On 12/2/2014 6:24 PM, Fred Smith wrote:
In reality, I had (in my ignorance) purchased a pair of WD drives that aren't intended to be used in a RAID array, and once in a long while (that was actually the only such instance in the 4-5 years I've had that RAID array) it doesn't respond to some HD command or other and gets dropped.
desktop class SATA drives will report 'write successful' when there's still data in its buffers, so the raid will happily continue, then if the drive actually gets a unrecoverable write error, things are toast, the raid is out of sync.
this is a major reason I'm leaning towards using ZFS for future raids (primarily via using FreeBSD), because ZFS checksums and timestamps every block it writes, so it can look at the two blocks that a regular raid can only say "something is wrong here, but what it is I ain't exactly sure" and go "A is good, B is bad/stale, lets replicate A back to B", as part of the zpool 'scrub' process.
-- john r pierce 37N 122W somewhere on the middle of the left coast _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
