Hi everyone, I’m currently hosting my restaurant-based website (focused on Texas Roadhouse menus and reviews) using a Dockerized setup on my VPS. The stack includes an Nginx reverse proxy, a Node.js application container, and a MySQL container. Everything was working smoothly until I updated the Docker engine and rebuilt my images last week. Since then, my website has been loading inconsistently — sometimes it loads instantly, and other times it hangs or times out completely.
The first issue I noticed was that the Nginx container started showing intermittent 502 Bad Gateway errors. The logs show that the Node.js container occasionally becomes unreachable, even though both containers are on the same custom Docker network. Restarting both containers temporarily fixes the issue, but it keeps coming back after a few hours. I’ve verified that all containers are running and healthy according to docker ps, so it doesn’t seem like a crash problem.
I also checked the Docker logs for the Node.js container and noticed several “ECONNRESET” and “EADDRINUSE” errors. It looks like the server process is not releasing ports properly after reloads. I tried rebuilding the container with a clean image and even changing the exposed port, but the behavior remains the same. I’m wondering if this might be related to a networking configuration issue in the latest Docker version or a conflict between the bridge network and the proxy container.
Another strange behavior is that my static assets (like images of the restaurant menu) are no longer being cached correctly. Before the update, Nginx handled caching efficiently, but now the assets reload from scratch each time. I’ve double-checked my Nginx config files, and the cache-control headers are still there. Could there be a problem with Docker volume mounts not syncing properly or the proxy not serving cached files due to permission issues?
Resource usage also seems unusually high. The containers collectively consume much more memory than before, even under low traffic. I ran docker stats, and it shows my Node container occasionally spiking to 100% CPU for a few seconds without any requests. I didn’t change any of the application code, so I’m thinking this could be caused by the new version of Docker, the overlay network, or some memory leak introduced by the way the containers communicate.
Has anyone else run into similar network and caching issues after updating Docker? Should I consider rolling back to a previous version of Docker Engine or changing the network driver type from bridge to host mode? I’d love to hear if there are best practices for keeping a web app like mine stable in Docker after updates, especially when using Nginx, Node.js, and MySQL together. Sorry for the long post