diff --git a/planpalto-helm/values.yaml b/planpalto-helm/values.yaml
index dfe96bc..b313f8e 100644
--- a/planpalto-helm/values.yaml
+++ b/planpalto-helm/values.yaml
@@ -60,11 +60,13 @@ apps:
     hpa: { min: 2, max: 5, cpu: 50 }
     pdb: { maxUnavailable: 1 }
 
+  # Workers declare no metrics port: the worker binaries don't serve /metrics
+  # (only planpal-server does). Adding one just yields a DOWN scrape target.
   schedule-worker:
     command: [planpal-schedule-worker]
     image: backend
     envFrom: [backend-env, aws-env]
-    ports: [{ name: metrics, containerPort: 9091 }]
+    ports: []
     resources:
       requests: { cpu: 25m, memory: 64Mi }
       limits: { cpu: 200m, memory: 256Mi }
@@ -74,7 +76,7 @@ apps:
     command: [planpal-notification-worker]
     image: backend
     envFrom: [backend-env, aws-env]
-    ports: [{ name: metrics, containerPort: 9092 }]
+    ports: []
     resources:
       requests: { cpu: 25m, memory: 64Mi }
       limits: { cpu: 200m, memory: 256Mi }
@@ -84,7 +86,7 @@ apps:
     command: [planpal-ai-worker]
     image: backend
     envFrom: [backend-env, aws-env]
-    ports: [{ name: metrics, containerPort: 9093 }]
+    ports: []
     resources:
       requests: { cpu: 25m, memory: 64Mi }
       limits: { cpu: 200m, memory: 256Mi }