How to Fix the Hostinger 503 Service Unavailable Error

What the 503 actually means

A 503 means the web server received your request and could not hand it to a process that was free to handle it. The site is not gone and the files are not missing. Something that should have answered your request was either busy, crashed or out of room.

On Hostinger, that something is almost always PHP. Hostinger runs LiteSpeed, and LiteSpeed serves each PHP request through a worker from a fixed pool. When every worker is occupied, incoming requests queue. When the queue overflows, LiteSpeed returns 503.

The fix is therefore not a single command. It is finding out which resource ran out, then either reducing demand or raising the ceiling.

The fastest triage test: load a static file — an image, a CSS file, a favicon — directly by URL. If static files load and PHP pages return 503, the server is healthy and your site is the bottleneck. If nothing loads at all, check Hostinger’s status page before you touch anything.

The causes, in order of likelihood

1. PHP worker exhaustion

This is the most common cause by a wide margin. Every plan comes with a fixed number of PHP workers, and that number is your concurrency ceiling.

Forty workers on Premium is a comfortable blog and a thin ecommerce site. It is not enough for a WooCommerce store with a large catalogue, an uncached membership site, or anything that makes outbound API calls inside a page request. Every slow page holds a worker for its entire duration. Ten requests that each take three seconds will occupy ten workers for three seconds.

2. A runaway plugin or bot traffic

Plugins cause 503s in two ways. A badly written plugin can loop, making the same query hundreds of times in a single request, which holds a worker for seconds instead of milliseconds. Or a plugin can make an external HTTP call on every page load, which turns a 50 ms request into a 2,000 ms request.

Bot traffic is the other half of this. Crawlers that ignore your robots file, scrapers, and badly configured uptime monitors can generate thousands of PHP requests per hour. On Premium, 40 concurrent bot requests is a full outage.

3. Memory limit hit

When a PHP process tries to allocate more memory than the plan allows, it dies. LiteSpeed sees a dead backend and returns 503. Premium caps PHP memory at 1,536 MB, Unlimited at 2,048 MB and Cloud Startup at 3,072 MB.

A single WordPress request rarely needs that much. A page builder rendering a complex layout, an image plugin resizing a large upload, or a backup plugin walking the database can all exceed it. The error appears in the log as an allowed memory size exhausted message.

4. A crashed PHP process

If PHP itself falls over — a segfault in an extension, a corrupted opcode cache — the backend disappears and LiteSpeed returns 503 for every request until the process restarts. This one resolves itself when the process manager restarts PHP, which is why “it fixed itself after five minutes” usually points here.

5. A genuine platform incident

The rarest cause. If Hostinger has a problem in the data center region your site sits in, every site there is affected. This is the only case where you wait.

Diagnosis, step by step

Check the resource usage graph

In hPanel, open your website’s management area and find the resource usage graph. It shows CPU, memory and entry processes over time. What you are looking for is shape, not numbers.

• Flat-topped peaks that touch the ceiling repeatedly: you are hitting a plan limit. This is structural and configuration will not fix it.
• One tall spike with normal values either side: a traffic event or a bot flood. Find the source.
• Sustained high memory with low CPU: usually a plugin holding a large dataset in memory.

Check whether the whole server is affected

Load a static asset directly. If it serves instantly while PHP pages fail, you have confirmed a PHP-side problem. If the static asset also fails, look at Hostinger’s status page and support chat before doing anything else.

Read the access log for bot floods

Hostinger keeps access and error logs for each site, reachable from the file manager or the logs area of hPanel. Sort by IP and by user agent and look for a single source generating hundreds or thousands of requests.

A typical bot flood looks like the same IP hitting many URLs in quick succession, or a user agent like a generic crawler requesting pages it has no reason to request. Blocking that IP or user agent in .htaccess, or at the CDN layer, often resolves the 503 within minutes.

Check the PHP error log

The error log tells you whether the cause is memory or a fatal error. Look for “Allowed memory size exhausted” (memory limit), “Maximum execution time exceeded” (a stuck request holding a worker), or a PHP fatal error naming a specific plugin file.

Disable plugins in bulk

Renaming the plugins folder through the file manager deactivates every plugin at once. If the site recovers immediately, the cause is a plugin. Reactivate them in small groups to identify which one.

The fixes that work

Enable caching first. On WordPress, install LiteSpeed Cache and turn on page caching. A cached page never invokes PHP, so it never occupies a worker. This is the single highest-leverage change available on any Hostinger plan. Pair it with the built-in CDN, which serves static assets from an edge location.

Cut the plugin count. Every active plugin adds code to every request. Sites running 30 plugins on Premium will hit 503 under load that a 10-plugin site handles without noticing.

Block the bots. Add rules to .htaccess to deny the offending user agents or IP ranges, or use the CDN’s firewall rules if your plan includes them. Also confirm that your own uptime monitor is not polling every 30 seconds — that is 720 PHP requests a day from one source, and it multiplies across monitors.

Raise the memory limit if that is the binding constraint. In the site’s PHP configuration area in hPanel you can raise the memory limit up to the plan maximum. Going beyond the maximum is not possible; that is what the upgrade is for.

Fix the slow query. If the graph shows CPU high and memory normal, look for a database query running on every page load. Database bloat and missing indexes are common on sites that have been running for years.

Restart PHP if you suspect a crash. Waiting a few minutes is usually enough, but support can restart the PHP service on request if the problem persists after a configuration change.

When to upgrade

Upgrade when the resource graph shows you saturating your plan on ordinary days, not on your worst day. The thresholds we use:

The upgrade path that most people take is Premium to Unlimited. It costs $1/mo more on the 48-month term and moves you from 40 PHP workers to 60, from 1 CPU core to 2, from 2 GB RAM to 3 GB, and from 20 GB SSD to 50 GB NVMe. It also switches backups from weekly to daily, which matters when you are debugging under pressure. Our Unlimited plan breakdown covers the detail.

If your 503s come with slow pages rather than outright failures, the problem may be performance rather than capacity — our 12 fixes for a slow Hostinger site works through that in priority order. If the site returns a blank page instead of a 503, start with the white screen guide instead.

A short prevention checklist

1. Install LiteSpeed Cache and enable page caching before you need it.
2. Turn on the built-in CDN.
3. Keep active plugins under about 15.
4. Set your uptime monitor to a sane interval — five minutes, not thirty seconds.
5. Check the resource graph once a month so you see trends rather than crises.
6. Know your plan’s worker count and memory limit, so you recognise the ceiling when you hit it.