multicast: make igmp query interval configurable

liu4480 · liu4480 · commit 96d569c1c3cb · 2022-05-24T17:32:39.000+08:00
Signed-off-by: Bin Liu &lt;biliu@vmware.com&gt;
diff --git a/build/charts/antrea/README.md b/build/charts/antrea/README.md
@@ -76,6 +76,7 @@ Kubernetes: `>= 1.16.0-0`
 | kubeAPIServerOverride | string | `""` | Address of Kubernetes apiserver, to override any value provided in kubeconfig or InClusterConfig. |
 | logVerbosity | int | `0` |  |
 | multicastInterfaces | list | `[]` | Names of the interfaces on Nodes that are used to forward multicast traffic. |
+| igmpQueryInterval | int | `125`| The interval for antrea-agent to send IGMP query to Pods, time unit is second.
 | noSNAT | bool | `false` | Whether or not to SNAT (using the Node IP) the egress traffic from a Pod to the external network. |
 | nodeIPAM.clusterCIDRs | list | `[]` | CIDR ranges to use when allocating Pod IP addresses. |
 | nodeIPAM.enable | bool | `false` | Enable Node IPAM in Antrea |
diff --git a/build/charts/antrea/conf/antrea-agent.conf b/build/charts/antrea/conf/antrea-agent.conf
@@ -217,12 +217,19 @@ tlsMinVersion: {{ .Values.tlsMinVersion | quote }}
 # 3. The Node IP
 transportInterface: {{ .Values.transportInterface | quote }}
 
+multicast:
+{{- with .Values.multicast }}
 # The names of the interfaces on Nodes that are used to forward multicast traffic.
 # Defaults to transport interface if not set.
-multicastInterfaces:
-{{- with .Values.multicastInterfaces }}
-{{- toYaml . | nindent 2 }}
-{{- end }}
+  multicastInterfaces:
+  {{- with .multicastInterfaces }}
+  {{- toYaml . | nindent 2 }}
+  {{- end }}
+
+# The interval for antrea-agent to send IGMP query to Pods, time unit is second.
+# Defaults to 125 seconds.
+  igmpQueryInterval: {{ .igmpQueryInterval }}
+{{- end}}
 
 # The network CIDRs of the interface on Node which is used for tunneling or routing the traffic across
 # Nodes. If there are multiple interfaces configured the same network CIDR, the first one is used. The
diff --git a/build/charts/antrea/values.yaml b/build/charts/antrea/values.yaml
@@ -28,8 +28,14 @@ transportInterface: ""
 # -- Network CIDRs of the interface on Node which is used for tunneling or
 # routing the traffic across Nodes.
 transportInterfaceCIDRs: []
-# -- Names of the interfaces on Nodes that are used to forward multicast traffic.
-multicastInterfaces: []
+
+multicast:
+  # -- Names of the interfaces on Nodes that are used to forward multicast traffic.
+  multicastInterfaces: []
+  # -- The interval for antrea-agent to send IGMP query to pods, time uint is second.
+  # Defaults to 125 seconds.
+  igmpQueryInterval: 125
+
 # -- Default MTU to use for the host gateway interface and the network interface
 # of each Pod. By default, antrea-agent will discover the MTU of the Node's
 # primary interface and adjust it to accommodate for tunnel encapsulation
diff --git a/build/yamls/antrea-aks.yml b/build/yamls/antrea-aks.yml
@@ -266,9 +266,14 @@ data:
     # 3. The Node IP
     transportInterface: ""
 
+    multicast:
     # The names of the interfaces on Nodes that are used to forward multicast traffic.
     # Defaults to transport interface if not set.
-    multicastInterfaces:
+      multicastInterfaces:
+
+    # The interval for antrea-agent to send IGMP query to Pods, time unit is second.
+    # Defaults to 125 seconds.
+      igmpQueryInterval: 125
 
