I think I may have killed my software RAID5

Sep

What is this storage?
Joined
Jan 15, 2011
Messages
1
I'm afraid that I've totalled my RAID partition. It all started when, after a reboot, my server told me that it couldn't mount my RAID. I then went into manual recovery and ran mdadm --examine, which came up with this.

I tested that drive, and found that it had no errors. I then folled the advice from post #11 of this thread.

Everything seemed to go fine, untill step 4, when the rebuild was much slower than the OP had reported. I then realised that I may have asked for it to rebuild the device with the disks in the wrong order, since the /dev/sd* definiteions change quite randomly on this system for some reason. I immediatly stopped the resync, but I fear that it may be too late.

I'm tired and depressed right now, and I just need to know wehter there's any chance at all to repair the damage, or whether I've messed it up too much to have any choice but to wipe it all and start over.
 

blakerwry

Storage? I am Storage!
Joined
Oct 12, 2002
Messages
4,203
Location
Kansas City, USA
Website
justblake.com
Your array is much larger than the example (8TB vs 1TB), so it does not surprise me that it might take many times longer. I've heard of large RAID 5 resync's taking days. Either way, I don't think this step would take longer in the event of user error.

The creation of the array in step 3 is basically updating only the superblock information - not a problem. Step 4 specifies "--update=resync"; From the mdadm man page:

Code:
The resync option will cause the array to be marked dirty meaning that
 any redundancy in the array (e.g. parity for raid5, copies for raid1) may
 be incorrect. This will cause the raid system to perform a "resync" pass
 to make sure that all redundant information is correct.

The way I read this is that the parity information is being recalculated. Unfortunately, if the disks are in the wrong order, mdadm may be updating what it believes to be parity information when, in fact, it is user data. Unfortunately, I not sure mdadm is smart enough to detect this condition.

My best advice is to try to get the order right and follow the steps and let it rebuild to see what you have left over.
 
Top