503 Service Unavailable

HTTP response status code 503 Service Unavailable is a server error indicating the server is temporarily unable to handle the request due to maintenance or overloading.

Table of Contents

Usage

The 503 Service Unavailable status code means the server is unable to fulfill the HTTP request for a temporary reason. The server is overloaded with a sudden surge of traffic, the resource is locked, the server is rebooting, or a firewall misconfiguration is blocking the request.

Many sites return the 503 Service Unavailable status code during site maintenance, which is the recommended approach to ensure search rankings are not affected.

When the server is aware of the problem, such as a locked or overburdened resource, the Retry-After HTTP header informs the client when to resubmit the request.

Amazon S3 returns 503 Slow Down when request rates exceed the per-prefix throughput limits. A sudden spike beyond these thresholds triggers throttling while S3 scales partitions in the background. Distributing objects across multiple prefixes, implementing exponential backoff, and ramping up request rates gradually avoids the error. S3 does not charge for 503 responses.

Fastly returns 503 Service Unavailable with the reason text "Loop detected" when a request appears to route back to the same Fastly service or has transited too many Fastly servers. Fastly tracks request hops using the Fastly-FF and CDN-Loop headers. A request is rejected as a loop when the request exceeds any of these limits: more than three prior visits to the same POP for the same service, more than six unique Fastly services in the request path, or more than twenty total hops across all services. Loops commonly occur when the same hostname is configured as both the domain and the backend for a service, and DNS for the hostname resolves to Fastly. Using a separate origin hostname (e.g., origin.example.re) or an IP address for the backend avoids the loop.

SEO impact

Googlebot treats 5xx errors as a signal to temporarily reduce the crawl rate across the site. Previously indexed URLs are preserved initially, but extended downtime causes Google to drop them from the index. Any content returned with a 503 response is ignored. Google treats 500, 502, and 503 identically. 503 has no special advantage over other 5xx codes for planned maintenance despite common advice suggesting otherwise. A 5xx error on the robots.txt file is far more severe, even when every other page on the site returns 200. Google halts crawling for up to 12 hours, then falls back to a cached copy for up to 30 days. If the error persists, Google fully deindexes the site. Bingbot respects Retry-After on 503 responses for scheduling return visits. Persistent 503 responses prevent indexing in Bing.

Example

The client requests a resource and the server responds with a 503 Service Unavailable because the resource is temporarily unavailable. The Retry-After header suggests resubmitting the request after 30 minutes.

Request

GET /latest_quotes HTTP/1.1
Host: www.example.re

Response

HTTP/1.1 503 Service Unavailable
Content-Type: text/html; charset=UTF-8
Content-Length: 124
Retry-After: 1800

<html>
  <head>
    <title>Resource Busy</title>
  </head>
  <body>
    <p>Try again after 30 minutes.</p>
  </body>
</html>

How to fix

Determine whether the 503 is intentional (maintenance mode) or unplanned (overload, crash). An intentional 503 paired with a Retry-After header requires no debugging, only waiting for maintenance to finish.

For unplanned 503 errors, check these common causes:

  • Server resource exhaustion. Run top, htop, or vmstat on the server. High CPU or memory usage means the server lacks capacity. In Apache, the error log shows "server reached MaxRequestWorkers setting" when all worker slots are occupied. Raise MaxRequestWorkers (and ServerLimit for prefork MPM) in the Apache configuration. In nginx, increase worker_connections in the events block.

  • Application pool crash. PHP-FPM, Gunicorn, or a similar process manager exiting under load leaves the web server with no backend to serve requests. Check process manager logs, then increase pool size (pm.max_children in PHP-FPM, --workers in Gunicorn).

  • Connection pool exhaustion. Database or external API connection pools running dry force the application to reject new requests. Increase pool size in the application configuration and confirm connections are properly released after use.

  • Rate limiting. Nginx limit_req and limit_conn modules return 503 (or a custom status) when thresholds are exceeded. Review limit_req_zone and limit_conn_zone directives. Raise burst values or add nodelay to absorb traffic spikes.

  • Load balancer health check failure. A failing health check pulls the server from rotation. In AWS ALB/NLB, check target health in the console. In nginx upstream blocks, review max_fails and fail_timeout settings.

  • Deployment-triggered restart. Rolling deployments with too few instances cause brief 503 windows. Increase minimum healthy instances or use blue-green deployments.

  • AWS S3 throttling (503 Slow Down). S3 throttles requests exceeding the per-prefix request rate limits. Distribute objects across multiple prefixes using randomized key prefixes (e.g. hex-hash/original-key). Implement exponential backoff starting at 1 second. Ramp up request rates gradually instead of sending burst traffic. The AWS SDK retries 503 SlowDown responses automatically with backoff when properly configured.

For traffic spikes, add a CDN or Caching layer to offload static requests. Set up monitoring and alerting on 503 rates so resource constraints surface before a full outage.

Code references

.NET

HttpStatusCode.ServiceUnavailable

Rust

http::StatusCode::SERVICE_UNAVAILABLE

Rails

:service_unavailable

Go

http.StatusServiceUnavailable

Symfony

Response::HTTP_SERVICE_UNAVAILABLE

Python3.5+

http.HTTPStatus.SERVICE_UNAVAILABLE

Java

java.net.HttpURLConnection.HTTP_UNAVAILABLE

Apache HttpComponents Core

org.apache.hc.core5.http.HttpStatus.SC_SERVICE_UNAVAILABLE

Angular

@angular/common/http/HttpStatusCode.ServiceUnavailable

Takeaway

The 503 Service Unavailable status code is a server error returned when an HTTP request fails due to a temporarily unavailable resource.

See also

Last updated: March 5, 2026