fix(ci): explicitly stop old app containers before redeploy

Existing containers lack compose project labels so Compose cannot claim
them — it tries to CREATE alongside them and hits a name conflict.

Fix: stop + rm only the 6 meezi app containers by name before compose up.
Postgres, redis, and all other projects are never touched.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

.gitea/workflows/ci-cd.yml

@@ -310,16 +310,20 @@ jobs:
           DOCKER_BUILDKIT: 1
           COMPOSE_DOCKER_CLI_BUILD: 1
 
-      - name: Deploy main app services
+      - name: Stop old app containers
-        # Only cycle the app containers — postgres and redis are long-running
+        # The existing containers were created before compose labels were added,
-        # infrastructure that must never be restarted by CI.
+        # so Compose can't claim them and hits a name conflict on 'up'.
-        # --force-recreate: always swap in the freshly-built images.
+        # This step removes only meezi's own 6 app containers — never touches
-        # --no-deps: do NOT pull in postgres/redis as dependencies.
+        # postgres, redis, or any other project's containers.
         run: |
-          docker compose up -d \
+          for name in meezi-api meezi-web meezi-website meezi-koja meezi-admin-api meezi-admin-web; do
-            --no-deps \
+            docker stop "$name" 2>/dev/null || true
-            --force-recreate \
+            docker rm   "$name" 2>/dev/null || true
-            api web website koja
+          done
+
+      - name: Deploy main app services
+        run: |
+          docker compose up -d --no-deps api web website koja
 
       - name: Deploy admin services
         run: |
@@ -328,7 +332,6 @@ jobs:
             -f docker-compose.admin.yml \
             up -d \
             --no-deps \
-            --force-recreate \
             admin-api admin-web
 
       - name: Wait for main API healthy