Merge branch 'k8snetworkplumbingwg:master' into master

.github/workflows/test.yml

@@ -100,6 +100,16 @@ jobs:
         # Required: the version of golangci-lint is required and must be specified without patch version: we always use the latest patch version.
         version: v1.55.2
 
+  shellcheck:
+    name: Shellcheck
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - name: Run ShellCheck
+        uses: ludeeus/action-shellcheck@master
+        with:
+          severity: error
+
   test-coverage:
     name: test-coverage
     runs-on: ubuntu-latest

bindata/manifests/webhook/002-rbac.yaml

@@ -21,7 +21,7 @@ rules:
 - apiGroups:
     - ""
   resources:
-    - configmap
+    - configmaps
   verbs:
     - 'watch'
     - 'list'

hack/deploy-setup.sh

@@ -22,7 +22,7 @@ load_manifest() {
     fi
     files="service_account.yaml role.yaml role_binding.yaml clusterrole.yaml clusterrolebinding.yaml configmap.yaml sriovoperatorconfig.yaml operator.yaml"
     for m in ${files}; do
-      if [ "$(echo ${EXCLUSIONS[@]} | grep -o ${m} | wc -w | xargs)" == "0" ] ; then
+      if [ "$(echo "${EXCLUSIONS[@]}" | grep -o ${m} | wc -w | xargs)" == "0" ] ; then
         envsubst< ${m} | ${OPERATOR_EXEC} apply ${namespace:-} --validate=false -f -
       fi
     done

hack/env.sh

@@ -1,3 +1,5 @@
+#!/bin/bash
+
 if [ -z $SKIP_VAR_SET ]; then
         export SRIOV_CNI_IMAGE=${SRIOV_CNI_IMAGE:-ghcr.io/k8snetworkplumbingwg/sriov-cni}
         export SRIOV_INFINIBAND_CNI_IMAGE=${SRIOV_INFINIBAND_CNI_IMAGE:-ghcr.io/k8snetworkplumbingwg/ib-sriov-cni}

hack/run-e2e-test-kind.sh

@@ -6,7 +6,7 @@ export SRIOV_NETWORK_OPERATOR_IMAGE="${SRIOV_NETWORK_OPERATOR_IMAGE:-sriov-netwo
 export SRIOV_NETWORK_CONFIG_DAEMON_IMAGE="${SRIOV_NETWORK_CONFIG_DAEMON_IMAGE:-origin-sriov-network-config-daemon:e2e-test}"
 export KUBECONFIG="${KUBECONFIG:-${HOME}/.kube/config}"
 INTERFACES_SWITCHER="${INTERFACES_SWITCHER:-"test-suite"}"
-SUPPORTED_INTERFACE_SWTICHER_MODES=("test-suite", "system-service")
+SUPPORTED_INTERFACE_SWITCHER_MODES=("test-suite", "system-service")
 RETRY_MAX=10
 INTERVAL=10
 TIMEOUT=300
@@ -16,9 +16,9 @@ while test $# -gt 0; do
     case "$1" in
     --device-netns-switcher)
         INTERFACES_SWITCHER="$2"
-        if [[ ! "${SUPPORTED_INTERFACE_SWTICHER_MODES[@]}" =~ "${INTERFACES_SWITCHER}" ]]; then
+        if [[ ! "${SUPPORTED_INTERFACE_SWITCHER_MODES[*]}" =~ "${INTERFACES_SWITCHER}" ]]; then
             echo "Error: unsupported interface switching mode: ${INTERFACES_SWITCHER}!"
-            echo "Supported modes are: ${SUPPORTED_INTERFACE_SWTICHER_MODES[@]}"
+            echo "Supported modes are: ${SUPPORTED_INTERFACE_SWITCHER_MODES[*]}"
             exit 1
         fi
         shift

hack/vf-netns-switcher.sh

@@ -95,7 +95,7 @@ It must be of the form <netns>:<pf1>,<pf2>. This flag can be repeated to specify
 done
 
 return_interfaces_to_default_namespace(){
-    for netns in ${netnses[@]};do
+    for netns in "${netnses[@]}";do
         for pf in ${pfs[$netns]};do
             return_interface_to_default_namespace "${netns}" "${pf}"
         done
@@ -277,19 +277,20 @@ switch_interface_vf_representors(){
         return 0
     fi
 
