Some comprehension problems with triggers

[krusty]

Hi,

i have installed zabbix 1.1 for a log time ago. Till this day if only checked the bandwith and availability of more than 1500 hosts. Now i want to check some more detailed.

I want to check two things.
First: If the value of one item is over 100 the last 10 checks the trigger should fire a message to the administrator.
I create the following trigger:

{<HOST>:<KEY>.avg(3000)>100

I check the item in a period of 300sec and i want to check the last 10 values (300*10=3000) Could this trigger work? Or is there any other way to check the last 10 values?

The second problem i have is:
How can i check all value of one day. The trigger must fireing if 50 values of this day are over 100. Should i use the function count in association of max??? Or something different?

I have checked the forum for answers but i haven't had found any help.

If anyone have some ideas i would appreciative. Maybe you have examples for me how to do this.

Greetz Krusty

[hegermi]

You may be in for some suprises there.

I have noticed that for instance with the "Simple check" "icmpping" this type of calculation doesn't work.

I goes fine regular when the host responds but becomes irregular when the host doesn't. Maybe due to a timeout setting somewhere.

I have uncommented and changed some values in zabbix_server.conf, maybe this will help.

# After how many seconds of unreachability treat a host as unavailable
UnreachablePeriod=90

# How ofter check host for availability during the unreachability period
UnavailableDelay=10

# How ofter check host for availability during the unavailability period
#UnavailableDelay=60

But maybe the type of checks you do remain more regular....

Cheers,

Michel

[krusty]

Hi,
i will not check icmpping or other simple checks.

The item is type of "snmpv2 agent". Does it work with this type?

With the values of this item (snmpv2 agent) i want to analysis this values.

Regards

[hegermi]

not sure until I get a host with problems

Until I get a host with problems I wont know. So far all hosts respond and it is regular so you can asume a x second period should contain y polls.

If this behaviour remains when there is a timeout I don't know.

Simply being able to get the last x values would be a nice enhancement I think.

Cheers,

Michel

[krusty]

Partial solution

Hi,

now i have create one trigger which should report if the last 10 values over 100.

{<SERVER>:<KEY>.max(3000)}>100

But have still problems with the other trigger.
I have try to use something like the follows but that wouldn't work.

{<SERVER>:<KEY>.max(86400)>100.count(288)}>50

Have you any idea why it wouldn't work?

288 = Count how many times the value will be return on one day.
50 = maximum count (How many times the value will be over 100)

Thanks

[Alexei]

I create the following trigger:

{<HOST>:<KEY>.avg(3000)>100

I check the item in a period of 300sec and i want to check the last 10 values (300*10=3000) Could this trigger work? Or is there any other way to check the last 10 values?

Yes! Use {<HOST>:<KEY>.avg(#10)}>100

[James Wells]

I create the following trigger:

{<HOST>:<KEY>.avg(3000)>100

I check the item in a period of 300sec and i want to check the last 10 values (300*10=3000) Could this trigger work? Or is there any other way to check the last 10 values?

A better way to do this one would be;

{<HOST>:<KEY>.avg(#10)}>100.0

This says to compare 100 to the average of the last 10 values. Please note that unless you are using my bulk patch or Vince's AVG on Integers patch, then the key must be of type float.

The second problem i have is:
How can i check all value of one day. The trigger must fireing if 50 values of this day are over 100. Should i use the function count in association of max??? Or something different?

Sadly, there is no easy way to do this. You could fire if the max value for the last day is over 100, or you could fire if the average value for the last day is over 100, or even the minimum value is over 100, but there is no mechanism to count the number of times it breaks 100 yet.

[krusty]

Trigger not working

Yes! Use {<HOST>:<KEY>.avg(#10)}>100

Hi,

i use this trigger but the trigger would not change the status.

I currently got values over 60.
My trigger locks like

{<HOST>:<KEY>.avg(#10)}>60

Well this trigger should change the status to TRUE but the trigger status is still UNKNOWN. Any idea why the trigger don't change the status?

[krusty]

Hi,

i have tested the example from Alexei, but the trigger doesn't work. Maybe there is one bug in the version 1.1.

Could someone please check if the trigger works with the version 1.1 of zabbix?

Thanks

[James Wells]

Can you post the item configuration and the full trigger as you have it in the UI please? I suspect I know what the issue is, but want to see all of the data first, just in case.

[Alexei]

Hi,

i have tested the example from Alexei, but the trigger doesn't work. Maybe there is one bug in the version 1.1.

Could someone please check if the trigger works with the version 1.1 of zabbix?

Thanks

It didn't work if item type is integer in some of 1.1.x releases, but I believe it is fixed in 1.1.6.

[James Wells]

It didn't work if item type is integer in some of 1.1.x releases, but I believe it is fixed in 1.1.6.

Small correction. All versions of 1.1.x require float for avg, unless you apply my bulk patch or Vince's avg patch.

[Alexei]

You're absolutely right!