add probe for workers

k8s/charts/seaweedfs/README.md

@@ -216,9 +216,9 @@ worker:
   workingDir: "/tmp/seaweedfs-worker"
   
   # Optional: configure admin server address
-  # If not specified, auto-discovers from admin service in the same namespace
-  # Replace <namespace> with your actual namespace
-  adminServer: "seaweedfs-admin.<namespace>:33646"
+  # If not specified, auto-discovers from admin service in the same namespace.
+  # Replace <namespace> with the namespace where the admin service is deployed.
+  adminServer: "seaweedfs-admin.<namespace>.svc:33646"
   
   # Workers need storage for task execution
   # Note: Workers use a Deployment, which does not support `volumeClaimTemplates` 
@@ -282,7 +282,7 @@ worker:
   maxConcurrent: 2
   # REQUIRED: Point to the admin service of your main SeaweedFS release
   # Replace <namespace> with the namespace where your main seaweedfs is deployed
-  adminServer: "seaweedfs-admin.<namespace>:33646"
+  adminServer: "seaweedfs-admin.<namespace>.svc:33646"
 ```
 
 **values-worker-balance.yaml** (for balance operations):
@@ -306,7 +306,7 @@ worker:
   maxConcurrent: 1
   # REQUIRED: Point to the admin service of your main SeaweedFS release
   # Replace <namespace> with the namespace where your main seaweedfs is deployed
-  adminServer: "seaweedfs-admin.<namespace>:33646"
+  adminServer: "seaweedfs-admin.<namespace>.svc:33646"
 ```
 
 Deploy the specialized workers as separate releases:

k8s/charts/seaweedfs/templates/worker/worker-deployment.yaml

@@ -182,6 +182,30 @@ spec:
           resources:
             {{- toYaml . | nindent 12 }}
           {{- end }}
+          {{- if .Values.worker.livenessProbe.enabled }}
+          livenessProbe:
+            {{- with .Values.worker.livenessProbe.tcpSocket }}
+            tcpSocket:
+              port: {{ .port }}
+            {{- end }}
+            initialDelaySeconds: {{ .Values.worker.livenessProbe.initialDelaySeconds }}
+            periodSeconds: {{ .Values.worker.livenessProbe.periodSeconds }}
+            successThreshold: {{ .Values.worker.livenessProbe.successThreshold }}
+            failureThreshold: {{ .Values.worker.livenessProbe.failureThreshold }}
+            timeoutSeconds: {{ .Values.worker.livenessProbe.timeoutSeconds }}
+          {{- end }}
+          {{- if .Values.worker.readinessProbe.enabled }}
+          readinessProbe:
+            {{- with .Values.worker.readinessProbe.tcpSocket }}
+            tcpSocket:
+              port: {{ .port }}
+            {{- end }}
+            initialDelaySeconds: {{ .Values.worker.readinessProbe.initialDelaySeconds }}
+            periodSeconds: {{ .Values.worker.readinessProbe.periodSeconds }}
+            successThreshold: {{ .Values.worker.readinessProbe.successThreshold }}
+            failureThreshold: {{ .Values.worker.readinessProbe.failureThreshold }}
+            timeoutSeconds: {{ .Values.worker.readinessProbe.timeoutSeconds }}
+          {{- end }}
           {{- if .Values.worker.containerSecurityContext.enabled }}
           securityContext: {{- omit .Values.worker.containerSecurityContext "enabled" | toYaml | nindent 12 }}
           {{- end }}

k8s/charts/seaweedfs/values.yaml

@@ -1296,6 +1296,28 @@ worker:
 
   extraEnvironmentVars: {}
 
+  # Health checks for worker pods
+  # Since workers do not have an HTTP endpoint, a tcpSocket probe on the metrics port is recommended.
+  livenessProbe:
+    enabled: true
+    tcpSocket:
+      port: metrics
+    initialDelaySeconds: 30
+    periodSeconds: 60
+    successThreshold: 1
+    failureThreshold: 5
+    timeoutSeconds: 10
+
+  readinessProbe:
+    enabled: true
+    tcpSocket:
+      port: metrics
+    initialDelaySeconds: 20
+    periodSeconds: 15
+    successThreshold: 1
+    failureThreshold: 3
+    timeoutSeconds: 10
+
   # ServiceMonitor annotations (separate from pod/deployment annotations)
   serviceMonitor:
     annotations: {}