Eric - ZFS monitoring

New About Yours API Help
1.8 KB, Plain text

I’ve been a listener for quite a while, and even had a tool mentioned on an episode a while ago (episode 183 / ethname tool) which was fun to hear. At first, I thought my google-fu has missed a tool someone else had already created, but then realized you were describing mine!

Unrelated to that, In your last episode, you put out a request for how people monitor their ZFS installations. I’ll throw my hat into that ring: I use Zabbix extensively for monitoring ZFS and pool status. I like Zabbix for its extensibility, cross-platform nature, and reporting capabilities. (For the record, I have no relationship with Zabbix beyond as a user.)

Explicitly, I use Zabbix‘s discovery functionality to enable the automatic detection and monitoring of the pools and ZFS filesystems on both FreeBSD and ZoL, as well as ARC stats.  Once properly configured, it can automatically detect the presence of and configure logging for pool utilization / status as well as filesystem snapshot and reference usage, although any “zfs/zpool get *” property is trivial to monitor at this point.

Triggers can be set up to email — or notify in some other fashion — when a pool status is not online, or snapshot usage is over some percent relative to “referenced” for a filesystem, for example. (A sign that I may want to reconfigure snapshot retention for a FS.)  With its forecasting features, I can have zabbix warn me when a pool is going to reach some threshold of capacity at some time in the future based on recent trends.

Just thought I’d pass that along. If there’s interest, I could look into (with employer) sharing the config scripts, but they are essentially a zfs spin on the Zabbix example discovery scripts. Zabbix is a bit of a beast to stand up, but once you’ve wrapped your head around how it wants to do things, it works great.
Pasted 10 months ago — Expires in 64 days