Introduction

Even though the hard disks are monitored via S.M.A.R.T. and the status of the ZFS pools is reported regularly with the periodic scripts, there is still no immediate notification in the event that one of the ZFS pools has a problem. So if any ZFS-specific problem occurs, we would like to know immediately. There is a small script for this and extension of the Monit configuration.

Last update:

• 24.11.2024: Initial document.

Prerequisites

• Monit is installed and set up

ZFS

I drew inspiration for this script from the following sources

• https://pawelrychlicki.pl/Article/Details/58/zfs-health-check-script-for-monit-0905
• https://calomel.org/zfs_health_check_script.html

The resulting script is a very lean modification with the following changes:

• Dynamised, one loop for all ZFS pools and actions
• Scrub and Trim have been removed as these are already handled by the periodic scripts
• E-mail sent Monit

Firstly, the following additional check is added at the bottom with ee /usr/local/etc/monitrc:

    check program zfs_health with path "/root/zfs_health_check.sh"
        if status != 0 then alert

The script stored there is created with ee /root/zfs_health_check.sh and the following content:

#! /bin/sh
/usr/bin/printf "%s\n\n" "$(/sbin/zfs list -o name,avail,used -d 0)"
/usr/bin/printf "%s\n\n" "$(/sbin/zpool list -o name,size,allocated,free,capacity,health)"
LISTPOOLS="$(/sbin/zpool list -H -o name)"
for POOL in ${LISTPOOLS}; do
  HEALTH="$(/sbin/zpool list -H -o health ${POOL})"
  ERROR="$(/sbin/zpool status ${POOL} | grep errors: | awk '{print $2}')"
  if [ ${HEALTH} != ONLINE ]; then
    exit 1
  fi
  if [ ${ERROR} != "No" ]; then
    exit 1
  fi
done

Finally, make it executable with chmod +x /root/zfs_health_check.sh and restart Monit with service monit restart. If an error occurs, Monit will then send an e-mail with the details of the error and the affected ZFS pool. The script can also be executed manually:

Test

• Remove a hard disc from the fakepool with zpool offline fakepool /poolfiles/file1:
  • Monit then sends an alarm by e-mail
• Add the hard drive back to the fakepool with zpool online fakepool /poolfiles/file1:
  • Monit then sends an all-clear by e-mail

Console

Voilá