	ginkgo.AfterEach(ginkgo.OncePerOrdered, func() {
 		ginkgo.By("Closing RPC server")
-		framework.ExpectNoError(server.Close())
+		framework.ExpectNoError(server.Shutdown(context.Background()))
 	})
 
 	ginkgo.It("Continuous NodePort HTTP testing", func() {
 		u := url.URL{
 			Scheme: "http",
-			Host:   util.JoinHostPort(suiteCtx.HostIP, suiteCtx.NodePort),
+			Host:   util.JoinHostPort(suiteCtx.NodeIP, suiteCtx.NodePort),
 			Path:   "/clientip",
 		}
 		ginkgo.By("GET " + u.String())
@@ -263,7 +282,7 @@ var _ = framework.OrderedDescribe("[group:disaster]", func() {
 		framework.ExpectNotEmpty(pods.Items, "no pod found in deployment ovn-central")
 
 		for _, pod := range pods.Items {
-			framework.Logf("pod %s is running on %s", pod.Name, pod.Spec.NodeName)
+			ginkgo.By("Deleting pod " + pod.Name + " running on " + pod.Spec.NodeName)
 			err = cs.CoreV1().Pods(pod.Namespace).Delete(context.TODO(), pod.Name, metav1.DeleteOptions{})
 			framework.ExpectNoError(err, "failed to delete pod "+pod.Name)
 		}
@@ -281,6 +300,90 @@ var _ = framework.OrderedDescribe("[group:disaster]", func() {
 		}
 	})
 
+	framework.DisruptiveIt("Stop ovn sb process", func() {
+		ginkgo.By("Getting deployment ovn-central")
+		deploymentClient := framework.NewDeploymentClient(cs, framework.KubeOvnNamespace)
+		deploy := deploymentClient.Get("ovn-central")
+
+		ginkgo.By("Getting pods of deployment ovn-central")
+		pods, err := deploymentClient.GetPods(deploy)
+		framework.ExpectNoError(err)
+		framework.ExpectNotEmpty(pods.Items, "no pod found in deployment ovn-central")
+
+		for _, pod := range pods.Items {
+			ginkgo.By("Getting ovn sb pid of pod " + pod.Name + " running on " + pod.Spec.NodeName)
+			cmd := []string{"cat", "/run/ovn/ovnsb_db.pid"}
+			stdout, stderr, err := framework.KubectlExec(pod.Namespace, pod.Name, cmd...)
+			framework.ExpectNoError(err, "failed to get ovn sb pid: %v, %s", err, string(stderr))
+			pid := string(bytes.TrimSpace(stdout))
+			framework.Logf("ovn sb pid: %s", pid)
+
+			ginkgo.By("Stopping ovn sb process by sending a STOP signal")
+			cmd = []string{"sh", "-c", fmt.Sprintf(`"kill -STOP %s"`, pid)}
+			_, stderr, err = framework.KubectlExec(pod.Namespace, pod.Name, cmd...)
+			framework.ExpectNoError(err, "failed to send STOP signal to ovn sb process: %v, %s", err, string(stderr))
+		}
+
+		ginkgo.By("Waiting 60s")
+		time.Sleep(60 * time.Second)
+
+		for _, pod := range pods.Items {
+			ginkgo.By("Getting ovn sb pid of pod " + pod.Name + " running on " + pod.Spec.NodeName)
+			cmd := []string{"cat", "/run/ovn/ovnsb_db.pid"}
+			stdout, stderr, err := framework.KubectlExec(pod.Namespace, pod.Name, cmd...)
+			framework.ExpectNoError(err, "failed to get ovn sb pid: %v, %s", err, string(stderr))
+			pid := string(bytes.TrimSpace(stdout))
+			framework.Logf("ovn sb pid: %s", pid)
+
+			ginkgo.By("Stopping ovn sb process by sending a CONT signal")
+			cmd = []string{"sh", "-c", fmt.Sprintf(`"kill -CONT %s"`, pid)}
+			_, stderr, err = framework.KubectlExec(pod.Namespace, pod.Name, cmd...)
+			framework.ExpectNoError(err, "failed to send CONT signal to ovn sb process: %v, %s", err, string(stderr))
+		}
+	})
+
+	framework.DisruptiveIt("Stop ovn-controller process", func() {
+		ginkgo.By("Getting DaemonSet ovs-ovn")
+		dsClient := framework.NewDaemonSetClient(cs, framework.KubeOvnNamespace)
+		ds := dsClient.Get("ovs-ovn")
+
+		ginkgo.By("Getting pods of DaemonSet ovs-ovn")
+		pods, err := dsClient.GetPods(ds)
+		framework.ExpectNoError(err)
+		framework.ExpectNotEmpty(pods.Items, "no pod found in DaemonSet ovs-ovn")
+
+		ginkgo.By("Getting ovs-ovn pod running on node " + suiteCtx.Node)
+		var pod *corev1.Pod
+		for i := range pods.Items {
+			framework.Logf("pod %s is running on %s", pods.Items[i].Name, pods.Items[i].Spec.NodeName)
+			if pods.Items[i].Spec.NodeName == suiteCtx.Node {
+				pod = &pods.Items[i]
+				break
+			}
+		}
+		framework.ExpectNotNil(pod, "no ovs-ovn pod running on node "+suiteCtx.Node)
+
+		ginkgo.By("Getting ovn-controller pid")
+		cmd := []string{"sh", "-c", `"pidof -s ovn-controller"`}
+		stdout, stderr, err := framework.KubectlExec(pod.Namespace, pod.Name, cmd...)
+		framework.ExpectNoError(err, "failed to get ovn-controller pid: %v, %s", err, string(stderr))
+		pid := string(bytes.TrimSpace(stdout))
+		framework.Logf("ovn-controller pid: %s", pid)
+
+		ginkgo.By("Stopping ovn-controller process by sending a STOP signal")
+		cmd = []string{"sh", "-c", fmt.Sprintf(`"kill -STOP %s"`, pid)}
+		_, stderr, err = framework.KubectlExec(pod.Namespace, pod.Name, cmd...)
+		framework.ExpectNoError(err, "failed to send STOP signal to ovn-controller process: %v, %s", err, string(stderr))
+
+		ginkgo.By("Waiting 60s")
+		time.Sleep(60 * time.Second)
+
+		ginkgo.By("Continuing the stopped ovn-controller process by sending a CONT signal")
+		cmd = []string{"sh", "-c", fmt.Sprintf(`"kill -CONT %s"`, pid)}
+		_, stderr, err = framework.KubectlExec(pod.Namespace, pod.Name, cmd...)
+		framework.ExpectNoError(err, "failed to send CONT signal to ovn-controller process: %v, %s", err, string(stderr))
+	})
+
 	framework.DisruptiveIt("Stop ovs-vswitchd process", func() {
 		ginkgo.By("Getting DaemonSet ovs-ovn")
 		dsClient := framework.NewDaemonSetClient(cs, framework.KubeOvnNamespace)