Tell me more ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.
up vote 1 down vote favorite

A drive dropped out of my raid 5 array yesterday. It looks like the reason was due to a bad controller so I switched it out and attempted to re-add the drive but mdadm claimed it couldn't do it. So I zerod the superblock and just added the drive normally and left it to resync.

When I came to check on the array this morning I was unable to mount it at all and it's now showing as CLEAN FAULTY with two drives missing. The two missing drives are listed as spare and faulty spare.

Is there anything I can do in this situation or is the array gone?

Update

The disks appear to be fine - except maybe for enough bad data on one of the disks for mdadm to get annoyed and kick that disk from the array too.

I was able to recreate the array by marking the disk as working and forcing the assembly so I'm currently just making sure that all my backups are up to date.

So I can probably change this question to: RAID5 seems to be a problem with large disks (3x3TB). I'm considering changing to mirrored RAID-Z arrays but is there anything else I should consider instead?

share|improve this question
You didn't say if the two drives are actually working, or not. If they are effectively dead, then there is no hope. Otherwise the answer would depend on what have actually happened - please attach the relevant syslogs. – Adam Ryczkowski Oct 3 '12 at 6:41
Don't have access to the logs at the minute. I'll post them tonight. – Dean Reilly Oct 3 '12 at 10:45
If you value your data, also stop using consumer-grade SATA drives. – Michael Hampton Oct 3 '12 at 13:07

1 Answer

up vote 2 down vote accepted

It is not recommended to use raid-5 on 3TB consumer disks because it take ages to resync (might be well more than 24h), and during that time it is likely that yet another one might fail, and in such case all your data (or at least the part, that haven't managed to get resynced) will be lost.

Raid-Z has small the advantage that it resyncs (resilvers) only that part of hard drive, which actually is used by files as compared to standard raid implementation which is filesystem-agnostic.

Another advantage of zfs is that it is said (I've never tried it myself. See article on serverfocus.org) that you can specify the order of files that are resilvered; files important can get high priority, which translates to being first to resilver.

I suggest to go with mirroring, which is faster and better error-proof for such large drives.

share|improve this answer
In general, RAID-6 is much preferred over RAID-5, but both of them have the issue of long rebuild times. Using "--bitmap=internal" on the arrays may help, but switching to a RAID 1+0 or RAID-10 array is also recommended (shorter rebuilds). – tgharold Mar 27 at 11:40

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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