One morning your site feels slower than usual. The server is up and there are no errors, yet every page hesitates for several seconds, and the CPU graph in your hosting panel is pinned to the top. This is one of the most common shared hosting situations: the site has used up its share of processor time. Here is why it happens, how to find the cause and how to stop it recurring.
How the CPU limit works
A shared server is divided among many sites, so each account receives a CPU share, usually expressed as a percentage of a core, where 100 percent equals one full core. When your site needs more, the system does not shut it down; it throttles your processes so they run in turn at limited speed. That is why the symptom is slowness rather than an outage: pages load slowly, the admin panel drags and some requests end in 503 or 504 timeouts. If the process limit fills up too, new visitors may not get a response at all.
Common causes
Bots
A large share of traffic often comes from search engine robots, SEO tools, AI data crawlers and scanners. On an uncached site every hit costs PHP and database work, and in online shops filter combinations create thousands of unique URLs for bots to walk through.
No caching
Without a cache, every page is built from scratch on each visit. With page caching, a ready page is served and the CPU barely works, which often solves half the problem.
WP-Cron and background jobs
WordPress checks scheduled tasks on every visit by default, and heavy jobs such as backup or sync plugins can occupy the CPU for long periods.
Brute-force attempts
Password guessing against wp-login.php and xmlrpc.php is invisible but expensive, because each attempt starts PHP and verifies a password hash.
Finding the culprit in the logs
Analyse the access log, usually found in the logs section of your hosting panel or the account's logs folder:
# IP addresses sending the most requests
awk '{print $1}' access.log | sort | uniq -c | sort -rn | head -20
# Most active User-Agents (spotting bots)
awk -F'"' '{print $6}' access.log | sort | uniq -c | sort -rn | head -20
# Most requested URLs
awk '{print $7}' access.log | sort | uniq -c | sort -rn | head -20
# Hits on xmlrpc.php and wp-login.php
grep -cE "xmlrpc.php|wp-login.php" access.log
One IP with thousands of requests suggests a bot or attack; the User-Agent list shows which crawlers are busiest; the URL list reveals heavy pages such as filters or search.
Practical fixes
Enable page caching first, then move WP-Cron to a real cron job so tasks run on a schedule rather than on every visit:
// in wp-config.php
define('DISABLE_WP_CRON', true);
# cron job in the hosting panel (every 15 minutes)
*/15 * * * * php /var/www/USER/data/www/example.uz/wp-cron.php >/dev/null 2>&1
Adjust the path to your actual folder, which you can check in the panel's file manager. If you do not need xmlrpc.php, for example because you do not publish via the mobile app or Jetpack, block it:
# .htaccess (Apache 2.4)
<Files xmlrpc.php>
Require all denied
</Files>
Close filter pages in robots.txt and block overly aggressive IPs. Remember that robots.txt only affects well-behaved bots; malicious scanners ignore it.
Example: a furniture shop in Namangan
A furniture shop's site in Namangan slowed every afternoon with CPU at the limit. The access log showed an SEO crawler walking every colour, size and price filter combination, tens of thousands of requests a day, plus password attempts against xmlrpc.php every minute. Blocking the crawler, closing filter URLs in robots.txt, disabling xmlrpc.php and enabling page caching brought CPU well below the limit without a plan upgrade.
When optimisation is not enough
If caching is on, bots are under control and slow queries are fixed but CPU still sits at the limit, your site has genuinely grown. A stronger plan or a VPS with dedicated resources is the next step, and you can change plans or order a VPS in the sayt.uz cabinet. Find the cause first, though, or a bigger plan will fill up just as quickly.