How to alert on disk going offline

[TigerNo3525]

Hi All,

We have an issue with some disks going offline and Zabbix isn't picking this up. I believe this is due to the disk status in Zabbix going to Unknown status which we aren't alerting on. I've looked at enabling alerts on unknown events but its alerting on lots of other items we don't want. Does anyone have a suggestion on how to proceed here?

Cheers for any assistance!!

[tim.mooney]

What OS, and how do you detect when a disk has gone offline? If there's a log message or something else Zabbix can watch for, then it should be possible to detect the event, but you'll have to identify how to programmatically detect when this happens.

[TigerNo3525]

Hi Tim,

Thanks for the reply, Windows Server 2016 mostly. Well currently we don't detect it, only notice due to the impact of the disk going offline (application issues). The main issue I'm trying to solve is so that we get alerted. If you see in the picture below the H: on this server has gone offline but we aren't alerted due to the Status being unknown.



Cheers

[TigerNo3525]

Actually I think if I could figure out where the below ones highlighted in yellow are coming from and how to exclude them I should be sorted and can then alert on items with Unknown status.



Any tips would be much appreciated!

[tim.mooney]

When the disk (or volume) goes offline, is there anything that gets logged to the event log?

One thing we've done for our windows hosts is to create a "canary file" at the root of any SAN volumes where they might suffer from a connectivity loss. We settled on naming the file "ZABBIX-DO-NOT-DELETE.TXT". We put some comments inside the text file about what its purpose is, and then used

vfs.file.exists[K:\ZABBIX-DO-NOT-DELETE.TXT]

for each SAN volume that we want to monitor for connectivity.

From there, you only need a trigger when that vfs.file.exists evaluates to false, to be alerted that you've lost a volume.

We put the item and the trigger in a template (with separate templates for each Windows drive letter, so we can have multiple templates assigned easily, for each mountpoint a particular Windows box might have).

[cyber]

Well, the simple answer to this question is, you do not monitor disk presence, you are just monitoring disk space..

And tim.mooney has here nicely presented a way to verify that... Or you should enable notifications on triggers turning to "unknown" or items to "not supported"...