chore(sysadvisor): disable some healthcheck rule (#539)

kubewharf · Apr 10, 2024 · 76d9f46 · 76d9f46
1 parent 72afaf3
commit 76d9f46
Show file tree

Hide file tree

Showing 3 changed files with 4 additions and 8 deletions.
diff --git a/pkg/agent/sysadvisor/plugin/inference/inference.go b/pkg/agent/sysadvisor/plugin/inference/inference.go
@@ -104,9 +104,6 @@ func NewInferencePlugin(pluginName string, conf *config.Configuration, extraConf
 }
 
 func (infp *InferencePlugin) Run(ctx context.Context) {
-	if len(infp.modelsResultFetchers) > 0 {
-		general.RegisterHeartbeatCheck(borweinfetcher.BorweinModelResultFetcherName, 3*infp.period, general.HealthzCheckStateNotReady, 3*infp.period)
-	}
 	wait.UntilWithContext(ctx, infp.fetchModelResult, infp.period)
 }
 
@@ -128,9 +125,6 @@ func (infp *InferencePlugin) fetchModelResult(ctx context.Context) {
 		go func(modelName string, fetcher modelresultfetcher.ModelResultFetcher) {
 			defer wg.Done()
 			err := fetcher.FetchModelResult(ctx, infp.metaReader, infp.metaWriter, infp.metaServer)
-			defer func() {
-				_ = general.UpdateHealthzStateByError(borweinfetcher.BorweinModelResultFetcherName, err)
-			}()
 			if err != nil {
 				general.Errorf("FetchModelResult for model: %s failed with error: %v", modelName, err)
 			}

diff --git a/pkg/agent/sysadvisor/plugin/qosaware/reporter/reporter_interface.go b/pkg/agent/sysadvisor/plugin/qosaware/reporter/reporter_interface.go
@@ -21,7 +21,7 @@ import (
 	"time"
 )
 
-const healthCheckTolerationDuration = 15 * time.Second
+const healthCheckTolerationDuration = 5 * time.Minute
 
 // Reporter is used to report resource
 type Reporter interface {

diff --git a/pkg/agent/sysadvisor/plugin/qosaware/resource/memory/plugin/memory_balancer.go b/pkg/agent/sysadvisor/plugin/qosaware/resource/memory/plugin/memory_balancer.go
@@ -313,7 +313,9 @@ func (m *memoryBalancer) getBalanceInfo() (balanceInfo *BalanceInfo, err error)
 	}
 
 	if balanceInfo.MaxLatencyNuma.ReadLatency == 0 || balanceInfo.MaxBandwidthNuma.ReadWriteBandwidth == 0 {
-		err = fmt.Errorf("all numas read latency or bandwidth are 0")
+		// TODO: this exception always happens in VM node since malachite can't provide related metric for VM node,
+		// maybe we should disable this whole feature when running in VM node.
+		general.Errorf("all numas read latency or bandwidth are 0")
 		return
 	}