     # The network CIDRs of the interface on Node which is used for tunneling or routing the traffic across
     # Nodes. If there are multiple interfaces configured the same network CIDR, the first one is used. The
@@ -3486,7 +3491,7 @@ spec:
         kubectl.kubernetes.io/default-container: antrea-agent
         # Automatically restart Pods with a RollingUpdate if the ConfigMap changes
         # See https://helm.sh/docs/howto/charts_tips_and_tricks/#automatically-roll-deployments
-        checksum/config: 4554a36b927c6e64fdbc53b4d4c64673d48c9c829ec444e3be6e699ade8481b6
+        checksum/config: 3b457b52eff11631f8afc718f43769caa51133a1c30a2bc858972e4ea79fd753
       labels:
         app: antrea
         component: antrea-agent
@@ -3726,7 +3731,7 @@ spec:
       annotations:
         # Automatically restart Pod if the ConfigMap changes
         # See https://helm.sh/docs/howto/charts_tips_and_tricks/#automatically-roll-deployments
-        checksum/config: 4554a36b927c6e64fdbc53b4d4c64673d48c9c829ec444e3be6e699ade8481b6
+        checksum/config: 3b457b52eff11631f8afc718f43769caa51133a1c30a2bc858972e4ea79fd753
       labels:
         app: antrea
         component: antrea-controller
diff --git a/build/yamls/antrea-eks.yml b/build/yamls/antrea-eks.yml
@@ -266,9 +266,14 @@ data:
     # 3. The Node IP
     transportInterface: ""
 
+    multicast:
     # The names of the interfaces on Nodes that are used to forward multicast traffic.
     # Defaults to transport interface if not set.
-    multicastInterfaces:
+      multicastInterfaces:
+
+    # The interval for antrea-agent to send IGMP query to Pods, time unit is second.
+    # Defaults to 125 seconds.
+      igmpQueryInterval: 125
 
     # The network CIDRs of the interface on Node which is used for tunneling or routing the traffic across
     # Nodes. If there are multiple interfaces configured the same network CIDR, the first one is used. The
@@ -3486,7 +3491,7 @@ spec:
         kubectl.kubernetes.io/default-container: antrea-agent
         # Automatically restart Pods with a RollingUpdate if the ConfigMap changes
         # See https://helm.sh/docs/howto/charts_tips_and_tricks/#automatically-roll-deployments
-        checksum/config: 4554a36b927c6e64fdbc53b4d4c64673d48c9c829ec444e3be6e699ade8481b6
+        checksum/config: 3b457b52eff11631f8afc718f43769caa51133a1c30a2bc858972e4ea79fd753
       labels:
         app: antrea
         component: antrea-agent
@@ -3728,7 +3733,7 @@ spec:
       annotations:
         # Automatically restart Pod if the ConfigMap changes
         # See https://helm.sh/docs/howto/charts_tips_and_tricks/#automatically-roll-deployments
-        checksum/config: 4554a36b927c6e64fdbc53b4d4c64673d48c9c829ec444e3be6e699ade8481b6
+        checksum/config: 3b457b52eff11631f8afc718f43769caa51133a1c30a2bc858972e4ea79fd753
       labels:
         app: antrea
         component: antrea-controller
diff --git a/build/yamls/antrea-gke.yml b/build/yamls/antrea-gke.yml
@@ -266,9 +266,14 @@ data:
     # 3. The Node IP
     transportInterface: ""
 
+    multicast:
     # The names of the interfaces on Nodes that are used to forward multicast traffic.
     # Defaults to transport interface if not set.
-    multicastInterfaces:
+      multicastInterfaces:
+
+    # The interval for antrea-agent to send IGMP query to Pods, time unit is second.
+    # Defaults to 125 seconds.
+      igmpQueryInterval: 125
 
     # The network CIDRs of the interface on Node which is used for tunneling or routing the traffic across
     # Nodes. If there are multiple interfaces configured the same network CIDR, the first one is used. The
@@ -3486,7 +3491,7 @@ spec:
         kubectl.kubernetes.io/default-container: antrea-agent
         # Automatically restart Pods with a RollingUpdate if the ConfigMap changes
         # See https://helm.sh/docs/howto/charts_tips_and_tricks/#automatically-roll-deployments
-        checksum/config: edef4c00e4f28a10dc1e077086ef68641a9a3b53d0fe7d47ff3dafc2ce5d5c9b
+        checksum/config: 78dbecf55ff93beb84ac29a187f617bd3853af405520b92cadd7a4d98c5b5ed3
       labels:
         app: antrea
         component: antrea-agent
@@ -3726,7 +3731,7 @@ spec:
       annotations:
         # Automatically restart Pod if the ConfigMap changes
         # See https://helm.sh/docs/howto/charts_tips_and_tricks/#automatically-roll-deployments
-        checksum/config: edef4c00e4f28a10dc1e077086ef68641a9a3b53d0fe7d47ff3dafc2ce5d5c9b
+        checksum/config: 78dbecf55ff93beb84ac29a187f617bd3853af405520b92cadd7a4d98c5b5ed3
       labels:
         app: antrea
         component: antrea-controller
diff --git a/build/yamls/antrea-ipsec.yml b/build/yamls/antrea-ipsec.yml
@@ -279,9 +279,14 @@ data:
     # 3. The Node IP
     transportInterface: ""
 
