HTTP 503 Service Unavailable

The server is temporarily unable to handle the request — overloaded, or down for maintenance. Often accompanied by Retry-After.

Defined in RFC 9110 §15.6.4 — 503 Service Unavailable · MDN reference

Where you meet HTTP 503 in production

Planned maintenance, a deploy in progress, a dependency that is down, or a load shedder deliberately turning traffic away to protect what is left.

Why you would test it

Unlike a 500, a 503 is explicitly retryable. Test that your client treats them differently instead of collapsing both into "error".

What your client should do about a 503

Retry with backoff, and honour Retry-After when the server sends one — a 503 with a Retry-After is a server telling you precisely when to come back, and ignoring it is how a recovering service is pushed back over. Fail open or degrade gracefully where the call is not essential.

503 versus the codes it gets confused with

• 503 vs 500 Internal Server Error — 500 is unplanned. 503 usually is not — it often means someone chose to return it.
• 503 vs 429 Too Many Requests — 429 limits you. 503 affects everyone.

Endpoints that return 503

2 endpoints in this playground answer with 503. Every one is free, needs no signup, and can be called from the browser or with curl.

• GET https://funapi.dev/api/protocol/v1/maintenance — Always, with a Retry-After header. Good retry/backoff practice. (Protocol Lab API) Open & run this endpoint
• GET https://funapi.dev/api/faults/v1/target — Configured outage (Fault Injection API) Open & run this endpoint

Questions about HTTP 503

Should maintenance pages be 503?

Yes. A maintenance page served as 200 tells search engines the thin page is the real content.

How long should I wait?

As long as Retry-After says. Without it, exponential backoff with jitter and an overall deadline.

Other server error codes

• 500 Internal Server Error
• 504 Gateway Timeout

All HTTP status codes · All 39 mock REST APIs · Getting started guide

Last updated 6 September 2026