5xx Server Error: What It Means, Who Broke It and How to Fix It

A 5xx server error means the website’s server received your request but could not complete it: the problem is on the website’s side, not the visitor’s. The HTTP standard (RFC 9110) puts it plainly: a 5xx code means “the server is aware that it has erred or is incapable of performing the requested method.” Which server, and why, is what the last two digits and this guide tell you.

Every page you open travels through a chain of machines: your browser, often Cloudflare or a CDN, the web server, the PHP code that runs WordPress, and the database. A 5xx is a message from the one link that gave up. Find that link and the fix is usually obvious, so this guide is built around the chain: decode your exact error, see what Google does while it lasts, fix it in the order that finds the cause fastest, then prove the fix the way Googlebot sees it.

What does a 5xx server error mean, and how is it different from a 4xx?

HTTP status codes are three-digit answers a server gives to every request. The first digit is the family: 2xx means “here you go”, 3xx means “it moved”, 4xx means “your request was the problem”, and 5xx means “I was the problem”. A 404 is a wrong address; a 500 or 503 means the address was fine and the server still could not deliver. That decides who fixes it: a 4xx by changing links or permissions, a 5xx by someone with access to the server, hosting account or code. One exception: Google says it treats a 429 (“too many requests”) as a server-overload signal and counts it as a server error, even though it sits in the 4xx family.

5xx error codes list: what each one means and whose job it is

You rarely need all of them. Four codes cover almost every real incident on a business website: 500, 502, 503 and 504. The rest are listed so you can recognise them when a host, a plugin or Cloudflare shows you one.

Definitions follow RFC 9110 (June 2022) and MDN; 509 is absent from both, which is why it is marked non-standard. The 508 “Resource Limit Is Reached” meaning comes from LiteSpeed’s CloudLinux documentation.

Find your code: the 5xx decoder

Pick the number you are seeing. You get the plain meaning, which link broke, whose job it is and your first three moves.

Why 5xx errors happen: the five links in the chain

Almost every 5xx on a small-business site comes from one of five places, and each has a tell.

1. Cloudflare or a CDN cannot reach your server (a 52x code). Cloudflare also notes that the cause of a 5xx is not always in your server’s logs, because the failure may be on an intermediate machine.

2. The web server is misconfigured or out of resources. A broken rule in .htaccess produces an instant 500 on Apache or LiteSpeed. Hitting the account’s process or memory limits produces 500 or 503 (LiteSpeed’s CloudLinux docs); Hostinger’s help page says a plan that reaches its process or RAM limit shows visitors a 503, while a CPU limit only makes the site sluggish.

3. PHP, the code that runs WordPress, crashed. A plugin update with a fatal error, a theme function removed in a new PHP version, or a script that needs more memory than allowed. WordPress only tries to give itself 40 MB of PHP memory on a single site (64 MB on multisite), which is why memory errors appear on heavy page builders and import tools.

4. Something took longer than a timer allows. This is the 504 and Cloudflare 524 story: every machine in the chain has a stopwatch, and the shortest one wins. PHP stops a script after 30 seconds by default; Nginx and Apache wait 60 seconds for the machine behind them; Cloudflare waits 19 seconds to open a connection and 125 seconds for a response. Google’s Search Console help puts it in one line: “Dynamic pages can take too long to respond, resulting in timeout issues.”

5. The database refused the connection. Too many connections, a crashed table or wrong credentials after a migration. WordPress shows its “Error establishing a database connection” page instead of your site.

Do 5xx errors hurt SEO? What Google actually does

Yes, in stages, and Google documents each one. The moment Googlebot starts receiving 5xx responses it slows down: Google’s HTTP status documentation says “5xx and 429 server errors prompt Google’s crawlers to temporarily slow down with crawling”, and that the slowdown “is proportionate to the number of individual URLs that are returning a server error.” Any content served with a 5xx is ignored, so a page that returns 500 with your beautiful HTML behind it is, to Google, an empty page.

Pages already in the index are protected at first: “already indexed URLs are preserved in the index, but eventually dropped.” Google publishes no number of days, only that its indexing pipeline “removes from the index URLs that persistently return a server error”. The Crawl Stats help page adds one concrete limit: do not return 503 or 429 “for more than two or three days”, or Google may crawl the site less often in the long term. After recovery the crawl rate comes back “gradually”, not instantly.

There is one file where the rules are stricter: robots.txt. Google’s robots.txt specification says that if that one file returns a 5xx, “for the first 12 hours, Google stops crawling the site but keeps trying to fetch the robots.txt file.” For the next 30 days Google uses the last good copy it saved; after that, a site with general availability problems stops being crawled. This is why Google’s maintenance guidance says never to serve a 503 for robots.txt: it blocks all crawling, not just one page.

How to find 5xx errors on your website

You only notice the errors you happen to see. Google and your server see all of them, so start there.

