This script is using notify-send to notify the user about problems with mdadm (software RAID) and ZFS.
To see how these notifications look like type in your terminal this:
notify-send "This is how the notification look."
It looks for suspicious strings in the output of the command sbin/zpool status and looks for the symbol _ (underscore) in the file /proc/mdstat.
The script must be started as the user who will receive the notification.
For example, if the script is located at /home/user/bin/zfs-health.bash the file /var/spool/cron/crontabs/user contains:
# zfs health monitor
* * * * * /home/user/bin/zfs-health.bash
This way the script is started by cron every minute.
Don't forget to chmod +x /home/user/bin/zfs-health.bash and test it by editing the script (after making a backup!) like this:
condition=$(/sbin/zpool status | egrep '(ONLINE|DEGRADED|FAULTED|OFFLINE|UNAVAIL|REMOVED|FAIL|DESTROYED|corrupt|cannot|unrecover)')
(I inserted "ONLINE|" before "DEGRADED".)
After about 1 minute or less you shold receve a notification. If you don't - something is not right. Check your cron configuration.
If you don't have ZFS (you use the script to monitor your software RAID) edit it like this:
if grep "U" /proc/mdstat ; then
(I replaced the underscore (_) symbol with U.)
After reverting the edits make sure the file is still with executable permission.