+    multicast:
     # The names of the interfaces on Nodes that are used to forward multicast traffic.
     # Defaults to transport interface if not set.
-    multicastInterfaces:
+      multicastInterfaces:
+
+    # The interval for antrea-agent to send IGMP query to Pods, time unit is second.
+    # Defaults to 125 seconds.
+      igmpQueryInterval: 125
 
     # The network CIDRs of the interface on Node which is used for tunneling or routing the traffic across
     # Nodes. If there are multiple interfaces configured the same network CIDR, the first one is used. The
@@ -3499,7 +3504,7 @@ spec:
         kubectl.kubernetes.io/default-container: antrea-agent
         # Automatically restart Pods with a RollingUpdate if the ConfigMap changes
         # See https://helm.sh/docs/howto/charts_tips_and_tricks/#automatically-roll-deployments
-        checksum/config: 50cc962db93c0354f5eaa088e51d690e779692979cbafac0c3e27a88fc2c0c7c
+        checksum/config: 0d141884a1852335a8a14916850b5b9d8924f0b205a6b9bf179399381bf04fea
         checksum/ipsec-secret: d0eb9c52d0cd4311b6d252a951126bf9bea27ec05590bed8a394f0f792dcb2a4
       labels:
         app: antrea
@@ -3775,7 +3780,7 @@ spec:
       annotations:
         # Automatically restart Pod if the ConfigMap changes
         # See https://helm.sh/docs/howto/charts_tips_and_tricks/#automatically-roll-deployments
-        checksum/config: 50cc962db93c0354f5eaa088e51d690e779692979cbafac0c3e27a88fc2c0c7c
+        checksum/config: 0d141884a1852335a8a14916850b5b9d8924f0b205a6b9bf179399381bf04fea
       labels:
         app: antrea
         component: antrea-controller
diff --git a/build/yamls/antrea.yml b/build/yamls/antrea.yml
@@ -266,9 +266,14 @@ data:
     # 3. The Node IP
     transportInterface: ""
 
+    multicast:
     # The names of the interfaces on Nodes that are used to forward multicast traffic.
     # Defaults to transport interface if not set.
-    multicastInterfaces:
+      multicastInterfaces:
+
+    # The interval for antrea-agent to send IGMP query to Pods, time unit is second.
+    # Defaults to 125 seconds.
+      igmpQueryInterval: 125
 