Search Console, Page indexing report. Open Indexing → Pages and look for the reason “Server error (5xx)”: in Google’s words, “Your server returned a 500-level error when the page was requested.” Click the row for the URL list and trend. Google adds that “server errors can be transient, so your live test might succeed when Google’s crawl failed”: a URL that loads for you now may still have failed at 3 a.m. when Googlebot visited.

Search Console, Crawl stats report. Open Settings → Crawl stats → Open report. The “By response” table shows the share of Googlebot’s requests that got a 5xx, and Host status flags availability problems; a green status means no significant crawl availability issues in the past 90 days.

Your hosting error log. In cPanel it is under Metrics → Errors; on other panels look for “error log” in the website’s management area. This log names the file and line that failed, which Search Console never will. The WordPress debug log (Step 3) does the same for PHP errors.

A crawler, for pages Google has not reached yet. Screaming Frog’s SEO Spider crawls up to 500 URLs free; filter Response Codes for “Internal Server Error (5XX)”. For a quick outside-in check of a page or a list of links, the free Broken Link Checker and Redirect Checker show the exact status code each URL returns right now.

How to fix a 5xx server error, step by step

Work down the chain from the outside in. Each step either clears the error or tells you which link to look at next, so you never spend an hour on the wrong machine.

Step 1. Confirm the error and read the exact code

Reload the page once, then open it in a private window so your browser’s cache and cookies are out of the picture. If it still fails, read the number: the browser tab, the page itself or your host’s error page will show it. Then check your hosting company’s status page, because a host-wide incident needs no work from you beyond waiting.

To see the real status code rather than a friendly error page, ask the server directly from a terminal (Terminal on Mac, PowerShell on Windows):

curl -I https://www.example.com/the-page/

The first line of the answer is what Google and every visitor get, for example HTTP/2 503. If you would rather not use a terminal, paste the URL into the free Redirect Checker, which prints the same status line.

Step 2. Work out which link in the chain broke

Three quick tells narrow it down in minutes.

Is Cloudflare in front? If the error page has Cloudflare’s branding and a Ray ID, the 502 or 504 came from your server; Cloudflare’s docs say a blank, unbranded page means Cloudflare itself produced it. A 52x code always means the trouble is between Cloudflare and your origin: 521 means your server refused the connection (often a firewall blocking Cloudflare’s IP ranges), 522 means it did not answer in 19 seconds, 525 and 526 are SSL certificate problems on your server, 530 means your origin hostname did not resolve.

Does the admin area load while the front end fails? Then PHP and the database are alive and the problem is in the theme or a front-end plugin. If /wp-admin/ fails too, suspect .htaccess, PHP memory or a plugin that loads everywhere. One page or every page? One URL points at that page’s content or a plugin only it uses; every page points at config, hosting limits or the database.

Step 3. The WordPress fix ladder for a 500 error

WordPress’s own documentation says the most likely cause of an Internal Server Error is “a corrupted .htaccess file”, and it gives the order below. Take a backup first, then work down. Stop as soon as the site comes back; the last thing you changed is the cause.

3a. Reset .htaccess. Connect with your host’s file manager or FTP, find .htaccess in the site’s root folder and rename it to .htaccess_old. Reload. If the site is back, go to Settings → Permalinks and click Save to generate a clean file. If you need to recreate it by hand, this is the default WordPress block:

# BEGIN WordPress
<IfModule mod_rewrite.c>
RewriteEngine On
RewriteBase /
RewriteRule ^index\.php$ - [L]
RewriteCond %{REQUEST_FILENAME} !-f
RewriteCond %{REQUEST_FILENAME} !-d
RewriteRule . /index.php [L]
</IfModule>
# END WordPress

3b. Deactivate all plugins. If you can log in, deactivate everything, then reactivate one at a time until the error returns. If you cannot log in, rename the /wp-content/plugins folder to plugins_off via FTP; WordPress will treat every plugin as deactivated. Rename it back and reactivate one by one.

3c. Switch to a default theme. Rename your active theme’s folder inside /wp-content/themes; WordPress falls back to a default theme such as Twenty Twenty-Four if one is installed. A site that comes back here has a theme problem, often after a PHP version change.

3d. Raise the PHP memory limit. Add this above the line that says “That’s all, stop editing!” in wp-config.php. The values are the ones WordPress documents; some hosts ignore these lines, in which case the host must raise the limit for you.

define( 'WP_MEMORY_LIMIT', '96M' );
define( 'WP_MAX_MEMORY_LIMIT', '128M' );

3e. Turn on the debug log to read the real error. Also in wp-config.php. Reload the broken page, then open /wp-content/debug.log: the last lines name the plugin or theme file and the line number. Remove these lines when you are finished.

define( 'WP_DEBUG', true );
define( 'WP_DEBUG_LOG', true );
define( 'WP_DEBUG_DISPLAY', false );

3f. Re-upload the WordPress core. Download a fresh copy of WordPress and upload only the wp-admin and wp-includes folders over the existing ones. This repairs damaged core files without touching your content, themes or plugins.

