We've been using Zabbix for a couple of years now, and we like it a lot.
One thing that we've never quite understood how to handle are "unsupported items". There are three scenarios:
1) Items that are just unsupported
We have templates that sometimes have items that are not applicable to the hosts they are applied to, so some of the items are unsupported
2) Items become unsupported because of transient error
I.e. An external script that connects to a networking port to monitor becomes unsupported if it can't connect (and the script doesn't handle it)
3) Items that seem to be designed with unsupported as fallback
vfs.file.time becomes unsupported if the file doesn't exist
----
For a long time we didn't check or alert on unsupported items at all. It caused problems more than once because items weren't being monitored at all because of some error and we had no idea. Then we tried turning on alerting when things became unsupported and it caused such a noise in Zabbix that we had to turn it off again. Then we went through every unsupported item and either disabled what we don't use or fixed what we use so that it wasn't unsupported any more.
This way we could turn on alerting for unsupported items, but that leaves scenario 3. We have checks that monitor the age of lock files to alert us if a lock file is older than some threshold. The only problem is that vfs.file.time becomes unsupported if the file doesn't exist, and the files in question only exists when there is processing job in progress. We can always write a very simple shell script to check the age and just return 0 if the file doesn't exist, that would fit our use case, but I'd like your experience in handling unsupported items in general.
One thing that we've never quite understood how to handle are "unsupported items". There are three scenarios:
1) Items that are just unsupported
We have templates that sometimes have items that are not applicable to the hosts they are applied to, so some of the items are unsupported
2) Items become unsupported because of transient error
I.e. An external script that connects to a networking port to monitor becomes unsupported if it can't connect (and the script doesn't handle it)
3) Items that seem to be designed with unsupported as fallback
vfs.file.time becomes unsupported if the file doesn't exist
----
For a long time we didn't check or alert on unsupported items at all. It caused problems more than once because items weren't being monitored at all because of some error and we had no idea. Then we tried turning on alerting when things became unsupported and it caused such a noise in Zabbix that we had to turn it off again. Then we went through every unsupported item and either disabled what we don't use or fixed what we use so that it wasn't unsupported any more.
This way we could turn on alerting for unsupported items, but that leaves scenario 3. We have checks that monitor the age of lock files to alert us if a lock file is older than some threshold. The only problem is that vfs.file.time becomes unsupported if the file doesn't exist, and the files in question only exists when there is processing job in progress. We can always write a very simple shell script to check the age and just return 0 if the file doesn't exist, that would fit our use case, but I'd like your experience in handling unsupported items in general.
Comment