Many Triggers ON for just a few seconds

[stomph]

Hello,

We see many triggers (processor load is too high) firing just for a couple of seconds and then being ok again for a while. Please see the attached screenshot from the Events-Screen.

I assume this is normal as some servers are quite active, but I just want to verify with others and am very greatful for experiences of others.

Do you also have this behaviour and do you do anything about it?

We are running Zabbix 1.8.15 on Centos 6.

Kind regards

-Stephan

[benbugohit]

i think it's not a real zabbix question, but OS related :
check beside your zabbix screen with "top" command, or reduce the interval between each "cpu usage" check ?

[heaje]

It's very easy for this kind of thing to happen. If the load spikes for one check, then the trigger goes off. What is the "Process load is too high" check based off of? CPU utilization? CPU load average? If it's based on CPU utilization, change the trigger to use an average value for the last 5-10 minutes or so. If it's based on load average, use the 5 minute or 15 minute load average in the trigger. Those are MUCH less susceptible to spikes.

[Bernd Hohmann]

We see many triggers (processor load is too high) firing just for a couple of seconds and then being ok again for a while.

The Zabbix preset templates are fine for unemployed servers but a pain in the ass in a productive environment where "to the universe and far beyond"-peaks are very common.

As heaje already mentioned, you have to replace most of the triggers with functions where the "avg" function plays a major rule.

For example: my gateway switch triggers an "Information" when the average in- or outbound traffic over the last 30 minutes is over 2Mbps.

Bernd