Add option to enable pprof endpoints

charts/opensearch-operator/templates/opensearch-operator-controller-manager-deployment.yaml

@@ -63,6 +63,8 @@ spec:
           value: {{ .Values.manager.dnsBase }}
         - name: PARALLEL_RECOVERY_ENABLED
           value: "{{ .Values.manager.parallelRecoveryEnabled }}"
+        - name: PPROF_ENDPOINTS_ENABLED
+          value: "{{ .Values.manager.pprofEndpointsEnabled }}"
         {{- if .Values.manager.extraEnv }}
         {{- toYaml .Values.manager.extraEnv | nindent 8 }}
         {{- end }}

charts/opensearch-operator/values.yaml

@@ -39,6 +39,9 @@ manager:
 
   # Set this to false to disable the experimental parallel recovery in case you are experiencing problems
   parallelRecoveryEnabled: true
+  # Set this to true to enable the standard go pprof endpoints on port 6060 (https://pkg.go.dev/net/http/pprof)
+  # Should only be used for debugging purposes
+  pprofEndpointsEnabled: false
 
   image:
     repository: opensearchproject/opensearch-operator

docs/userguide/main.md

@@ -100,6 +100,17 @@ manager:
   #    value: somevalue
 ```
 
+### Pprof endpoints
+
+There have been situations reported where the operator is leaking memory. To help diagnose these situations the standard go [pprof](https://pkg.go.dev/net/http/pprof) endpoints can be enabled by adding the following to your `values.yaml`:
+
+```yaml
+manager:
+  pprofEndpointsEnabled: true
+```
+
+The access the endpoints you will need to use a port-forward as for security reasons the endpoints are only exposed on localhost inside the pod: `kubectl port-forward deployment/opensearch-operator-controller-manager 6060`. Then from another terminal you can use the [go pprof tool](https://pkg.go.dev/net/http/pprof#hdr-Usage_examples), e.g.: `go tool pprof http://localhost:6060/debug/pprof/heap`.
+
 ## Configuring OpenSearch
 
 The main job of the operator is to deploy and manage OpenSearch clusters. As such it offers a wide range of options to configure clusters.

opensearch-operator/main.go

@@ -37,6 +37,9 @@ import (
 	ctrl "sigs.k8s.io/controller-runtime"
 	"sigs.k8s.io/controller-runtime/pkg/healthz"
 	"sigs.k8s.io/controller-runtime/pkg/log/zap"
+
+	"net/http"
+	_ "net/http/pprof"
 )
 
 var (
@@ -87,6 +90,11 @@ func main() {
 			setupLog.Info("Enabled debug logging via environment variable OPERATOR_DEV_LOGGING")
 		}
 	}
+	pprofEndpoints, err := strconv.ParseBool(os.Getenv("PPROF_ENDPOINTS_ENABLED"))
+	if err != nil {
+		// by default to not enable endpoints
+		pprofEndpoints = false
+	}
 
 	ctrl.SetLogger(zap.New(zap.UseFlagOptions(&opts)))
 
@@ -188,6 +196,13 @@ func main() {
 		os.Exit(1)
 	}
 
+	if pprofEndpoints {
+		go func() {
+			listenError := http.ListenAndServe("localhost:6060", nil)
+			setupLog.Error(listenError, "Failed to start pprof endpoint listener")
+		}()
+	}
+
 	setupLog.Info("Starting manager")
 	if err := mgr.Start(ctrl.SetupSignalHandler()); err != nil {
 		setupLog.Error(err, "problem running manager")