-    for interface in $(ls /sys/class/net);do
-        phys_switch_id=$(cat /sys/class/net/$interface/phys_switch_id)
+    for interface in /sys/class/net/*;do
+        phys_switch_id=$(cat $interface/phys_switch_id)
         if [[ "$phys_switch_id" != "${pf_switch_ids[$pf_name]}" ]]; then
             continue
         fi
-        phys_port_name=$(cat /sys/class/net/$interface/phys_port_name)
+        phys_port_name=$(cat $interface/phys_port_name)
         phys_port_name_pf_index=${phys_port_name%vf*}
         phys_port_name_pf_index=${phys_port_name_pf_index#pf}
         if [[ "$phys_port_name_pf_index" != "${pf_port_names[$pf_name]:1}"  ]]; then
             continue
         fi
-        echo "Switching VF representor $interface of PF $pf_name to netns $worker_netns"
-        switch_vf $interface $worker_netns
+        interface_name=${interface##*/}
+        echo "Switching VF representor $interface_name of PF $pf_name to netns $worker_netns"
+        switch_vf $interface_name $worker_netns
     done
 }
 
@@ -348,7 +349,7 @@ variables_check(){
 check_empty_var(){
     local var_name="$1"
 
-    if [[ -z "${!var_name[@]}" ]];then
+    if [[ -z "${!var_name[*]}" ]];then
         echo "Error: $var_name is empty..."
         return 1
     fi
@@ -360,7 +361,7 @@ main(){
     trap return_interfaces_to_default_namespace INT EXIT TERM
 
     while true;do
-        for netns in ${netnses[@]};do
+        for netns in "${netnses[@]}";do
             switch_pfs "$netns" "${pfs[$netns]}"
             sleep 2
             switch_netns_vfs "$netns"
@@ -388,7 +389,7 @@ if [[ "$status" != "0" ]];then
     exit $status
 fi
 
-for netns in ${netnses[@]};do
+for netns in "${netnses[@]}";do
     netns_create "$netns"
     let status=$status+$?
     if [[ "$status" != "0" ]];then
@@ -397,13 +398,13 @@ for netns in ${netnses[@]};do
     fi
 done
 
-for netns in ${netnses[@]};do
+for netns in "${netnses[@]}";do
     get_pcis_from_pfs "$netns" "${pfs[$netns]}"
     get_pf_switch_dev_info "$netns" "${pfs[$netns]}"
 done
 
 if [[ "${#pcis[@]}" == "0" ]];then
-    echo "Error: could not get pci addresses of interfaces ${pfs[@]}!!"
+    echo "Error: could not get pci addresses of interfaces ${pfs[*]}!!"
     exit 1
 fi
 

pkg/daemon/daemon.go

@@ -120,6 +120,7 @@ func New(
 		eventRecorder:   er,
 		featureGate:     featureGates,
 		disabledPlugins: disabledPlugins,
+		mu:              &sync.Mutex{},
 	}
 }
 
@@ -159,7 +160,6 @@ func (dn *Daemon) Run(stopCh <-chan struct{}, exitCh <-chan error) error {
 
 	var timeout int64 = 5
 	var metadataKey = "metadata.name"
-	dn.mu = &sync.Mutex{}
 	informerFactory := sninformer.NewFilteredSharedInformerFactory(dn.sriovClient,
 		time.Second*15,
 		vars.Namespace,
@@ -683,7 +683,6 @@ func (dn *Daemon) restartDevicePluginPod() error {
 	defer dn.mu.Unlock()
 	log.Log.V(2).Info("restartDevicePluginPod(): try to restart device plugin pod")
 
