Articles in this section

See more

High CPU and memory usage by Apache or PHP processes in Plesk

Avatar
Alexander Tsmokalyuk
Updated
Follow
Plesk Onyx for Linux kb: technical

Applicable to:

  • Plesk Onyx for Linux

Symptoms

  • High CPU and memory usage by Apache process is reported by top command:

    # top
    2043 example 20 0 0 0 0 S 100.0 70.5 0:05.43 httpd

  • High CPU usage by PHP-FPM process is reported by top command: 

    # top
    PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
    ...
    2089 example 20 0 0 0 0 S 96.8 60.5 0:02.43 php-fpm

Cause

Apache module is used to serve PHP. It is the slowest engine available.

Resolution

Log into Plesk and apply a solution according to handler type in Domains > example.com > PHP Settings

if mod_php or FastCGI is used

Switch websites' PHP engine to FastCGI application served by Apache or to FPM application served by Apache (better option) via the tab Domains > example.com > Hosting Settings:
slowCPU.png

if "FPM Application server by Apache" is used

Warning: Nginx does not support .htaccess rules. Code of some websites may not support FastCGI or PHP-FPM, so it is recommended to consult with website developer before switching the handler.

1. Enable nginx web server 

2. Switch to PHP 7.0 or newer at Domains > example.com > PHP Settings

3. Enable to Nginx for affected domain: open Subscriptions > example.com > Apache & nginx Settings, check option Proxy mode and press OK:slowCPU2.png

Note: on servers with high load, switch to Event mode at Tools & Settings > Apache Web server > MPM Mode

Comments

4 comments

Sort by Date Votes
  • Avatar
    Michael Pratt

    Be cool if we could narrow down the cluprit/domain in the backend...

    0
    Comment actions Permalink
  • Avatar
    Bato Tsydenov

    @Michael Pratt

    Please check the following article:

    How to check which domain PHP process consumes CPU resources?

    0
    Comment actions Permalink
  • Avatar
    Michael Pratt

    Hi Bato, 

    I am familiar with the -c command. 

    This is not helpful if, for example, I am trying to determine a server crash the night before. Also top jumps around, so it really isnt ideal for troubleshooting in real time either, although it might help narrow it down, you still have to be in a position to SSH into the server as it is happening. 

    Would be so much easier, and productive, to have these stats and history in the backend, no?

    0
    Comment actions Permalink
  • Avatar
    Alexandr Bashurov

    @Michael Pratt

    Indeed, in case the error happened a while ago running top -c might not help.

    However, there is also atop, which can run as a daemon, logging all activity in the background and allowing for auditing and extracting reports from captured data. You may install, configure and use it as described here: How to monitor usage of system resources in a period of time using atop?

    0
    Comment actions Permalink

Please sign in to leave a comment.

Have more questions? Submit a request

Related articles