Host unreachable for a certain period

[Taz-Matt]

Hi!

I have made a search through these forums and found out other threads talking about the question that I am about to ask but none had a concrete or working answer to the problem so here is the question (if someone made a better search than me and found another thread which has the answer, please refer to it and I'll read it carefully):

The question: Is it possible to use the host's "status" item to create a trigger that tells if the server has been unreachable for at least (example) 5 minutes.

Problems that I previously encountered:

1- Other items than "status" stop being polled when zabbix_agentd is unreachable so the triggers are not triggered so they can't be used to tell that the host is unreachable, they can only be use to tell that a certain service/port/etc. is unavailable.

2- I can't use the status item with standard trigger arguments such as, the minimum value for the past 5 minutes was higher than 0 since the status item's value is not stored as long as there is no change so if the server has been up for let's say a week and now becomes unreachable, there will be only 2 values stored for the status item during the last week and therefore, the trigger will be triggered right away and not 5 minutes after the host has become unreachable.

-----

Guess what, I finally found my answer while writing this message by searching even more... I am still posting it if it can help some people like me to find the answer, here is the way I found and did it (should've thaught about it, LOL!):

{HostOrTemplateName:agent.ping.nodata(300)}=1

I think this is a basic thing but should be written in the manual as an example since most people will use a trigger like this and it brought me (and others as I see in other threads) to search for an answer to this...

Sorry for the others who already knew the answer. You can also post other ways you guys did it. Maybe there are better ways it can be done.

Thanks!

[bbrendon]

Good job. You can eliminate the agent.ping item all together and just use something like cpu load or any reliable item. If you have 200 servers saves 200 items and many more in your history tables!

[exkg]

Hi,

{HostOrTemplateName:agent.ping.nodata(300)}=1

I think this is a basic thing but should be written in the manual as an example since most people will use a trigger like this and it brought me (and others as I see in other threads) to search for an answer to this...

I think its already in manual:

nodata sec any Returns:
1 – if no data received during
period of time in seconds. The
period should not be less than 30
seconds.
0 - otherwise

But u didnt need create two triggers, use an 'OR' to compose a trigger.


[]s,
Luciano

[Taz-Matt]

You can eliminate the agent.ping item

Excellent idea! We don't have hundreds of servers but even on the 12 servers we have, it is 1440 requests a day (every 60 secs.) per server so, 17280 requests. Thanks bbrandon!

I think its already in manual

What I meant by putting this in the manual is that there is a place where they explain how triggers work and they have put the following example :

Example 7 Server is unreachable
{zabbix.zabbix.com:status.last(0)}=2

There, they should add something like:

Example 8 Server was unreachable for 5 minutes
{zabbix.zabbix.com:agent.ping.nodata(300)}=1

Just to point users in the right direction instead of having to understand every little commands and putting the pieces together since it is a trigger that most people will use... Just a suggestion. Thanks for the reply exkg!