I have a RAID5 array with three disks. One of the disks appeared to have overheated and shut itself down. Then, md failed the drive (as expected) and I got a message. After letting the system cool down, I resynced the failed drive. Well, apparently, it was pretty hot in there, because, it appears that another drive went offline during the resync and it got kicked off, too, leaving only one drive in the array.

So, I let it *really* cool down (like, overnight) and am now trying to get it back together.

Here's the particulars:
/dev/md1 (RAID5 array)
/dev/sda1 (the first overheated disk)
/dev/hdc2 (the OK disk)
/dev/hda5 (the second overheated disk?)

I can reassemble the array using:

After this I get my array back, but it is "inactive". It has assembled /dev/hdc2 into the array, set /dev/sda1 as a "spare" and dmesg reports that /dev/hda5 is "non-fresh" and has been kicked from the array.

If I "freshen" /dev/hda5 and reassemble into the array, will I be able to read the data?

Subsequently, why does /dev/sda1 show up as a "spare" and how do I rejoin it to the array as one of the active disks?

Extra info:
-- Linux 2.6.18-1.2257.fc5
-- No writing was going on to these disks during this whole debacle (well, none that I know of)

Thanks for your help,

DS

You can assemble the drive with the force option (which will accept an out of date superblock). Then run fsck on it to replay the journal. Mount, and you should be back in business while the drives sync up.

Hey, thanks a million, macemoneta!

I was guessing that this would work, but was reluctant to try without the benefit of someone else that was confident that --force would not destroy all my data.

I ran the assemble again with the --force option and it immediately reassembled itself and began to resync. In an hour's time, the array was rebuilt. I rebooted and everything came up normal.

Thanks, again.

--DS