Triggers for web monitoring question

[lukeiam]

Hello,

I am using Zabbix 2.0.2 on CentOS 5U8. I have set up web monitoring with the following trigger:

{host:web.test.fail[Scenario].last(0)}#0

as per:
http://www.zabbix.com/documentation/...nitoring/items

It works great, however I am monitoring some sites that are far away and it is possible they might fail a check or two. With the config above, I get a notification for every failure. I would like to set up this trigger in a way, that after 3 failed checks it triggers the notification. Is that possible?

Regards,

Luke

[mbsit]

Yes, its possible.

Read DOC: http://www.zabbix.com/documentation/...gers/functions
especially "last" function.

Best,
Grzegorz

[lukeiam]

Triggers for web monitoring question

Thank you for the reply.

Good find. I will bookmark that page.

I read through the notes for the last parameter, but I am a little confused.

since it's talking about # and a number do I to this:

{host:web.test.fail[Scenario].last(#3)}#0

or this:

{host:web.test.fail[Scenario].last(0)}#3

Assuming I want it to trigger after 3 fails.

Regards,

Luke

[lukeiam]

Yes, its possible.

Read DOC: http://www.zabbix.com/documentation/...gers/functions
especially "last" function.

Best,
Grzegorz

Thank you again for the reply. I don't know why my posts take so long to show up on the thread. Anyway, I reviewed that page, and it contains under the "last" option the following:

"last(#3) - third most recent value (not three latest values)"

This means it will look at the third value not compare last three. Am I understanding this correctly? If so, then this will not do as intend it to.

Can you shed some more light on this?

Regards,

Luke

[mbsit]

Sure

You can trigger like this:
{host:web.test.fail[Scenario].last(#3)}#0 & {host:web.test.fail[Scenario].last(#2)}#0 &
{host:web.test.fail[Scenario].last(#1)}#0

& = logical AND

Maybe not "brilliant" but it's simple and easy to understand.

[lukeiam]

I'm sorry for the late response. My wife had a baby

Anyway, the simplicity of this makes be embarrassed. Sorry I didn't figure it out and thank you kindly for the help!

Regards,

Luke

Sure

You can trigger like this:
{host:web.test.fail[Scenario].last(#3)}#0 & {host:web.test.fail[Scenario].last(#2)}#0 &
{host:web.test.fail[Scenario].last(#1)}#0

& = logical AND

Maybe not "brilliant" but it's simple and easy to understand.

[angelhc]

Congrats for the baby!

I think this issue could be better if check the last 3 values, for example:

.-If you check the web each minute:
{host:web.test.fail[Scenario].count(180,0) - the number of values for last 3 minutes that equal '0'

Hope this help.

[mbsit]

Hi

Hmm, yes we can do that, too.
{host:web.test.fail[Scenario].count(180,0)}=3

Bests,
Grzegorz

[lukeiam]

Thank you.

This seems to work well:

{host:web.test.fail[Scenario].last(#3)}#0 & {host:web.test.fail[Scenario].last(#2)}#0 &
{host:web.test.fail[Scenario].last(#1)}#0

But I will try yours too as it is shorter. Since I check every 5 mins, I would do:

{host:web.test.fail[Scenario].count(900,0)}=3

Is that correct?

Regards,

Luke

[novregen]

hi,

why don´t you use this one ?

{:.avg(900)}#1

So you get an alarm, if the web site can´t reach 3 time by an intervall of
300 seconds.

[davidca]

Hello-
While this is an older thread, I can't seem to find a similar one related to the structure on this thread that does seem to be near my question

Would wildcard values in the expression make sense or does the trigger always need to be a specific item?

I have setup a trigger for {host:web.test.fail[Scenario].last(0)}<>200 where the [Scenario] in the live expression is an ITEM but I wonder if this should be a wildcard variable like {host:web.test.fail[ANY-ON-HOST].last(0)}<>200 and thus trigger on any problem for any of the 20 website monitors based on not receiving and 200 response code.

But then I begin to wonder if all triggers should be custom to one web scenario and I would want one or many triggers active for every monitor? My overall objective:

My goal is monitoring web scenarios with importance on
1) not getting anything except a 200 (all OK) code from the web page
2) and the other alert would be for {host:web.test.in[SCENARIO,,bps].last(#2)}>=500 with objective of a warning if time is over 500ms.