multicast: set igmp query interval by adding igmpQueryInterval in agent.conf

liu4480 · liu4480 · commit fe885c81bca9 · 2022-05-23T15:37:54.000+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 uint 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
@@ -224,6 +224,10 @@ multicastInterfaces:
 {{- toYaml . | nindent 2 }}
 {{- end }}
 
+# The interval for antrea-agent to send IGMP query to pods, time uint is second.
+# Defaults to 125 seconds.
+igmpQueryInterval: {{ .Values.igmpQueryInterval }}
+
 # 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
 # IP address used for tunneling or routing traffic to remote Nodes is decided in the following order of
diff --git a/build/charts/antrea/values.yaml b/build/charts/antrea/values.yaml
@@ -30,6 +30,9 @@ transportInterface: ""
 transportInterfaceCIDRs: []
 # -- 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
@@ -270,6 +270,10 @@ data:
     # Defaults to transport interface if not set.
     multicastInterfaces:
 
+    # The interval for antrea-agent to send IGMP query to pods, time uint 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
     # IP address used for tunneling or routing traffic to remote Nodes is decided in the following order of
@@ -3486,7 +3490,7 @@ 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: 20facf53caa00bcd92c0756c1f7ea86b9788ec46f66519a482ac468e86ba7f70
       labels:
         app: antrea
         component: antrea-agent
@@ -3726,7 +3730,7 @@ 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: 20facf53caa00bcd92c0756c1f7ea86b9788ec46f66519a482ac468e86ba7f70
       labels:
         app: antrea
         component: antrea-controller
diff --git a/build/yamls/antrea-eks.yml b/build/yamls/antrea-eks.yml
@@ -270,6 +270,10 @@ data:
     # Defaults to transport interface if not set.
     multicastInterfaces:
 
+    # The interval for antrea-agent to send IGMP query to pods, time uint 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
     # IP address used for tunneling or routing traffic to remote Nodes is decided in the following order of
@@ -3486,7 +3490,7 @@ 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: 20facf53caa00bcd92c0756c1f7ea86b9788ec46f66519a482ac468e86ba7f70
       labels:
         app: antrea
         component: antrea-agent
@@ -3728,7 +3732,7 @@ 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: 20facf53caa00bcd92c0756c1f7ea86b9788ec46f66519a482ac468e86ba7f70
       labels:
         app: antrea
         component: antrea-controller
diff --git a/build/yamls/antrea-gke.yml b/build/yamls/antrea-gke.yml
@@ -270,6 +270,10 @@ data:
     # Defaults to transport interface if not set.
     multicastInterfaces:
 
+    # The interval for antrea-agent to send IGMP query to pods, time uint 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
     # IP address used for tunneling or routing traffic to remote Nodes is decided in the following order of
@@ -3486,7 +3490,7 @@ 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: 2fc5286dcb73c16d939a6cd67143f2ab86b895127e0806ec036b27231130a217
       labels:
         app: antrea
         component: antrea-agent
@@ -3726,7 +3730,7 @@ 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: 2fc5286dcb73c16d939a6cd67143f2ab86b895127e0806ec036b27231130a217
       labels:
         app: antrea
         component: antrea-controller
diff --git a/build/yamls/antrea-ipsec.yml b/build/yamls/antrea-ipsec.yml
@@ -283,6 +283,10 @@ data:
     # Defaults to transport interface if not set.
     multicastInterfaces:
 
+    # The interval for antrea-agent to send IGMP query to pods, time uint 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
     # IP address used for tunneling or routing traffic to remote Nodes is decided in the following order of
@@ -3499,7 +3503,7 @@ 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: 9246c1868a5f1847e24fe1bf84cad2c05dc020dd6e4480b65bf9bfb5dd58f54b
         checksum/ipsec-secret: d0eb9c52d0cd4311b6d252a951126bf9bea27ec05590bed8a394f0f792dcb2a4
       labels:
         app: antrea
@@ -3775,7 +3779,7 @@ 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: 9246c1868a5f1847e24fe1bf84cad2c05dc020dd6e4480b65bf9bfb5dd58f54b
       labels:
         app: antrea
         component: antrea-controller
diff --git a/build/yamls/antrea.yml b/build/yamls/antrea.yml
@@ -270,6 +270,10 @@ data:
     # Defaults to transport interface if not set.
     multicastInterfaces:
 
