Try replacing --replace in prod deployment stage

.forgejo/workflows/ci.yml

@@ -293,19 +293,19 @@ jobs:
           docker.io/nginx:alpine \
           sh -lc 'nginx -t -c /etc/nginx/nginx.conf'
 
-    # APPLY/RE-APPLY THE POD (no explicit "down"; use --replace)
+    # If --replace is NOT supported in your Podman, use this fallback instead:
-    - name: Apply pod (kube play --replace)
+    - name: Recreate pod (fallback)
       run: |
         set -euo pipefail
-        # If your Podman supports --replace, this is the cleanest:
+        podman --remote pod rm -f sharenet-production-pod 2>/dev/null || true
-        envsubst < deploy/prod-pod.yml | podman --remote kube play --replace -
+        envsubst < deploy/prod-pod.yml | podman --remote kube play -
 
-    # If --replace is NOT supported in your Podman, use this fallback instead:
+    # If --replace IS supported in your Podman, use this instead:
-    # - name: Recreate pod (fallback)
+    # - name: Apply pod (kube play --replace)
     #   run: |
     #     set -euo pipefail
-    #     podman --remote pod rm -f sharenet-production-pod 2>/dev/null || true
+    #     # If your Podman supports --replace, this is the cleanest:
-    #     envsubst < deploy/prod-pod.yml | podman --remote kube play -
+    #     envsubst < deploy/prod-pod.yml | podman --remote kube play --replace -
 
     # VERIFY (install curl first)
     - name: Verify in-pod Nginx