-	var podToDelete string
 	pods, err := dn.kubeClient.CoreV1().Pods(vars.Namespace).List(context.Background(), metav1.ListOptions{
 		LabelSelector:   "app=sriov-device-plugin",
 		FieldSelector:   "spec.nodeName=" + vars.NodeName,
@@ -702,35 +701,37 @@ func (dn *Daemon) restartDevicePluginPod() error {
 		log.Log.Info("restartDevicePluginPod(): device plugin pod exited")
 		return nil
 	}
-	podToDelete = pods.Items[0].Name
 
-	log.Log.V(2).Info("restartDevicePluginPod(): Found device plugin pod, deleting it", "pod-name", podToDelete)
-	err = dn.kubeClient.CoreV1().Pods(vars.Namespace).Delete(context.Background(), podToDelete, metav1.DeleteOptions{})
-	if errors.IsNotFound(err) {
-		log.Log.Info("restartDevicePluginPod(): pod to delete not found")
-		return nil
-	}
-	if err != nil {
-		log.Log.Error(err, "restartDevicePluginPod(): Failed to delete device plugin pod, retrying")
-		return err
-	}
-
-	if err := wait.PollImmediateUntil(3*time.Second, func() (bool, error) {
-		_, err := dn.kubeClient.CoreV1().Pods(vars.Namespace).Get(context.Background(), podToDelete, metav1.GetOptions{})
+	for _, pod := range pods.Items {
+		podToDelete := pod.Name
+		log.Log.V(2).Info("restartDevicePluginPod(): Found device plugin pod, deleting it", "pod-name", podToDelete)
+		err = dn.kubeClient.CoreV1().Pods(vars.Namespace).Delete(context.Background(), podToDelete, metav1.DeleteOptions{})
 		if errors.IsNotFound(err) {
-			log.Log.Info("restartDevicePluginPod(): device plugin pod exited")
-			return true, nil
+			log.Log.Info("restartDevicePluginPod(): pod to delete not found")
+			continue
 		}
-
 		if err != nil {
-			log.Log.Error(err, "restartDevicePluginPod(): Failed to check for device plugin exit, retrying")
-		} else {
-			log.Log.Info("restartDevicePluginPod(): waiting for device plugin pod to exit", "pod-name", podToDelete)
+			log.Log.Error(err, "restartDevicePluginPod(): Failed to delete device plugin pod, retrying")
+			return err
+		}
+
+		if err := wait.PollImmediateUntil(3*time.Second, func() (bool, error) {
+			_, err := dn.kubeClient.CoreV1().Pods(vars.Namespace).Get(context.Background(), podToDelete, metav1.GetOptions{})
+			if errors.IsNotFound(err) {
+				log.Log.Info("restartDevicePluginPod(): device plugin pod exited")
+				return true, nil
+			}
+
+			if err != nil {
+				log.Log.Error(err, "restartDevicePluginPod(): Failed to check for device plugin exit, retrying")
+			} else {
+				log.Log.Info("restartDevicePluginPod(): waiting for device plugin pod to exit", "pod-name", podToDelete)
+			}
+			return false, nil
+		}, dn.stopCh); err != nil {
+			log.Log.Error(err, "restartDevicePluginPod(): failed to wait for checking pod deletion")
+			return err
 		}
-		return false, nil
-	}, dn.stopCh); err != nil {
-		log.Log.Error(err, "restartDevicePluginPod(): failed to wait for checking pod deletion")
-		return err
 	}
 
 	return nil

pkg/daemon/daemon_test.go

@@ -32,6 +32,8 @@ import (
 	"github.com/k8snetworkplumbingwg/sriov-network-operator/test/util/fakefilesystem"
 )
 
