Plugins Plugin hub Backup SEO Caching Security Forms WooCommerce Images Best backup plugins
Hosting Hosting directory Shared Managed Cloud India
Themes & builders Compare Rank Math vs Yoast
Tools All tools Hosting finder Plugin stack builder Cost calculator Salt generator Theme detector Who is hosting this? Speed test Headline analyzer robots.txt generator Redirect generator Uptime calculator Blog name generator
Best lists Best backup plugins Deals
Guides All guides Speed up WordPress Choose hosting Back up WordPress
How we test

Troubleshooting · Explainer

What Causes the 503 Service Unavailable Error in WordPress?

Short answer

A 503 Service Unavailable error means the server is not ready to handle the request. In WordPress that usually means the site is in maintenance mode, the server is overloaded, the PHP worker pool is full, or a resource limit such as memory, CPU or database connections was hit. Unlike a 500, a 503 says the problem is temporary, so the server may also send a Retry-After header.

Key takeaways

  • A 503 says the server is not ready, not that it is broken for good.
  • The maintenance message means a stuck .maintenance file, and nothing else shows that exact text.
  • A 503 under traffic spikes points at overload, a full PHP worker pool, or a hit resource limit.
  • A Retry-After header is the server's own estimate of when to try again.
  • A 503 should not be cached, or visitors may keep seeing the error page after the fix.

What does a 503 Service Unavailable error mean in WordPress?

A 503 Service Unavailable error means the server is not ready to handle the request.

MDN defines the 503 status this way and names two common causes: a server down for maintenance and a server that is overloaded (MDN 503 status).

The word “temporary” matters. A 503 says the service should come back, which is why the server may send a Retry-After header with an estimate of when to try again. MDN notes that the header should hold the estimated recovery time when possible.

In WordPress, a 503 has a narrower set of usual causes than a 500. The site is often in maintenance mode, the server is out of PHP workers, or a resource limit was hit. Each of these looks different, and the next sections show how to tell them apart.

What are the most common causes of a 503 in WordPress?

The most common WordPress causes are maintenance mode, server overload, a full PHP worker pool, and a hit resource limit.

Maintenance mode is the simplest cause. WordPress creates a .maintenance file in the site root during core, theme and plugin updates.

While the file exists, visitors see "Briefly unavailable for scheduled maintenance. Please check back in a minute." The WordPress common errors guide says the file may not be removed properly after an update, and deleting it from the site root clears the message.

Server overload is the second cause. MDN explains that some server software rejects requests with a 503 when resource limits such as memory, CPU or connection pool limits are met, which creates backpressure and stops the server from running out of resources entirely (MDN 503 status).

A full PHP worker pool is the third cause. When every PHP worker is busy, new requests cannot start and the server answers with a 503. A slow plugin, a stuck cron job, or a traffic spike can fill the pool.

Resource limits form the fourth group. A database that hit its connection limit, a host plan that capped CPU use, or a PHP memory limit that is too low can each produce a 503. WordPress tries to give PHP 40MB on a single site and 64MB on multisite, per the wp-config handbook.

How can you tell which cause you have?

Start with the message on the page and the timing of the first error.

The “Briefly unavailable for scheduled maintenance” message means the .maintenance file is still there. If the message appeared during an update and stayed, the update likely failed partway and the file was never removed.

ClueLikely causeWhere to check
Maintenance message on every pageStuck .maintenance fileSite root over FTP
Error under traffic spikes onlyOverload or full worker poolHost CPU and worker stats
Error on one heavy actionMemory or connection limitPHP error log
Error after an updateFailed update, stuck maintenanceSite root, update log
Error with a Retry-After headerPlanned maintenance or rate limitResponse headers
Whole server unreachableHost outageHost status page

If the error comes and goes with traffic, overload is the likely cause. If it is constant since an update, maintenance mode is the likely cause. If it hits one action every time, a limit on that action is the likely cause.

The fix steps for the WordPress side are in our guide on how to fix the 503 Service Unavailable error.

Why does maintenance mode cause a 503?

Maintenance mode causes a 503 because WordPress deliberately blocks visitors while files are being replaced.

When WordPress starts a core, theme or plugin update, it writes a .maintenance file to the site root. While that file exists, WordPress answers every request with the maintenance message and a 503 status. This stops visitors from loading a half-updated site.

The file should be deleted when the update finishes, which usually takes seconds. The WordPress common errors guide notes that the file may not be removed if the update fails, and the fix is to delete the file from the site root over FTP.

The clue is the message itself. If the page says “Briefly unavailable for scheduled maintenance”, the cause is maintenance mode and nothing else.

No other 503 cause shows that exact message. Our guide on how to fix a site stuck in maintenance mode walks through the check.

How does server overload cause a 503?

Server overload causes a 503 when the server refuses new requests to protect itself from running out of memory, CPU or database connections.

MDN describes this as backpressure: dropping requests stops the server's resources from being exhausted and avoids a worse failure (MDN 503 status). A server that accepted every request during a spike would slow down for everyone and might crash.

A traffic spike is the most common trigger. A post shared widely, a sale, or a bot crawl can push a small plan past its limits.

