diff --git a/k8s/apps/comfyui/deployment.yaml b/k8s/apps/comfyui/deployment.yaml
index 2e723a0..017b757 100644
--- a/k8s/apps/comfyui/deployment.yaml
+++ b/k8s/apps/comfyui/deployment.yaml
@@ -22,19 +22,19 @@ spec:
       nodeSelector:
         kubernetes.io/hostname: worker-1
       runtimeClassName: nvidia
+      # k8s Service named 'comfyui' injects COMFYUI_PORT=tcp://... into pod env,
+      # which clobbers ai-dock's own COMFYUI_PORT variable (expects a port number).
+      # Disable service link injection to avoid the collision.
+      enableServiceLinks: false
       containers:
         - name: comfyui
           image: ghcr.io/ai-dock/comfyui:v2-cuda-12.1.1-base-22.04
           ports:
-            - containerPort: 18188
+            - containerPort: 8188
               protocol: TCP
           env:
             - name: NVIDIA_VISIBLE_DEVICES
               value: "all"
-            # ai-dock adds 10000 to port: ComfyUI listens on 18188 (localhost)
-            # caddy is NOT configured to proxy 8188->18188 in this image
-            - name: COMFYUI_PORT_HOST
-              value: "18188"
           resources:
             requests:
               cpu: "4"
@@ -52,14 +52,14 @@ spec:
           readinessProbe:
             httpGet:
               path: /
-              port: 18188
+              port: 8188
             periodSeconds: 10
             initialDelaySeconds: 30
           startupProbe:
             httpGet:
               path: /
-              port: 18188
-            failureThreshold: 60
+              port: 8188
+            failureThreshold: 120
             periodSeconds: 10
       volumes:
         - name: models
diff --git a/k8s/apps/comfyui/service.yaml b/k8s/apps/comfyui/service.yaml
index 4bb9691..7afee74 100644
--- a/k8s/apps/comfyui/service.yaml
+++ b/k8s/apps/comfyui/service.yaml
@@ -10,5 +10,5 @@ spec:
     app: comfyui
   ports:
     - port: 80
-      targetPort: 18188
+      targetPort: 8188
       protocol: TCP