+var SriovDevicePluginPod corev1.Pod
+
 func TestConfigDaemon(t *testing.T) {
 	RegisterFailHandler(Fail)
 	RunSpecs(t, "Config Daemon Suite")
@@ -107,19 +109,6 @@ var _ = Describe("Config Daemon", func() {
 			},
 		}
 
-		SriovDevicePluginPod := corev1.Pod{
-			ObjectMeta: metav1.ObjectMeta{
-				Name:      "sriov-device-plugin-xxxx",
-				Namespace: vars.Namespace,
-				Labels: map[string]string{
-					"app": "sriov-device-plugin",
-				},
-			},
-			Spec: corev1.PodSpec{
-				NodeName: "test-node",
-			},
-		}
-
 		err = sriovnetworkv1.AddToScheme(scheme.Scheme)
 		Expect(err).ToNot(HaveOccurred())
 		kClient := kclient.NewClientBuilder().WithScheme(scheme.Scheme).WithRuntimeObjects(&corev1.Node{
@@ -130,7 +119,7 @@ var _ = Describe("Config Daemon", func() {
 					Namespace: vars.Namespace,
 				}}).Build()
 
-		kubeClient := fakek8s.NewSimpleClientset(&FakeSupportedNicIDs, &SriovDevicePluginPod)
+		kubeClient := fakek8s.NewSimpleClientset(&FakeSupportedNicIDs)
 		snclient := snclientset.NewSimpleClientset()
 		err = sriovnetworkv1.InitNicIDMapFromConfigMap(kubeClient, vars.Namespace)
 		Expect(err).ToNot(HaveOccurred())
@@ -175,6 +164,22 @@ var _ = Describe("Config Daemon", func() {
 			err := sut.Run(stopCh, exitCh)
 			Expect(err).ToNot(HaveOccurred())
 		}()
+
+		SriovDevicePluginPod = corev1.Pod{
+			ObjectMeta: metav1.ObjectMeta{
+				Name:      "sriov-device-plugin-xxxx",
+				Namespace: vars.Namespace,
+				Labels: map[string]string{
+					"app": "sriov-device-plugin",
+				},
+			},
+			Spec: corev1.PodSpec{
+				NodeName: "test-node",
+			},
+		}
+		_, err = sut.kubeClient.CoreV1().Pods(vars.Namespace).Create(context.Background(), &SriovDevicePluginPod, metav1.CreateOptions{})
+		Expect(err).ToNot(HaveOccurred())
+
 	})
 
 	AfterEach(func() {
@@ -286,6 +291,34 @@ var _ = Describe("Config Daemon", func() {
 
 			Expect(sut.desiredNodeState.GetGeneration()).To(BeNumerically("==", 777))
 		})
+
+		It("restart all the sriov-device-plugin pods present on the node", func() {
+			otherPod1 := SriovDevicePluginPod.DeepCopy()
+			otherPod1.Name = "sriov-device-plugin-xxxa"
+			_, err := sut.kubeClient.CoreV1().Pods(vars.Namespace).Create(context.Background(), otherPod1, metav1.CreateOptions{})
+			Expect(err).ToNot(HaveOccurred())
+
+			otherPod2 := SriovDevicePluginPod.DeepCopy()
+			otherPod2.Name = "sriov-device-plugin-xxxz"
+			_, err = sut.kubeClient.CoreV1().Pods(vars.Namespace).Create(context.Background(), otherPod2, metav1.CreateOptions{})
+			Expect(err).ToNot(HaveOccurred())
+
+			err = sut.restartDevicePluginPod()
+			Expect(err).ToNot(HaveOccurred())
+
+			Eventually(func() (int, error) {
+				podList, err := sut.kubeClient.CoreV1().Pods(vars.Namespace).List(context.Background(), metav1.ListOptions{
+					LabelSelector: "app=sriov-device-plugin",
+					FieldSelector: "spec.nodeName=test-node",
+				})
+
+				if err != nil {
+					return 0, err
+				}
+
+				return len(podList.Items), nil
+			}, "1s").Should(BeZero())
+		})
 	})
 })