The WordPress common errors guide describes the same pattern for timeouts: a site trying to do more than the server can manage, common on shared hosting with a restricted memory limit.

A slow plugin can cause the same overload without a spike. A plugin that makes a slow remote call on every page load holds a PHP worker for seconds, and a handful of visitors can fill the whole pool.

The host control panel usually shows CPU and memory use, and a flat line at the plan limit during the errors confirms overload.

What do PHP workers and resource limits have to do with a 503?

PHP workers and resource limits decide how many requests a WordPress site can handle at once, and a 503 appears when that number is passed.

On PHP-FPM setups, a fixed pool of workers handles PHP requests, one at a time. When all workers are busy, the next request has nowhere to go. Some servers answer that request with a 503 instead of making the visitor wait.

Resource limits work the same way. MDN lists memory, CPU and connection pool limits as the thresholds that trigger a 503 (MDN 503 status). A database that ran out of connections, a host plan that capped CPU minutes, or a PHP memory limit set too low can each trip the same response.

PHP ships with a default memory_limit of 128M, according to the PHP manual. Hosts can set it lower, and the WordPress handbook notes that some hosts set the PHP limit as low as 8MB (wp-config handbook).

The clue is the pattern. A 503 that appears at the same time each day, or only when traffic peaks, points at a limit. A 503 that appears on a quiet site at 3 a.m. points at a stuck cron job or a runaway script instead.

If CPU use is the limit, our guide on how to fix high CPU usage covers the WordPress side.

What does the error log tell you about a 503?

The PHP error log shows whether a script hit a limit, while the server log shows whether the server refused the request.

WordPress writes PHP errors to wp-content/debug.log when WP_DEBUG and WP_DEBUG_LOG are true in wp-config.php, with WP_DEBUG_DISPLAY set to false so errors stay out of the page HTML (WordPress debugging handbook).

A log line that says “Allowed memory size exhausted” points at the memory limit. A log line that names a plugin and a slow remote call points at that plugin. A log with no entries at the time of the error points at the server refusing requests before PHP ran.

The host control panel often shows resource graphs for CPU, memory and database connections. A graph that hits its ceiling at the same time as the errors is strong evidence. If the error came with a Retry-After header, note its value: it is the server's own estimate of the wait.

Cloudflare's 5xx guide advises giving the host the error code, the time and timezone, and the failing URL, and checking the logs of any load balancer, cache, proxy or firewall between the visitor and the origin (Cloudflare 5xx errors).

What mistakes should you avoid?

Do not confuse a 503 with a 429, and do not delete files you have not identified.

A 429 Too Many Requests means rate limiting for a specific client, while a 503 means the server is not ready for anyone. MDN notes that rate limiting for specific clients should use 429, not 503 (MDN 503 status).

The two need different fixes, so check the status code before acting. Our guide on how to fix the 429 Too Many Requests error covers the rate-limit case.

Do not delete the .maintenance file while an update is still running. Wait a few minutes first. Deleting it mid-update can leave half-written files behind and turn a short maintenance window into a broken site.

Do not raise limits blindly. The wp-config handbook warns that raising the memory limit can hide the real fault, which returns as you add more plugins. Check the log and the resource graphs first.

Finally, do not cache the error page. MDN warns that a 503 response should not usually be cached, since visitors may keep seeing the error page after the fix (MDN 503 status). If your cache plugin or CDN cached the 503, purge the cache after the fix.

Frequently asked questions

What is the difference between a 503 and a 500 error?

A 503 means the server is not ready to handle the request and should recover, often because of maintenance or overload. A 500 means the server hit an unexpected condition and could not finish the request. A 503 points at capacity or maintenance, while a 500 points at a fault in code, config or files.

Can a 503 error fix itself?

Often, if the cause was a short maintenance window or a brief traffic spike. The server may send a Retry-After header with an estimate. If the error stays for more than a few minutes, check for a stuck .maintenance file, a full PHP worker pool, or a resource limit that was hit.

Should I wait before deleting the .maintenance file?

Yes, if the update is still running. WordPress creates the .maintenance file during updates and should delete it when done. Wait a few minutes first. If the message stays after ten minutes, the update likely failed and the file should be deleted from the site root over FTP.

Is a 503 error always a hosting problem?

Not always. A short 503 during an update is normal. A 503 that repeats under traffic, blocks wp-admin, or lasts more than a few minutes needs attention. Repeated 503s point at a plan that is too small, a slow plugin holding workers, or a limit that keeps being hit.

Should I contact my host about a 503 error?

Yes, if the error is site-wide, repeats under traffic, or the resource graphs show a limit being hit. Give the host the error code, the time and timezone, the failing URL, and the last lines of your error log. That detail lets them check server logs you cannot see.

Sources

  1. 503 Service Unavailable MDN Web Docs
  2. Common WordPress Errors WordPress Developer Resources
  3. Debugging in WordPress WordPress Developer Resources
  4. Editing wp-config.php WordPress Developer Resources
  5. Description of core php.ini directives PHP Manual

Is your product missing, or something out of date?

Tell us what to add or correct. We read every note and reply when we need a source. A message does not buy a ranking.

About the editors

WPLegible editors

Researched from the primary sources listed above and checked against our editorial rules. About the site · How we test