Step 4. Fix the slow thing behind a 502, 504 or Cloudflare 524

These codes mean a machine in front waited for one behind it and gave up, so the question is never “why did Nginx time out?” but “what was PHP or the database doing for 60 seconds?”. The usual suspects: an import, a backup running in a browser tab, an uncached page with dozens of queries, a call to an outside service that hangs, or cron jobs piling up.

Start with the error log and the debug log from Step 3; a timeout usually leaves a “Maximum execution time exceeded” line. WordPress’s documentation gives the lift for that limit, in .htaccess on Apache or LiteSpeed (back the file up first):

php_value max_execution_time 60

Raising timers buys time; it does not fix the slow task. Cache public pages so the heavy work runs once, run imports and backups in smaller batches or from the server’s scheduler, and give outside services a short timeout so one slow vendor cannot hold your page. If a Cloudflare 524 only appears on admin tasks, bypass Cloudflare for /wp-admin/ with a page rule.

Step 5. Fix the limit behind a 503 or 508

On shared hosting a 503 or 508 is usually your account bumping into its ceiling, not a broken site. LiteSpeed’s CloudLinux documentation explains the pattern: CPU and disk limits make the site slower, while memory and process limits produce 500 or 503, and the concurrent-connection limit produces “508 Resource Limit Is Reached”. Hostinger’s own help page says a plan that reaches its process or RAM limit shows visitors a 503.

What eats those processes on a small site is rarely real customers: bots hammering search and filter URLs, WordPress cron jobs firing on every visit, uncached pages, and image or backup plugins working in the background. So: turn on page caching, block or slow aggressive bots at the server or Cloudflare level, move WP-Cron to a real server cron, and only then look at the plan. If you need the host, send everything they need in one message:

Step 6. Prove it is fixed, then ask Google to look again

A fix you cannot prove is a guess. Check in the same order Google will.

Two more things worth doing. For your most important URLs, click “Request indexing” after the live test so Google returns sooner, and remember Google says crawling resumes “gradually” after a fix, so a day or two of lower crawl activity is normal. Our guide on how to check if a page is indexed shows what each URL Inspection verdict means.

Planned maintenance: the right way to take a site down

If you must take the site offline, Google’s guidance for pausing an online business is specific: for an urgent one or two day shutdown, “return an informational error page with a 503 HTTP response status code instead of all content”, and “use the retry-after HTTP header with a best effort date or duration.” It calls a site-wide 503 “an extreme measure that should only be taken for a very short period of time (a few days at most)”, warns that Google cannot refresh titles, descriptions or structured data while a page returns 503, and says never to return 503 for robots.txt.

The common mistake is the opposite: an “Under construction” or error page served with a 200 code. Google has warned about this since 2011: it sees many sites returning 200 for server errors and placeholder pages, which tells Google the placeholder is your real content. Returning the right code is one line in most maintenance plugins and worth checking with the status test from Step 1.

How to stop 5xx errors coming back

Most repeat incidents come from four gaps: no uptime monitor (so Google learns about the outage before you do), updates applied straight to the live site instead of a staging copy, no page caching or bot rule on a shared plan, and nobody looking at Search Console’s Crawl stats host status, Google’s own 90-day record of whether your server kept up.

For a second pair of eyes, the free Website SEO Audit checks status codes, response time and crawlability of your key pages in one pass, and the manual website audit service goes through server logs, Search Console history and hosting configuration when errors are intermittent. For the wider crawl picture, read what crawl budget is and robots.txt best practices.

FAQs

Why does Search Console show a 5xx error when my site works fine?

Because the error happened when Googlebot visited, not when you did. Google says server errors can be transient and a live test can pass while the earlier crawl failed. Check your host’s error log for the time of the crawl, look for overnight backups or cron jobs, and watch the Crawl stats report for repeats.

What is the difference between a 502, 503 and 504 error?

A 503 is the server saying it is overloaded or in maintenance and expects to recover. A 502 and a 504 are reported by a machine in front of your server: 502 means it received an invalid answer from the server behind it, 504 means it waited and got no answer in time. For 502 and 504, look at the machine behind the one that reported it.

How long until Google removes pages that return a 5xx error?

Google does not publish a fixed number of days. Its documentation says indexed URLs are kept at first “but eventually dropped”, and that URLs which persistently return a server error are removed. For maintenance codes, Google advises not returning 503 or 429 for more than two or three days. Fix it in hours, not weeks, and crawling returns gradually.

Is a 5xx error my fault or my hosting company’s?

It depends on the link that broke. A 500 after a plugin or theme update is on the site; a 503 or 508 from resource limits is shared, since your site caused the load but the host sets the ceiling; a host-wide outage is theirs. Use the decoder above and Step 2 to name the link before you open a ticket, and send the exact code, time and URLs.

What does “5xx server error” mean on Instagram?

It means Instagram’s own servers failed to handle the request; it is not your phone, account or internet connection. There is nothing to fix on your side. Wait a few minutes, close and reopen the app, and try again. If it continues for everyone, Instagram is having an outage and will restore service itself.