Troubleshooting guide
Nginx Proxy Manager 502 After Restart: Verify the Upstream Is Actually Listening
Category: Reverse Proxy & SSL
You restarted your stack, the upstream service is running on its direct port, but the proxy host now returns 502 before you can run a controlled diagnosis.
Check the upstream listening address
On the Docker host, run docker ps and note the upstream container name. Then run docker exec <container> netstat -tlnp (or ss -tlnp) inside it. Look at the local address column. If it shows 127.0.0.1:8080, the service only listens on loopback inside its container. Nginx Proxy Manager connects from another container, so that loopback address is unreachable and you get a nginx proxy manager 502.
Confirm the port and protocol
From the NPM container, run docker exec <npm-container> curl -v http://<upstream-container>:8080. A connection refused or timeout confirms the address mismatch. A successful HTTP response means the port is correct and you should check the scheme in the proxy host (HTTP vs HTTPS).
Apply the fix
Edit the upstream service configuration to listen on 0.0.0.0:8080 instead of 127.0.0.1:8080, then restart only that container. In NPM, open the proxy host and confirm the forward hostname is the container name and the forward port matches the listening port. Save the host.
Verification test
Run the same curl -v from the NPM container. You should see an HTTP status line. Then load the proxy host URL in a browser; it should return the upstream page, not 502.
Rollback note
If the service fails to start after changing the bind address, revert to the previous configuration and restart it. If the 502 persists,

Ask for help with the curl output and container names.