How do you diagnose 502 and 504 errors in Nginx?

Question

IT Interview Practice · Accepted Answer

Find where the request failed before increasing timeouts. A 502 often points to failed upstream communication; a 504 often points to an upstream wait that expired. Logs distinguish the actual failure.

## Example: the application is not listening

Suppose an error-log excerpt says:

```text
connect() failed (111: Connection refused) while connecting to upstream
upstream: "http://127.0.0.1:3000/products"
```

Nginx tried to connect to port 3000 and was refused. Check whether the application is running and listening on that address. From Nginx's own machine or container, test `curl -i http://127.0.0.1:3000/products`. Inside a container, that address refers to the container itself, not automatically to another application's container.

## Example: connected, but no response headers in time

```text
upstream timed out while reading response header from upstream
```

This points to waiting for response headers. Correlate the request with application logs and slow database queries. Check connection time and upstream response time to separate network setup from application work.

## Confirm the fix

Correct the address or restore the failed application for the first case; investigate slow work for the second. Retry the affected route and monitor error rate and latency. A longer `proxy_read_timeout` permits a longer gap between reads; it does not create backend capacity. Check CDN and application logs too if either produced the error.

[Official documentation](https://nginx.org/en/docs/http/ngx_http_proxy_module.html#proxy_read_timeout).

Example: the application is not listening

Example: connected, but no response headers in time

Confirm the fix

Related questions