Question Apache High CPU, but total CPU low

[Caspar]

Hi all,

I am baffled by the following:

I get frequent e-mails that there is HIGH (red) CPU usage for "Apache CPU" (where I see the process is sw-engine). In the monitoring I see this for the Apache also, but when I look there is hardly any CPU usage in the total (everything below 10%)

The e-mail that is send should send only red messages if CPU is above 50%

This system has 2 vCPU cores so it isn't something that should trigger it if I look at it, even if it would be using 50% of CPU of one core (which should be 25%)

Also trying to find this, I only sometimes see a high CPU of the sw-engine-fpm module for Plesk pool, but even that is not that high (while I haven't gotten an e-mail from "red to green")

Looking with htop I don't see that high usage ( load avarage while I'm looking now is .03 .22 .22)

When running the following I also am not wiser:

ps -o %cpu,%mem,cmd -C sw-engine
%CPU %MEM CMD
 0.0  1.0 /usr/bin/sw-engine -c /opt/psa/admin/conf/php.ini /usr/lib/plesk-9.0/psa-health-monitor-notification.php
 0.0  0.8 /usr/bin/sw-engine -c /opt/psa/admin/conf/php.ini /opt/psa/admin/bin/modules/watchdog/wdcollect -c /opt/psa/etc/modules/watchdog/wdcollect.inc.php
 0.0  1.3 /usr/bin/sw-engine -c /opt/psa/admin/conf/php.ini -dauto_prepend_file=sdk.php /opt/psa/admin/plib/modules/plesk-mobile/scripts/push_worker.php

I looked for updates, but I don't see anything, and the monitoring e-mails keep coming quite often.

Can anyone help me either to pinpoint the issue, or to figure out if this a. is a bug and b. solve this "bug".

Using:
‪Ubuntu 16.04.2 LTS‬
Plesk Onyx Version 17.5.3 Update #13, last updated on July 14, 2017 06:26 AM

Thanks!

 

[Caspar]

Can anyone help me with this?

Currently my times on the load average are as followed: .06 .16 .17
But if I would look in the apache graph, it says that apache is using 70-80% cpu (see uploaded picture)

I would like to pinpoint where my problem is so I can make a permanent solution

 

[Caspar]

Hmm,

I managed to "solve" it by using these actions:

service sw-cp-server restart; but it failed since it was running but didn't think it was running
Killed all processes sw-cp-server
service sw-cp-server start
service sw-collectd stop
Killed all proccesses with sw-collectd
service sw-collectd start
service sw-engine restart

After this it looked like everything is working properly again.

Anyone who will have any clue on what might have been wrong?

 

[UFHH01]

Hi Caspar,

consider to adjust the standart settings for the Plesk Health monitor with your very own definitions/settings, to avoid receiving notifications, which might not point to any issues/errors/problems, but depend on too low standart settings:

=> How to change alarm levels for Health Monitor​