High CPU on a VPS — common causes and fixes

High CPU on a VPS means something is consuming processor time continuously—a runaway process, traffic spike, crypto miner, or inefficient query loop. Find the process first; rebooting without diagnosis often brings the problem back.

Step 1: Identify the consumer

top -o %CPU
# or
htop
ps aux --sort=-%cpu | head -15

Note PID, USER, and COMMAND.

Step 2: Common causes

Process type	Typical fix
App workers (node, php-fpm, java)	Scale workers, fix infinite loop, optimize query
mysql / postgres	Slow query log, missing index, run EXPLAIN
find, tar, backup job	Schedule off-peak, ionice/nice
kdevtmpfsi, unknown miner	Security incident—kill, audit, rotate keys
Syn flood / bot traffic	Rate limit at CDN, fail2ban, review access logs

Step 3: Deeper look at one PID

sudo strace -p PID -c   # brief syscall summary—careful on prod
ls -l /proc/PID/exe
cat /proc/PID/cmdline

For web traffic spikes: tail -f /var/log/nginx/access.log and check for abusive IPs.

Step 4: Mitigate now

• Restart only the misbehaving service after you know what it is
• Temporary: renice or cgroup CPU limit on non-critical jobs
• Long-term: caching, DB indexes, horizontal scale, CDN

Avoid

• Rebooting the whole VPS as step one
• Killing systemd or database processes without understanding load
• Unlimited worker counts in app config on a small VPS

AI-assisted diagnosis

Describe “CPU at 100%” to an AI DevOps Copilot and approve read-only commands first (top, ps, log tails). Ohuriya AI is built for VPS owners who need that workflow safely. Connect · 502 errors