503 Service Unavailable: Meaning and How to Fix It

A 503 Service Unavailable error means the web server is reachable but temporarily cannot handle your request, usually because it is overloaded, down for maintenance, or waiting on a backend service that has failed. The problem is on the server side, not your device, and it is meant to be temporary.
If you are visiting the site: wait a minute and reload. If it persists for more than a few minutes, the site owner has to fix it.
If you run the site: check, in this order, whether you are in maintenance mode, whether the server has run out of CPU, memory or worker processes, and whether a backend service (application server, database, upstream API) is down. The sections below walk through each one.
Quick fixes at a glance
| Who you are | Try this | Why it helps |
|---|---|---|
| Visitor | Wait 1 to 5 minutes, then reload | Most 503s clear on their own once load drops |
| Visitor | Check whether it is down for everyone | Rules out your own network or cache |
| Site owner | Turn off maintenance mode | A forgotten maintenance flag returns 503 to everyone |
| Site owner | Check CPU, memory and worker limits | Exhausted resources are the most common cause |
| Site owner | Restart the application server or app pool | Clears crashed or hung workers |
| Site owner | Check backend services and their logs | A failed dependency often surfaces as a 503 |
| Site owner | Check your CDN, load balancer or host status page | The 503 may come from a layer in front of your server |
What a 503 error actually means
The HTTP standard (RFC 9110) defines 503 as a server that is "currently unable to handle the request due to a temporary overload or scheduled maintenance". Two words matter there: temporarily and unable. The server is running and answered you, it just chose not to process the request right now.
That makes 503 different from most other server errors:
| Code | Name | What it means | How it differs from 503 |
|---|---|---|---|
| 500 | Internal Server Error | The application crashed while handling the request | A bug or failure, not deliberate unavailability |
| 502 | Bad Gateway | A proxy got an invalid response from the server behind it | The upstream answered badly |
| 503 | Service Unavailable | The server cannot handle requests right now | Temporary, and can tell clients when to retry |
| 504 | Gateway Timeout | A proxy waited too long for the server behind it | The upstream did not answer in time |
| 429 | Too Many Requests | This client is being rate limited | Applies to one client, not everyone |
You may also see the same error worded differently, such as "503 Service Temporarily Unavailable", "HTTP Error 503. The service is unavailable" (IIS) or "Service Unavailable: The server is temporarily unable to service your request" (Apache). They all mean the same thing.
Fixes for visitors
Because a 503 comes from the server, there is little a visitor can fix, but these steps rule out your side quickly:
- Wait and reload. Give it a minute or two, then refresh. Overload and short restarts usually clear quickly.
- Check if it is down for everyone. Try the site on mobile data or another device, or use an online website uptime checker. If it fails everywhere, it is the site's problem.
- Do a hard refresh (Ctrl+F5, or Cmd+Shift+R on a Mac) or open the page in a private window, in case your browser cached the error page.
- Check the service's status page or social accounts for a maintenance or outage notice.
If the site is down for everyone, the only real fix is time. The owner is very likely already working on it.
Fixes for site owners
Work through these roughly in order. The first two cover most real cases.
1. Check whether maintenance mode is on
Many platforms return 503 on purpose during updates. WordPress shows "Briefly unavailable for scheduled maintenance" while it updates, and if an update fails part way, a leftover .maintenance file in the site root keeps returning that page. Delete the file and reload. Check for similar maintenance flags in your CMS, framework or load balancer.
2. Check server resources and worker limits
When CPU, memory or the pool of worker processes runs out, servers reject new requests with 503 rather than falling over completely.
- On Linux, run
toporhtopfor CPU and memory,df -hfor disk space, and check how many PHP-FPM, Gunicorn, Puma or Node workers are busy. - In your web server's error log, look for messages about reaching the maximum number of workers or children.
- If resources are maxed out because of a traffic spike, scale up or out, enable caching, or put a CDN in front of the site. If they are maxed out by a runaway process or bad query, fix that first.
3. Restart the application server or app pool
A crashed or hung application process can leave the web server with nothing to send requests to. Restart the application service (for example PHP-FPM, your Node or Python process manager, or the container) and watch whether it stays up. If it dies again, its logs will say why.
4. Check backend services and dependencies
If your application relies on a database, cache, authentication service or third-party API, a failure there often reaches users as a 503, especially behind a proxy or load balancer that has no healthy backends left to route to.
- Check each dependency's health and logs.
- Check your load balancer's target health. A failing health check endpoint can take every server out of rotation at once.
- Add timeouts, circuit breakers and fallbacks so one failing dependency degrades a feature instead of the whole site.
