# HTTP 408 Request Timeout

**Code:** `408` · **Category:** HTTP · **Seen on:** Nginx, Apache

The server closed the connection because the client took too long to send the complete request.

## Meaning

408 is about the client being slow to *send* the request (headers/body), not the server being slow to respond (that’s 504). It’s common on slow mobile networks, large uploads, or idle keep-alive connections.

## Common causes

- Slow or unstable client network during upload
- Large request body with a short server body-read timeout
- Idle keep-alive connection reused after the server closed it
- Load balancer idle timeout shorter than the upload duration

## Quick fix

1. Retry the request (408 is generally safe to retry for idempotent requests)
2. Increase body/header read timeouts on the server or load balancer
3. Upload large files in chunks or via direct-to-storage (pre-signed URLs)

## Detailed fix by platform

### Nginx

1. Raise read timeouts:

```nginx
client_body_timeout 120s;
client_header_timeout 60s;
```


### Apache

1. Increase `Timeout` and check `mod_reqtimeout` (`RequestReadTimeout header=20-40,MinRate=500 body=60,MinRate=500`).

## How to diagnose

1. **Payload** — Is this a large upload or slow client?
2. **Timeouts** — Which hop (LB, proxy, server) has the shortest read timeout?
3. **Retry** — Does an immediate retry succeed?

---
Source: [WhatIsMyError](https://whatismyerror.com/errors/http/408-request-timeout)