     # The network CIDRs of the interface on Node which is used for tunneling or routing the traffic across
     # Nodes. If there are multiple interfaces configured the same network CIDR, the first one is used. The
@@ -3486,7 +3491,7 @@ spec:
         kubectl.kubernetes.io/default-container: antrea-agent
         # Automatically restart Pods with a RollingUpdate if the ConfigMap changes
         # See https://helm.sh/docs/howto/charts_tips_and_tricks/#automatically-roll-deployments
-        checksum/config: f7414e9171ab246b09dc380bc7934ebac81af7a5ef7bd3f73d661b6301040768
+        checksum/config: 37d55103cfa223781c44d6d05c7094cc8038fc6c4c03770bca35065acd4ba81a
       labels:
         app: antrea
         component: antrea-agent
@@ -3726,7 +3731,7 @@ spec:
       annotations:
         # Automatically restart Pod if the ConfigMap changes
         # See https://helm.sh/docs/howto/charts_tips_and_tricks/#automatically-roll-deployments
-        checksum/config: f7414e9171ab246b09dc380bc7934ebac81af7a5ef7bd3f73d661b6301040768
+        checksum/config: 37d55103cfa223781c44d6d05c7094cc8038fc6c4c03770bca35065acd4ba81a
       labels:
         app: antrea
         component: antrea-controller
diff --git a/cmd/antrea-agent/agent.go b/cmd/antrea-agent/agent.go
@@ -589,9 +589,10 @@ func run(o *Options) error {
 			nodeConfig,
 			ifaceStore,
 			multicastSocket,
-			sets.NewString(append(o.config.MulticastInterfaces, nodeConfig.NodeTransportInterfaceName)...),
+			sets.NewString(append(o.config.Multicast.MulticastInterfaces, nodeConfig.NodeTransportInterfaceName)...),
 			ovsBridgeClient,
-			podUpdateChannel)
+			podUpdateChannel,
+			o.igmpQueryInterval)
 		if err := mcastController.Initialize(); err != nil {
 			return err
 		}
diff --git a/cmd/antrea-agent/options.go b/cmd/antrea-agent/options.go
@@ -47,6 +47,7 @@ const (
 	defaultFlowPollInterval        = 5 * time.Second
 	defaultActiveFlowExportTimeout = 30 * time.Second
 	defaultIdleFlowExportTimeout   = 15 * time.Second
+	defaultIGMPQueryInterval       = 125 * time.Second
 	defaultStaleConnectionTimeout  = 5 * time.Minute
 	defaultNPLPortRange            = "61000-62000"
 )
@@ -68,8 +69,10 @@ type Options struct {
 	idleFlowTimeout time.Duration
 	// Stale connection timeout to delete connections if they are not exported.
 	staleConnectionTimeout time.Duration
-	nplStartPort           int
-	nplEndPort             int
+	// IGMP query interval
+	igmpQueryInterval time.Duration
+	nplStartPort      int
+	nplEndPort        int
 }
 
 func newOptions() *Options {
@@ -271,6 +274,14 @@ func (o *Options) setDefaults() {
 			o.config.NodePortLocal.PortRange = defaultNPLPortRange
 		}
 	}
+
+	if features.DefaultFeatureGate.Enabled(features.Multicast) {
+		if o.config.Multicast.IGMPQueryInterval == 0 {
+			o.igmpQueryInterval = defaultIGMPQueryInterval
+		} else {
+			o.igmpQueryInterval = time.Duration(o.config.Multicast.IGMPQueryInterval) * time.Second
+		}
+	}
 }
 
 func (o *Options) validateAntreaProxyConfig() error {
diff --git a/pkg/agent/multicast/mcast_controller.go b/pkg/agent/multicast/mcast_controller.go
@@ -153,20 +153,20 @@ func (c *Controller) checkLastMember(group net.IP) {
 	c.queue.AddAfter(group.String(), igmpMaxResponseTime)
 }
 
-// clearStaleGroups checks the stale group members which have not been updated for mcastGroupTimeout, and then notifies worker
+// clearStaleGroups checks the stale group members which have not been updated for c.mcastGroupTimeout, and then notifies worker
 // to remove them from groupCache.
 func (c *Controller) clearStaleGroups() {
 	now := time.Now()
 	for _, obj := range c.groupCache.List() {
 		status := obj.(*GroupMemberStatus)
 		diff := now.Sub(status.lastIGMPReport)
-		if diff > mcastGroupTimeout {
+		if diff > c.mcastGroupTimeout {
 			// Notify worker to remove the group from groupCache if all its members are not updated before mcastGroupTimeout.
 			c.queue.Add(status.group.String())
 		} else {
 			// Create a "leave" event for a local member if it is not updated before mcastGroupTimeout.
 			for member, lastUpdate := range status.localMembers {
-				if now.Sub(lastUpdate) > mcastGroupTimeout {
+				if now.Sub(lastUpdate) > c.mcastGroupTimeout {
 					ifConfig := &interfacestore.InterfaceConfig{
 						InterfaceName: member,
 					}
@@ -222,6 +222,11 @@ type Controller struct {
 	installedGroupsMutex sync.RWMutex
 	mRouteClient         *MRouteClient
 	ovsBridgeClient      ovsconfig.OVSBridgeClient
+
+	// queryInterval is the interval to send IGMP query messages.
+	queryInterval time.Duration
+	// mcastGroupTimeout is the timeout to detect a group as stale if no IGMP report is received within the time.
+	mcastGroupTimeout time.Duration
 }
 