5. Check the layer in front of your server
The 503 may not come from your server at all. CDNs, reverse proxies, load balancers and hosting platforms generate their own 503 pages when they cannot reach your origin. The branding on the error page, or a response header such as Server or Via, usually tells you which layer sent it. Check that provider's status page and your origin's reachability from it.
6. Read the logs
Look at the error logs from the moment the 503s started:
- Nginx:
/var/log/nginx/error.log - Apache:
/var/log/apache2/error.log(or/var/log/httpd/error_log) - IIS: Event Viewer (Windows Logs, Application) and
C:\Windows\System32\LogFiles\HTTPERR - Your application's own logs, and any recent deploy or configuration change around the same time
A 503 that started at the exact time of a deploy, config change or certificate renewal is rarely a coincidence.
Fixing 503 errors on IIS
On Windows servers, "HTTP Error 503. The service is unavailable" almost always means the application pool behind the site has stopped.
- Open IIS Manager, select Application Pools, and check whether the site's pool is Stopped.
- Start it. If it stops again within seconds, IIS's Rapid-Fail Protection has disabled it after repeated crashes. Check Event Viewer for the reason.
- A common cause is an expired or changed password on the account the pool runs as. Update it under the pool's Advanced Settings, Identity.
- Other causes include a missing or broken IIS module after a Windows update (repair the URL Rewrite module, for example) and a conflicting URL reservation, which you can list with
netsh http show urlacl.
How to return a 503 properly during maintenance
If you take a site down on purpose, a 503 is the right status code, and sending a Retry-After header tells browsers, API clients and search engines when to come back. It can be a number of seconds or an HTTP date:
HTTP/1.1 503 Service Unavailable
Retry-After: 3600
Content-Type: text/htmlNginx:
location / {
add_header Retry-After 3600 always;
return 503;
}
error_page 503 /maintenance.html;
location = /maintenance.html {
root /var/www/html;
internal;
}The always flag matters: without it, Nginx does not add headers to error responses.
Apache (.htaccess):
ErrorDocument 503 /maintenance.html
RewriteEngine On
RewriteCond %{REQUEST_URI} !^/maintenance\.html$
RewriteRule ^ - [R=503,L]
Header always set Retry-After "3600"Keep the maintenance page short and useful: say what is happening, when you expect to be back, and where to get updates, such as your status page. For APIs, return a JSON body with the same information.
Does a 503 error hurt SEO?
Not if it is short. Google treats a 503 as a temporary signal, slows down crawling and comes back later, so a maintenance window of a few hours does no lasting harm. If a site keeps returning 503 for days, Google can start dropping pages from its index. That is why maintenance should use 503 (not 200 with an error message, and not 404), and why unplanned 503s need to be caught fast.
How to prevent 503 errors
- Load test before launches with a tool such as k6, Locust or JMeter so you know your capacity limits before real traffic finds them.
- Autoscale application servers and containers on CPU, memory or request queue length, with headroom for spikes.
- Cache aggressively at the CDN and application level so traffic spikes do not all reach your origin.
- Degrade gracefully by switching off non-essential features under load instead of failing every request.
- Plan maintenance for low-traffic hours, announce it on your status page and use a proper 503 with
Retry-After. - Monitor from outside so you hear about 503s before your users tell you.
Catch 503 errors before your users do
A 503 often appears in bursts, during a traffic spike or a failed deploy, and disappears before anyone checks. Odown checks your site and API endpoints from multiple locations every minute and alerts you by email, SMS, Slack, Discord, Telegram or webhook as soon as they start returning 5xx status codes. Pair it with a public status page so users know you are on it while you fix the cause.
For other errors in the same family, see our guides to common website errors and ERR_CONNECTION_REFUSED.
FAQ
What does 503 Service Unavailable mean? The server is online but temporarily cannot handle requests, usually because of overload, maintenance or a failed backend service. It is a server-side problem and is meant to be temporary.
How long does a 503 error last?
Usually seconds to minutes for overload or restarts, and the length of the maintenance window for planned downtime. If a Retry-After header is sent, it says how long to wait.
Is a 503 error my fault? Almost never. It comes from the website's server or a service in front of it. Waiting, reloading or checking from another network is all a visitor can do.
What is the difference between 502 and 503? A 502 means a proxy got a bad response from the server behind it. A 503 means the server, or the proxy itself, is temporarily unable to handle the request at all.
How do I fix a 503 error in WordPress?
Delete any leftover .maintenance file in the site root, then deactivate recently updated plugins or switch to a default theme to find the culprit. If the server is out of resources, raise your hosting plan's limits or add caching.