+    # The interval for antrea-agent to send IGMP query to pods, time uint 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
     # IP address used for tunneling or routing traffic to remote Nodes is decided in the following order of
@@ -3486,7 +3490,7 @@ 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: 842778dc6a141b7868ad1dba2bc680e78ce5a5473db6c589334c88b1ec91181b
       labels:
         app: antrea
         component: antrea-agent
@@ -3726,7 +3730,7 @@ 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: 842778dc6a141b7868ad1dba2bc680e78ce5a5473db6c589334c88b1ec91181b
       labels:
         app: antrea
         component: antrea-controller
diff --git a/cmd/antrea-agent/agent.go b/cmd/antrea-agent/agent.go
@@ -591,7 +591,8 @@ func run(o *Options) error {
 			multicastSocket,
 			sets.NewString(append(o.config.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.IGMPQueryInterval == 0 {
+			o.igmpQueryInterval = defaultIGMPQueryInterval
+		} else {
+			o.igmpQueryInterval = o.igmpQueryInterval * time.Second
+		}
+	}
 }
 
 func (o *Options) validateAntreaProxyConfig() error {
diff --git a/pkg/agent/multicast/mcast_controller.go b/pkg/agent/multicast/mcast_controller.go
@@ -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,
@@ -231,7 +236,8 @@ 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)
 	groupCache := cache.NewIndexer(getGroupEventKey, cache.Indexers{
@@ -250,6 +256,7 @@ func NewMulticastController(ofClient openflow.Client,
 		queue:            workqueue.NewNamedRateLimitingQueue(workqueue.NewItemExponentialFailureRateLimiter(minRetryDelay, maxRetryDelay), "multicastgroup"),
 		mRouteClient:     multicastRouteClient,
 		ovsBridgeClient:  ovsBridgeClient,
+		queryInterval:    igmpQueryInterval,
 	}
 	podUpdateSubscriber.Subscribe(c.removeLocalInterface)
 	return c
@@ -279,10 +286,10 @@ 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++ {
diff --git a/pkg/agent/multicast/mcast_controller_test.go b/pkg/agent/multicast/mcast_controller_test.go
@@ -364,7 +364,7 @@ func newMockMulticastController(t *testing.T) *Controller {
 	mockOFClient.EXPECT().RegisterPacketInHandler(gomock.Any(), gomock.Any(), gomock.Any()).Times(1)
 	groupAllocator := openflow.NewGroupAllocator(false)
 	podUpdateSubscriber := channel.NewSubscribableChannel("PodUpdate", 100)
-	mctrl := NewMulticastController(mockOFClient, groupAllocator, nodeConfig, mockIfaceStore, mockMulticastSocket, sets.NewString(), ovsClient, podUpdateSubscriber)
+	mctrl := NewMulticastController(mockOFClient, groupAllocator, nodeConfig, mockIfaceStore, mockMulticastSocket, sets.NewString(), ovsClient, podUpdateSubscriber, queryInterval)
 	return mctrl
 }
 
diff --git a/pkg/agent/multicast/mcast_discovery.go b/pkg/agent/multicast/mcast_discovery.go
@@ -38,7 +38,7 @@ const (
 	// queryInterval is the interval to send IGMP query messages.
 	queryInterval = time.Second * 125
 	// mcastGroupTimeout is the timeout to detect a group as stale if no IGMP report is received within the time.
-	mcastGroupTimeout = queryInterval * 3
+	mcastGroupTimeout = time.Second * 125 * 3
 )
 
 var (
diff --git a/pkg/config/agent/config.go b/pkg/config/agent/config.go
@@ -185,6 +185,9 @@ type AgentConfig struct {
 	// The names of the interfaces on Nodes that are used to forward multicast traffic.
 	// Defaults to transport interface if not set.
 	MulticastInterfaces []string `yaml:"multicastInterfaces,omitempty"`
+	// The interval for antrea-agent to send IGMP query to pods, time uint is second.
+	// Defaults to 125 seconds.
+	IGMPQueryInterval int `yaml:"igmpQueryInterval"`
 	// AntreaProxy contains AntreaProxy related configuration options.
 	AntreaProxy AntreaProxyConfig `yaml:"antreaProxy,omitempty"`
 	// Egress related configurations.
