Autoscale and delaying an initial check

[structurefall]

Hello.

I have an AWS environment with autoscale instances. We're doing a healthcheck on them, which works, except that the amount of time it takes an autoscale instance to actually come into service properly after the Zabbix agent starts sometimes exceeds the amount of time between triggers. I'd like to tell Zabbix to only trigger if it has already seen at least one successful check from the healthcheck item.

My trigger formula now is:

{templatename:keyname.max(#3)}<1

I have a couple of other possible workaround ideas as well, but most of them involve altering our autoscale setup, which is less desirable. That said, any other suggestions about how to approach this are welcome!

I'm running Zabbix 2.0.8 on CentOS 6.3.

[zychonatic]

hi,

i had the same problem some months ago for an availability check (which returns 0 for failure and 1 for ok):

{templatename:keyname.nodata(300)}=1&{templatename :keyname.last(0)}=1


-) nodata(300) alerts if the service couldn´t be reached for more than 5 mins.
-) last(0)=1 ensures that there was 1 "healthy" check before.

i hope this helps you

br matthias

[structurefall]

Ah, good idea, thanks!

I actually wound up coming up with a different, slightly more awkward solution yesterday after I posted- the healthcheck in question triggers a script on the monitored host, so I built in some functionality to drop a dummy file in /tmp once a successful check has been returned, and only return a fail result if the dummy file is already in place.