High CPU on a Linux VPS: find which process and account causes it

The problem

The load average climbs, every website on the server gets slow, and top over SSH shows a wall of php-fpm and httpd processes that all look the same. On a shared server with dozens of hosting accounts, the real question is never “is the CPU busy?” — it is which account, site or script is making it busy, and whether that is normal work or a runaway process.

Doing this by hand means juggling top, ps, per-pool PHP-FPM status pages and Apache logs — while the server is on fire.

The solution

VpsMonitor.Pro polls CPU, memory, IO-wait and network every few seconds over SSH and draws live charts with history, so you see the spike the moment it starts — from the Windows tray, no terminal needed.

One click on Details next to the CPU chart takes a full “who is loading the server?” snapshot: top processes, top users, PHP-FPM pools, running MySQL queries, busiest domains and long-running maintenance jobs (like certbot or backups) — each normalised so 100% means the whole server, not one core.

The Accounts view goes further: per-account CPU and memory, charted hourly, daily and monthly from the panel’s own statistics, so you can prove which customer’s site caused last night’s spike. If you want a second opinion, the built-in Analyze with AI button sends the snapshot to Claude or ChatGPT for a plain-language verdict and prioritised next steps.