Issue Server health parameter "CPU > Total usage" changed its status from "green" to "red".

[Chrisa]

I am seeing this across several servers. It does appear to be random.

Watchdog is reports (top) psaadm consuming all CPU resources
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
29898 psaadm 20 0 291264 41164 34828 R 25.0 0.7 0:00.04 sw-engine
29899 psaadm 20 0 291236 40864 34528 R 25.0 0.7 0:00.04 sw-engine
29900 psaadm 20 0 291236 41080 34748 R 18.8 0.7 0:00.03 sw-engine
29901 psaadm 20 0 291396 42316 35812 R 18.8 0.7 0:00.04 sw-engine
29897 psaadm 20 0 291396 42652 36140 R 12.5 0.7 0:00.03 sw-engine
1 root 20 0 38080 6064 3900 S 0.0 0.1 0:03.50 systemd

I have top already running so I can switch to it within a second of receiving these warning. top never shows what watchdog reports so I suspect that watchdog 'waits' before reporting these issues.

This watchdog warning seems to be reported with 2 variations:
Server health parameter "CPU > Total usage" changed its status from "green" to "red".
Server health parameter "Services > Apache CPU usage" changed its status from "green" to "red".

Plesk Onyx v17.0.17
Ubuntu 16.04.2 LTS

 

[Bitpalast]

Server health monitor checks short-term, mid-term or long-term intervals according to what has been defined in the .xml configuration file that you can download from (and upload to) the server health monitor configuration. The alarm goes off only if a defined threshold is exceeded throughout the defined interval. For that reason the realtime data does not necessarily reflect server health monitor behavior.

It is normal to see some peaks in CPU and Apache CPU usage throughout a day. There might even sometimes be unreasonably high peaks for a short period of time. This is nothing to worry about. Only if a warning state is maintained for some while one should check what is going on.

 

[Chrisa]

Hello Peter,
I agree that peaks might be normal throughout the day if we are dealing with valid traffic/activities. When I look at the frequency of these logs across all my servers it doesn't 'feel' normal. It certainly doesn't leave me with the warm fuzzies.

My follow-up questions would be:
What exactly was 'psaadm' doing when it was taking 100.1% of the CPU resources?
How do I determine this task was a valid activity and not a problem this was caused by a misconfiguration or some sort of external interference?

Thanks. Further suggestions would be appreciated.

 

[Bitpalast]

Difficult to tell what the services are doing. They could be executing maintenance tasks, they could also be reacting on events coming from the outside, e.g. login attempts. They will react to any action in the Plesk GUI.

 

[Bitpalast]

Two other similar cases in the forum today.
https://talk.plesk.com/threads/sw-engine-status-red.341821/
and
https://talk.plesk.com/threads/health-manager-high-apache-cpu-alarms-sw-engine.341634/
Maybe you want to "watch" these threads so that you get notified on updates in those.