Issue CPU Load sw-engine

[haaid]

Dear all,


I have noticed that when I "top" I get this result:

op - 23:22:49 up 43 min,  1 user,  load average: 52.67, 47.56, 80.66
Tasks: 527 total, 144 running, 380 sleeping,   0 stopped,   3 zombie
Cpu(s): 81.0%us, 18.9%sy,  0.0%ni,  0.0%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:  32815940k total,  4605928k used, 28210012k free,   276812k buffers
Swap: 16777084k total,        0k used, 16777084k free,  1035984k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND           
 2548 mysql     20   0 4239m  95m 7096 S 16.9  0.3   6:14.48 mysqld             
 6218 vlov_adm  20   0  403m  74m  44m R  6.3  0.2   0:01.25 php-cgi           
 9298 psaadm    20   0     0    0    0 R  4.3  0.0   0:00.13 sw-engine         
 9300 psaadm    20   0     0    0    0 R  4.3  0.0   0:00.13 sw-engine         
 9301 psaadm    20   0  243m  29m  18m R  4.3  0.1   0:00.13 sw-engine         
 9320 psaadm    20   0  249m  30m  19m R  4.3  0.1   0:00.13 sw-engine         
 9311 psaadm    20   0  243m  29m  18m R  4.0  0.1   0:00.12 sw-engine         
 9312 psaadm    20   0  249m  30m  19m R  4.0  0.1   0:00.12 sw-engine         
 9318 psaadm    20   0  243m  29m  18m R  4.0  0.1   0:00.12 sw-engine         
 9323 psaadm    20   0  254m  35m  19m R  4.0  0.1   0:00.12 sw-engine         
 9335 psaadm    20   0  257m  38m  19m R  4.0  0.1   0:00.12 sw-engine         
 9337 psaadm    20   0  257m  38m  19m R  4.0  0.1   0:00.12 sw-engine         
 9342 psaadm    20   0  251m  32m  19m R  4.0  0.1   0:00.12 sw-engine         
 9345 psaadm    20   0  256m  37m  19m R  3.7  0.1   0:00.11 sw-engine         
 9349 psaadm    20   0  256m  37m  19m R  3.7  0.1   0:00.11 sw-engine         
 9352 psaadm    20   0  257m  38m  19m R  3.7  0.1   0:00.11 sw-engine         
 9353 psaadm    20   0  256m  37m  19m R  3.7  0.1   0:00.11 sw-engine         
 9355 psaadm    20   0  256m  37m  19m R  3.7  0.1   0:00.11 sw-engine         
 9356 psaadm    20   0  257m  37m  19m R  3.7  0.1   0:00.11 sw-engine         
 9362 psaadm    20   0  256m  37m  19m R  3.7  0.1   0:00.11 sw-engine         
 9364 psaadm    20   0  256m  36m  19m R  3.7  0.1   0:00.11 sw-engine         
 9374 psaadm    20   0  256m  37m  19m R  3.7  0.1   0:00.11 sw-engine         
 9376 psaadm    20   0  256m  37m  19m R  3.7  0.1   0:00.11 sw-engine         
 9358 psaadm    20   0  255m  36m  19m R  3.3  0.1   0:00.10 sw-engine         
 9359 psaadm    20   0  256m  36m  19m R  3.3  0.1   0:00.10 sw-engine         
 9384 psaadm    20   0  255m  36m  19m R  3.3  0.1   0:00.10 sw-engine         
 9391 psaadm    20   0  255m  35m  19m R  3.3  0.1   0:00.10 sw-engine         
 9398 psaadm    20   0  255m  36m  19m R  3.3  0.1   0:00.10 sw-engine         
 9399 psaadm    20   0  255m  36m  19m R  3.3  0.1   0:00.10 sw-engine         
 5087 psaadm    20   0  480m  37m  19m S  3.0  0.1   0:00.47 sw-engine-fpm     
 5120 psaadm    20   0  477m  31m  16m S  3.0  0.1   0:00.38 sw-engine-fpm     
 5123 psaadm    20   0  477m  31m  16m S  3.0  0.1   0:00.39 sw-engine-fpm     
 9382 psaadm    20   0  255m  36m  19m R  3.0  0.1   0:00.09 sw-engine         
 9390 psaadm    20   0  254m  35m  19m R  3.0  0.1   0:00.09 sw-engine

that is, to say it nicely, a whole bunch of these sw-engine threads. My CPU warning is constantly very high as well and my server and the domains running on it is rather slow. When I manually kill sw-engine, all goes well for a while but the process seems to mysteriously re-enable itself.

A few questions:

- is it normal that it has so many threads going on the same time?
- can i do something to not have it impact my cpu usage (and thus serverspeed) so much?
- do you have any other hints, tips or tricks?

Thanks a lot,


Klaas.

ps. i tried searching for "sw-engine" but the "sw" gets dropped because of too short,...

 

[IgorG]

Do you have any related suspicious records in /var/log/sw-cp-server/sw-engine.log ?
Have you tried restart sw-engine service?

 

[SibProgrammer]

Reported as PPP-26712
Fixed in the scope of microupdate #6

 

[Giorgos Kontopoulos]

I have a very similar situation

I am getting frequent alert emails from my plesk ONYX web admin edition
with lots of sw-engine lines in them consuming high CPU

OS: ‪CentOS 6.8 (Final)‬
Product: Plesk Onyx 17.0.17 Update #12 , last updated at Dec 26, 2016 03:34 AM
Checked at Jan 10, 2017 03:11 AM.

I am attaching one such alert

I know the plesk panel is not used MOST of the times when those alarms arrive
and the server is not really used much yet

I am running another 2 servers with plesk 12.5 and 12 respectively and did not notice such thing

I have done a restart of the sw-engine

/etc/init.d/psa restart

and will report back but if you have any suggestions they are very welcome

 

[Giorgos Kontopoulos]

Still getting alert emails with sw-engine occupying too much CPU
meanwhile I have not logged in the PLESK PANEL for some time

any suggestions ?

 

[Oto Tortorella]

Same here, any advice?

 

[Bitpalast]

According to the log file from #4, the load comes from high Apache CPU load, not from sw-engine. Can you identify the processes that are causing a high load by
# ps aux
or can you see them live by
watch "ps aux | sort -nrk 3,3 | head -n 20"
?

 

[Oto Tortorella]

I'm not able to see it live, I get this warning several times a day and then it returns to normal in 10/20 seconds.

What can cause the sw-engine burst?

 

[Bitpalast]

Please run
# ps aux | grep sw-engine
and report output here.

 

[Bitpalast]

This file includes some text, but it is not the output of # ps aux | grep sw-engine.

 

[Oto Tortorella]

Sorry, here is the correct file:

 

[Bitpalast]

According to that output, sw-engine is not using any CPU power. What you will need to do is to run # ps aux | grep sw-engine at the same time when the issue occurs in # top.

 

[Giorgos Kontopoulos]

@Peter Debik
Not coming up that often and its not possible to RUN this command when it occurs (unless I am lucky) because even if I get the warning and jump into the server the problem is gone
any other method to see what caused the high cpu on the last 30 minutes lets say ?
or anything that continuesly checks and then I can find the report waiting for me ?
thanks

 

[Bitpalast]

It is not possible to react on the server health notification? You could try to create your own script and a crontab job every minute to test the situation, evaluate the load caused by all jobs that refer to sw-engine and mail or text you if a threshold that you define in that script is passed. There is no "ready" solution for that beyond server health monitor.