sockmem: optimize the unnecessary fail handling

enhancement: Too many unnecessary failed logs in the current sockmem plugin Signed-off-by: Robin Lu <[email protected]>
kubewharf · Feb 12, 2025 · 1325ea9 · 1325ea9
1 parent fb6d1d8
commit 1325ea9
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 7 deletions.
diff --git a/pkg/agent/qrm-plugins/memory/dynamicpolicy/policy.go b/pkg/agent/qrm-plugins/memory/dynamicpolicy/policy.go
@@ -410,7 +410,7 @@ func (p *DynamicPolicy) Start() (err error) {
 		general.Infof("setSockMem enabled")
 		err := periodicalhandler.RegisterPeriodicalHandlerWithHealthz(memconsts.SetSockMem,
 			general.HealthzCheckStateNotReady, qrm.QRMMemoryPluginPeriodicalHandlerGroupName,
-			sockmem.SetSockMemLimit, 120*time.Second, healthCheckTolerationTimes)
+			sockmem.SetSockMemLimit, 240*time.Second, healthCheckTolerationTimes)
 		if err != nil {
 			general.Infof("setSockMem failed, err=%v", err)
 		}

diff --git a/pkg/agent/qrm-plugins/memory/handlers/sockmem/sockmem_linux.go b/pkg/agent/qrm-plugins/memory/handlers/sockmem/sockmem_linux.go
@@ -200,19 +200,17 @@ func SetSockMemLimit(conf *coreconfig.Configuration,
 			continue
 		}
 		for _, containerStatus := range pod.Status.ContainerStatuses {
-			podUID, containerID := string(pod.UID), native.TrimContainerIDPrefix(containerStatus.ContainerID)
+			if !containerStatus.Ready {
+				continue
+			}
 
+			podUID, containerID := string(pod.UID), native.TrimContainerIDPrefix(containerStatus.ContainerID)
 			memLimit, err := helper.GetPodMetric(metaServer.MetricsFetcher, emitter, pod, coreconsts.MetricMemLimitContainer, -1)
 			if err != nil {
-				errList = append(errList, err)
-				general.Infof("memory limit not found:%v..\n", podUID)
 				continue
 			}
-
 			memTCPLimit, err := helper.GetPodMetric(metaServer.MetricsFetcher, emitter, pod, coreconsts.MetricMemTCPLimitContainer, -1)
 			if err != nil {
-				errList = append(errList, err)
-				general.Infof("memory tcp.limit not found:%v..\n", podUID)
 				continue
 			}