bug (imho): two different hosts in one expression

[corehosting]

Hello @all,

I monitor two hosts with icmppingsec.

I created a trigger with both hosts. This trigger changes status without a reason (imho). So I created another two test-triggers.

test-trigger 1:

{host1:icmppingsec.count(#2,50,"gt")}<2
trigger ON > OFF > ON (21 Jan 06:21:00)
>> this is OK

test-trigger 2:

{host2:icmppingsec.count(#2,15,"gt")}<2
trigger permanent ON
>> this is OK

trigger:

({host1:icmppingsec.count(#2,50,"gt")}<2)&({host2: icmppingsec.count(#2,15,"gt")}<2)
trigger ON > OFF > ON (21 Jan 06:21:00)
>> this is imho NOT OK

Is this a bug or made I an error?


Values host1:

2009.Jan.21 06:21:16 16.8000
2009.Jan.21 06:21:07 33.3000
2009.Jan.21 06:20:58 33.6000
2009.Jan.21 06:20:46 464.0000
2009.Jan.21 06:20:36 62.0000
2009.Jan.21 06:20:30 34.7000
2009.Jan.21 06:20:16 16.4000

Values host2:

2009.Jan.21 06:21:20 5.2000
2009.Jan.21 06:21:07 5.1200
2009.Jan.21 06:20:58 4.9500
2009.Jan.21 06:20:50 6.8000
2009.Jan.21 06:20:40 19.4000
2009.Jan.21 06:20:30 7.0000
2009.Jan.21 06:20:20 9.3400
2009.Jan.21 06:20:10 6.3300

Kind Regards

Fabian

[nelsonab]

From what I can see this is working as expected. If you have two high latency pings in a row your trigger will turn off. Is there a reason why you have setup triggers that are almost always on? Also because your sample size is very small you will likely see a fair bit of trigger bounce.

As for the final trigger this is working as expected too here's a logic table for and operations.

1     2        Result
---------------------
T     T          T
T     F          F
F     T          F
F     F          F

[corehosting]

@nelsonab,

thank you very much for your fast reply! Now I see and understand the logic.

My final goal is this:

I would like to monitor 3 external and foreign hosts. Next I would like to create a trigger which is (almost always) true, if 2 of 3 hosts are reachable in normal ranges (icmppingsec).

This trigger is my indicator if the gateway of ZABBIX-Server ist full online, so I would use this trigger for all alarms as 2nd trigger. So I will get less "wrong" alarms, because of a tmp. bad gateway.

How should a trigger looks like, which solves my problem? Is there perhapes another way?


Kind Regards

Fabian