Zabbix 3.0 system.cpu.util trigger windows

[Zabbixtrainee]

Hello,

I am monitoring my windows laptop for testing purposes. I am using the latest Zabbix 3.0.1 version and the agent is also the newest one. I am getting the data from Templates so all my configurations should be correct.

I want the trigger to alert if CPU utilization goes over 75% for 10min.

So my item key is:
"system.cpu.util"
it displays the CPU utilization% and it seems to be working. (I am getting correct data). The default parameters are [all,system,avg1]

My trigger expression for this is: "system.cpu.util.avg(600)}>75"
Which should alert if the average CPU utilization goes over 75% for 10minutes, right?

But when I use prime95 for stressing the CPU at 100% and the trigger alerts almost immediately. I guess my trigger expression is wrong?

BTW. I can't use perfounter because the port 10051 isn't open yet and I'd like to use this since it is supported in Zabbix 3.0 (not sure was it in previous versions)

I appreciate any given help!

[chewy]

I was testing zabbix a few months back (got sidetracked for a while now) but I did experience the same problem as you and I thought exactly the same thing with regards to the trigger but sadly zabbix doesn't agree with us.

Back in time I think I found the solution but never tried it. I found this:

http://blog.zabbix.com/no-more-flapping-define-triggers-the-smart-way/1488/

and if my understanding is correct, instead of using avg, try using min.

Let me know if this solves your problem as it's something I'll need to get working eventually as well.

[Zabbixtrainee]

Hi, chewy

After the relaxing weekend I quickly figured that the trigger:
"system.cpu.util.avg(600)>75" was actually working but the 75 was just too low for my testing purposes. In the other word I was just being stupid... (for my defence, it was a long friday afterall)

But anyway I implemented the hysteresis to my trigger and now it looks like this:
({TRIGGER.VALUE}=0 and {hostname:system.cpu.util.avg(15m)}>95)
or
({TRIGGER.VALUE}=1 and {hostname:system.cpu.util.avg(10m)}>75)

It fires when CPU util average goes over 95% for 15minutes and recovers when CPU util average is below 75% for 10minutes. And now it's working!

I can't believe how long I struggled with this "problem".

[sakibmas]

try system.cpu.load

Actually 'System Idle Process' is causing the high value. Please try system.cpu.load[,avg1]

$ zabbix_get -s 192.168.2.3 -k system.cpu.util[,,avg1]
100.000000

$ zabbix_get -s 192.168.2.3 -k system.cpu.load[,avg1]
2.166667