A someone kludgey way to do it could be to monitor /proc/mdstat just as if were a normal file, and setup Hobbit to monitor it just like you would /var/log/messages, triggering on a pattern matched error string. This would allow you to have it setup server-side like you want.
Not the best solution, I know, but it popped into my head :) You could also configure mdadm on the client machines to send an email where there are RAID issues.
-Charles
▸ quoted from Dennis Ortsen
Dennis Ortsen wrote:
Hi All,
I've got a shell script that checks the status of the Linux md devices (cat
/proc/mdstat). It is currently intended to be run on the client itself (drop
it in $BBHOME/ext and add some config to $BBHOME/etc/clientlaunch.cfg). It's
a script from the RAID category on www.deadcat.net. Is it possible to have
the script installed only on the hobbit server and use the clientchannel to
retrieve the contents of /proc/mdstat and let the hobbit server do the test?
A new column (raid i.e.) would be preferable. A rule in
$BBHOME/etc/hobbit-clients.cfg would be very nice.