 func NewMulticastController(ofClient openflow.Client,
@@ -231,25 +236,28 @@ func NewMulticastController(ofClient openflow.Client,
 	multicastSocket RouteInterface,
 	multicastInterfaces sets.String,
 	ovsBridgeClient ovsconfig.OVSBridgeClient,
-	podUpdateSubscriber channel.Subscriber) *Controller {
+	podUpdateSubscriber channel.Subscriber,
+	igmpQueryInterval time.Duration) *Controller {
 	eventCh := make(chan *mcastGroupEvent, workerCount)
-	groupSnooper := newSnooper(ofClient, ifaceStore, eventCh)
+	groupSnooper := newSnooper(ofClient, ifaceStore, eventCh, igmpQueryInterval)
 	groupCache := cache.NewIndexer(getGroupEventKey, cache.Indexers{
 		podInterfaceIndex: podInterfaceIndexFunc,
 	})
 	multicastRouteClient := newRouteClient(nodeConfig, groupCache, multicastSocket, multicastInterfaces)
 	c := &Controller{
-		ofClient:         ofClient,
-		ifaceStore:       ifaceStore,
-		v4GroupAllocator: v4GroupAllocator,
-		nodeConfig:       nodeConfig,
-		igmpSnooper:      groupSnooper,
-		groupEventCh:     eventCh,
-		groupCache:       groupCache,
-		installedGroups:  sets.NewString(),
-		queue:            workqueue.NewNamedRateLimitingQueue(workqueue.NewItemExponentialFailureRateLimiter(minRetryDelay, maxRetryDelay), "multicastgroup"),
-		mRouteClient:     multicastRouteClient,
-		ovsBridgeClient:  ovsBridgeClient,
+		ofClient:          ofClient,
+		ifaceStore:        ifaceStore,
+		v4GroupAllocator:  v4GroupAllocator,
+		nodeConfig:        nodeConfig,
+		igmpSnooper:       groupSnooper,
+		groupEventCh:      eventCh,
+		groupCache:        groupCache,
+		installedGroups:   sets.NewString(),
+		queue:             workqueue.NewNamedRateLimitingQueue(workqueue.NewItemExponentialFailureRateLimiter(minRetryDelay, maxRetryDelay), "multicastgroup"),
+		mRouteClient:      multicastRouteClient,
+		ovsBridgeClient:   ovsBridgeClient,
+		queryInterval:     igmpQueryInterval,
+		mcastGroupTimeout: igmpQueryInterval * 3,
 	}
 	podUpdateSubscriber.Subscribe(c.removeLocalInterface)
 	return c
@@ -279,10 +287,10 @@ func (c *Controller) Run(stopCh <-chan struct{}) {
 		if err := c.igmpSnooper.queryIGMP(net.IPv4zero, queryVersions); err != nil {
 			klog.ErrorS(err, "Failed to send IGMP query")
 		}
-	}, queryInterval, stopCh)
+	}, c.queryInterval, stopCh)
 
 	// Periodically check the group member status, and remove the groups in which no members exist
-	go wait.NonSlidingUntil(c.clearStaleGroups, queryInterval, stopCh)
+	go wait.NonSlidingUntil(c.clearStaleGroups, c.queryInterval, stopCh)
 	go c.eventHandler(stopCh)
 
 	for i := 0; i < int(workerCount); i++ {
@@ -421,11 +429,11 @@ func (c *Controller) syncGroup(groupKey string) error {
 	return nil
 }
 
-// groupIsStale returns true if no local members in the group, or there is no IGMP report received after mcastGroupTimeout.
+// groupIsStale returns true if no local members in the group, or there is no IGMP report received after c.mcastGroupTimeout.
 func (c *Controller) groupIsStale(status *GroupMemberStatus) bool {
 	membersCount := len(status.localMembers)
 	diff := time.Now().Sub(status.lastIGMPReport)
-	if membersCount == 0 || diff > mcastGroupTimeout {
+	if membersCount == 0 || diff > c.mcastGroupTimeout {
 		return true
 	}
 	return false
diff --git a/pkg/agent/multicast/mcast_controller_test.go b/pkg/agent/multicast/mcast_controller_test.go
diff --git a/pkg/agent/multicast/mcast_discovery.go b/pkg/agent/multicast/mcast_discovery.go
diff --git a/pkg/config/agent/config.go b/pkg/config/agent/config.go
diff --git a/test/e2e/framework.go b/test/e2e/framework.go
