Fix port conflict in zero-downtime deployment

- Remove port mapping for temporary container to avoid conflicts
- Use docker exec for health checks instead of external port access
- Eliminates 'port already allocated' error
- Maintains zero-downtime functionality without port conflicts

.gitea/workflows/ci-cd.yml

@@ -109,12 +109,11 @@ jobs:
           if [ "$CURRENT_CONTAINER_RUNNING" = "true" ]; then
             echo "🔄 Performing rolling update..."
             
-            # Start new container with temporary name
+            # Start new container with temporary name (no port mapping needed for health check)
             docker run -d \
               --name portfolio-app-new \
               --restart unless-stopped \
               --network portfolio_net \
-              -p 3001:3000 \
               -e NODE_ENV=${{ vars.NODE_ENV }} \
               -e LOG_LEVEL=${{ vars.LOG_LEVEL }} \
               -e DATABASE_URL=postgresql://portfolio_user:portfolio_pass@postgres:5432/portfolio_db?schema=public \
@@ -133,9 +132,9 @@ jobs:
             echo "⏳ Waiting for new container to be ready..."
             sleep 15
             
-            # Health check new container
+            # Health check new container using docker exec
             for i in {1..20}; do
-              if curl -f http://localhost:3001/api/health > /dev/null 2>&1; then
+              if docker exec portfolio-app-new curl -f http://localhost:3000/api/health > /dev/null 2>&1; then
                 echo "✅ New container is healthy!"
                 break
               fi