503 Service Unavailable
The server is temporarily unable to handle the request, typically because it is overloaded or down for maintenance.
Status
HTTP/1.1 503 Service Unavailable
Details
- Category: 5xx
- Status: Standard. Registered with IANA and defined by a current standards-track RFC.
- Specification: RFC 9110 §15.6.4
- Cacheable: Only with explicit cache headers
- Response body: Allowed
- Retry: Retry after a delay
503 says the condition is temporary and expected to clear. It is the correct code for planned maintenance, for shedding load under pressure, and for a service whose dependencies are unavailable. Retry-After tells the client when to come back and, during maintenance, tells search engines not to deindex the page.
It is also the standard response from anti-bot and DDoS protection layers while they evaluate traffic, which is why automated clients meet it during challenge flows.
Note: Send Retry-After during maintenance. Without it, crawlers can treat a long outage as content removal rather than a temporary condition.
Headers
- Retry-After: Seconds or an HTTP date. Important for both clients and crawlers during maintenance windows.
Common causes
- Planned maintenance with the application deliberately serving a maintenance page.
- Overload: worker pool exhausted, queue full, or an autoscaler still catching up.
- A critical dependency such as a database or cache is unavailable.
- Load shedding or a circuit breaker refusing new work to protect the system.
- A protection layer holding the request while it evaluates the client.
How to fix it
As the client
- Honour Retry-After and back off. Retrying hard against an overloaded server makes recovery slower.
- Add jitter to retries so a fleet of clients does not return in a synchronized wave.
- Reduce concurrency rather than only spacing retries, since 503 under load is usually about parallelism.
As the server
- Always send Retry-After during maintenance. Without it, crawlers may treat the outage as content removal.
- Serve a static maintenance page from the edge so the origin is not doing work while it is meant to be resting.
- Scale the bottleneck rather than the whole stack, and check queue depth before adding capacity.
Examples
Planned maintenance
A request during a maintenance window:
HTTP request
GET / HTTP/1.1 Host: www.example.com
Retry-After tells both clients and crawlers that the condition is temporary:
HTTP response
HTTP/1.1 503 Service Unavailable Retry-After: 120 Content-Type: text/html; charset=utf-8
Notes for proxy users
A 503 that appears only for automated traffic and clears for a browser is usually a protection layer rather than an outage. Lower concurrency, add delay between requests, and distribute across more addresses before assuming the site is down.