Be alerted in case of problem with auto discovery

[jcurty]

Hello

To monitor our infra, that have hundreds of servers and services, we use auto discovery a lot, to create automatically items and triggers.
Though, sometimes, it happens that there is a 'bug' / error in the script called by autodiscovery.
in that case, ... it leads to problems because, resources are not updated / created the way it should, and .. we have no clue that this is happening. Sometimes, we find 1 month later that the script didn't work correctly for that period, and that our monitoring was not correct ..

Is there a way to define a trigger, be alerted by zabbix if it gets an error calling the auto discovery script ? (like a 'no value' for a trigger for an intem for ex.)
what could be the best way to avoid having problem like this running for long before discovering them ,

many thanks for your help
regards
Jade

[ISiroshtan]

Hey there.

From top of my head can think only one way to handle it: instead of running the scripts in discovery itself one can create a master item that would collect and store JSON data. Then discovery can be switched to dependent item type. This way you could add extra trigger with nodata() onto new master item to be alerted if data for discovery failed to be collected and discovery itself will still work mostly unchanged.