up vote 3 down vote favorite
1
share [g+] share [fb]

ZDNET put a very controversial article about RAID5 doomed in 2009 because of bigger and bigger capacities some time ago (2007).

2009 has arrived and I would like to know if it's possible to recheck a RAID array online : read the whole discs and eventually detect a disk failure in advance (ie not subject to any SMART or data-needed basis).

I can think of a very simple way of checking : do a dd if=/dev/md0 of=/dev/null and monitor the RAID logs, but I'm searching of a more integrated way, that could even do some magic with the errors (like relocating them).

link|improve this question
92% accept rate
feedback

2 Answers

up vote 3 down vote accepted

Linux software RAID (md) in many distributions (debian & ubuntu at least) will often perform a resync monthly (first sunday) just in case; look for the "checkarray" script.

NetApp filer's (by default) do a low priority resync every sunday morning.

In practice I don't believe I've ever had either of these catch a problem even with hundreds of systems in production.

link|improve this answer
That's exactly what I was looking for. Incidently I am with Debian, have the checkarray script in crontab, but didn't know it was already running under the hood. – Steve Schnepp May 26 '09 at 11:02
1  
If you install something like the logcheck package you'll be e-mailed the "interesting" parts of syslog, such as the messages md prints while it does this. – LapTop006 May 26 '09 at 11:18
feedback
up vote 2 down vote

Most modern array controllers provide this as the controller level. Dell PERC raid controllers have something called "Patrol Read" which exercises the disks in the array.

ZFS has a feature called "resilvering" which does something similar.

link|improve this answer
feedback

Your Answer

 
or
required, but never shown
discard

Not the answer you're looking for? Browse other questions tagged or ask your own question.