The short version
- 01As a visitor, wait and retry later — the delay is at the site's own server.
- 02If you own the site, log response times to find which endpoint is slow.
- 03Ask your hosting provider about long-running processes or server overload.
- 04Replace a long synchronous request with a job plus status polling.
- 05Move genuinely lengthy processes to a DNS-only subdomain that bypasses the proxy.
- 06On Enterprise, raise the read timeout through Cache Rules or the zone settings API.
Raw message
Error 524 — A timeout occurred
Likely causes
- A long-running process at the origin that has not finished in time.
- An overloaded origin server without the resources to respond promptly.
- A large data query that takes longer than the read timeout allows.
- An endpoint that is slow only under particular conditions or particular inputs.
Connected, then silent
Cloudflare defines error 524 as a timeout that occurs when Cloudflare successfully connected to the origin web server, but the origin did not provide an HTTP response before the default 125 seconds.
That first clause is the diagnosis. The network is fine, DNS is fine, the firewall is letting Cloudflare through, and the server is up and accepting connections. It simply has not finished producing an answer. This is a performance problem wearing the costume of a connectivity problem.
Cloudflare publishes the timeouts alongside it: a read timeout of 125 seconds by default, adjustable on Enterprise plans, and a write timeout of 30 seconds which is not adjustable — 6.5 seconds in the case of Cloudflare Images. It also notes to expect roughly a second of variance between the configured timeout and the actual one.
If you are a visitor
Cloudflare is explicit that this error typically requires action from the site owner rather than visitors. Nothing about your browser, your connection or your device contributed to it, and nothing you change will shorten the time that server takes to respond.
Retrying later is reasonable here in a way it is not for some other codes, because Cloudflare's named causes include an overloaded origin and large queries. A request that timed out under load can succeed on a quieter server an hour afterwards.
If it happens consistently on one particular page or action — a report, an export, a search — that is worth telling the site owner, because it points them straight at the slow endpoint.
If you own the site
Cloudflare's causes are all about time: long-running processes, an overloaded origin lacking resources, and large data queries requiring extended processing. Its first practical step is to log response times to identify slowness, which is the only way to find out whether one endpoint is responsible or the whole server is struggling.
The architectural fix Cloudflare recommends is to implement status polling for large HTTP processes — start the work, return immediately with a job identifier, and let the client poll for completion. A request that cannot reliably finish inside two minutes should not be a synchronous HTTP request at all, and raising the timeout only defers the problem.
Cloudflare's other options are worth knowing in order. You can move lengthy processes to a DNS-only subdomain, which takes them out of the proxy and out of the timeout entirely. On Enterprise plans you can increase the timeout through Cache Rules — up to 6,000 seconds — or the zone settings API. And Origin Analytics will show you response times so you can optimise the slow endpoints rather than guess at them.
If it is the whole server rather than one endpoint, that is a hosting conversation: Cloudflare's step is to contact your provider about long-running processes or server overload.
Sources and review notes
Checked on . Based on Cloudflare's support documentation for error 524. The 125-second read timeout, the 30-second write timeout, the 6,000-second Enterprise ceiling and the resolution list are as published and are subject to change — check Cloudflare's current page before relying on a figure. The framing of 524 as a performance problem, and the argument against raising the timeout, are editorial. Glitchary is independent of Cloudflare.