How should look unavailable trigger.

[pietro54]

Hi,
In zabbix 2.0 I was using (internal) item

zabbix[host,agent,available]

and trigger

 {Linux:zabbix[host,agent,available].min(9)}=1

.

But there is a problem, this doesent work well in new zabbix 2.2

Can some one tell how should look correct item and triger to check zabbix agentd is available?

[pc99096]

just use the default zabbix agent template

item - agent.ping

trigger (if there is no data for more than 5 minutes) - agent.ping.nodata(5m)=1

[pietro54]

Hi,
thanks, this is very helpful.

But this is generating some false alarm on windows hosts.
Is there any possibility to create something like this

Zabbix server ask host are you alive: yes im alive
Zabbix server ask host are you alive: no answer
Zabbix server ask ... no answer
Zabbix server ask ... no answer

and if there 3 check one by one give no answer then trigger should to problem.

[pc99096]

i am usually using icmpping instead of agent.ping

https://www.zabbix.com/documentation/2.0/manual/config/items/itemtypes/simple_checks

and a trigger like icmpping.sum(#3)<1 (sum of the last 3 values)

[aib]

Hi,
thanks, this is very helpful.

But this is generating some false alarm on windows hosts.
Is there any possibility to create something like this

Zabbix server ask host are you alive: yes im alive
Zabbix server ask host are you alive: no answer
Zabbix server ask ... no answer
Zabbix server ask ... no answer

and if there 3 check one by one give no answer then trigger should to problem.

{hostname:agent.ping.nodata(#3)}=1

[pietro54]

Are you sure it work like you want it?
Because it work like this:
if in any of 3 last values was no data then trigger goes to problem.
No meter if this was first or third value. (that's just mine opinion)

[steveboyson]

nodata(#3) is the 3rd last value.

Better use count(#3,1,"lt")
It will count the number of values of the last 3 item calls which have less than "value=1". Is probably that what you want.

[pietro54]

Yes, this sounds correctly, but i can not add trigger like this:
{Windows :agent.ping.nodata(#3,1,"lt")}=1

Windows is a template name.

[steveboyson]

But you can use {Windows.agent.ping.count(#3,1,"lt")}>0

[pietro54]

Hi,
I try this, but is doesn't work well...
Please let me understood this

agent.ping.count(#3,1,"lt")}>0
agent.ping.count(<numer of recived values>, <what means 1>, < what means lt>?

I can`t find it on https://www.zabbix.com/documentation...ers/expression

where do you get such knowledge?

[steveboyson]

https://www.zabbix.com/documentation/2.0/manual/appendix/triggers/functions

agent.ping.count(#3,1,"lt")}>0

Trigger fires, if the number of values from the last 3 results which are "less than" (lt) 1 is more than 0.
(means: if more than "zero" value from the last three values have failed)

[pietro54]

But if there will be "no data".
no data#(value)0.

I mean, if host is down, server receives <no data>, so server cannot compare 1or0 to <no data>.
Thats how I`m thinking.

[steveboyson]

Simply add a 2nd "nodata" trigger.

[pietro54]

Hi,
Thanks for your time and patient for this thread.

I think it`s not thats simple.
If i will create something like this:

agent.ping.count(#3,1,"lt")}>0 | agent.ping.nodata(0)}=1

This trigger will generate same number of false alarms like only agent.ping.nodata(10)}=1

But if I will create

agent.ping.count(#3,1,"lt")}>0 & agent.ping.nodata(0)}=1

I think trigger will not work correctly.

trigger should implicit in one count and no data.
Sth like that:

agent.ping.count,nodata(#3,1,"lt")}>0

[steveboyson]

Simply add a 2nd "nodata" trigger.

"A 2nd trigger" means a 2nd trigger. Not one trigger with two statements, combined by "|" or "&" ...