It's really important to periodically scrub raid arrays. From http://gentoo-wiki.com/HOWTO_Gentoo_Install_on_Software_RAID#Data_Scrubbing: During raid reconstruction, if you run across a previously undetected unreadable block, you may not be able to reconstruct your array without data corruption. The larger the disk, the higher the odds that passive bad block detection will be inadequate. Therefore, with today's large disks it is important to actively perform data scrubbing on your array. Unfortunately, we cannot do this on sme 7.x because of the kernel not having the required feature. It could be done on sme8, with a cron job (daily or weekly) that simply repeats echo check >> /sys/block/mdX/md/sync_action for each array. I've been bitten by this problem, defective disk sectors were in an unused PE at the end of a disk. When the other disk failed I wasn't able to sync and I had to resort to a restore from backup over a new install.
Closing as Bug 6160 shows that this is active.
(In reply to comment #3) > Upstream fixed this bug, but we fall in a case where we still need an > unreleased fix, our / raid set is never checked. > https://bugzilla.redhat.com/show_bug.cgi?id=523000 > > I think we will pick this up with a future update. AFAIK this is not true anymore, according to bug 6320. See the link in that bug to RedHat's bugzilla for a more detailed explanation. Should this be closed?
Upstream fixed this bug, but we fall in a case where we still need an unreleased fix, our / raid set is never checked. https://bugzilla.redhat.com/show_bug.cgi?id=523000 I think we will pick this up with a future update.
(In reply to comment #1) > Two bugs are dealing with this upstream: > https://bugzilla.redhat.com/show_bug.cgi?id=233972 > https://bugzilla.redhat.com/show_bug.cgi?id=474436 > > Should we patch mdadm (like fedora) or simply add the cron job to e-smith-base? I don't see any value in patching mdadm just to get a cron job script. That script could be added to e-smith-base. We would need to anticipate an updated mdadm eventually arriving that includes the cron job. We could, for instance, use a different path for the script, and in our copy of the script, check for the existence of the one from mdadm.
Two bugs are dealing with this upstream: https://bugzilla.redhat.com/show_bug.cgi?id=233972 https://bugzilla.redhat.com/show_bug.cgi?id=474436 Should we patch mdadm (like fedora) or simply add the cron job to e-smith-base?