Restoring a RAID 5

I would expect this question to be answered in the FAQ, but the site seems to be down. At least I can't get to it now. I'm sure this is simple, but I want to be sure I get it right the first time and don't mess up the data.

I have a system with 5 drives: a system/boot drive and a RAID5 composed of 3 active drives and a spare (formated in ext3). The boot/system drive died on me the other day and I've restored it, re-installing Debian Sarge. The RAID drives were not effected by the failure.

Now that the system is up and running, I want to restore the RAID. I've been searching and tried:

mdadm --examine --scan /dev/hde /dev/hdf /dev/hdg /dev/hdh

and got:

ARRAY /dev/md0 level=raid5 num-devices=3
UUID=d6fd4af2:5e5d0da2:9dec7c5f:ad60c25d
devices=/dev/hdf

I'm a little puzzled by this, since I created the drive specifying hde first, and would expect mdadm to find that drive and not hdf. It also finds only that one drive and does not seem to see hde or hdg as part of the RAID.

When I tried

mdadm --examine /dev/hdf

I got:

/dev/hdf:

          Magic : a92b4efc
        Version : 00.90.00
           UUID : d6fd4af2:5e5d0da2:9dec7c5f:ad60c25d

Creation Time : Fri Dec 9 12:41:00 2005
Raid Level : raid5
Raid Devices : 3
Total Devices : 4
Preferred Minor : 0

Update Time : Fri Dec 9 15:44:21 2005
State : clean
Active Devices : 3
Working Devices : 4
Failed Devices : 0
Spare Devices : 1
Checksum : ba586641 - correct
Events : 0.82

Layout : left-symmetric
Chunk Size : 64K

Number Major Minor RaidDevice State
this 3 34 64 3 spare /dev/hdh

   0     0      33        0        0      active sync   /dev/hde
   1     1      33       64        1      active sync   /dev/hdf
   2     2      34        0        2      active sync   /dev/hdg
   3     3      34       64        3      spare   /dev/hdh

That made me feel better, since it shows me that at least mdadm "knows", somehow, the structure and all the disks in the RAID.

When I type the same for any other drive, I get (depending on the drive letter):

mdadm: No super block found on /dev/hdg (Expected magic a92b4efc, got 00000000)

Is this correct? Is there only one superblock and not one per drive? (I may have misunderstood that.)

I thought, from the man page, that I could type:

mdadm --assemble --scan and have it assemble the RAID again, but that doesn't. I've tried a few combinations, but nothing seems to work. I get errors and it seems I need to include info in the config file to make it work.

I've seen a fair number of tutorials on setting up RAID drives, but almost nothing on restoring one that exists. I guess this would be treated the same as if I were moving it from one computer to another. I've found some info on this in the archives of this list, but nothing that actually addressed this issue (I'm still searching, though).

Before anyone asks about backups, I don't have an exact backup (don't yet have the space and equipment), but I can restore the data. Doing so will take time and I'd really like to know how to take care of this so if it happens in the future with a drive I need restored quickly, I'll be sure to know how to handle it.

I feel like I'm being lazy asking this outright, but should there be just a single command using --scan and --assemble and a few options that will rebuild this and mount it as md0?

Thanks for any help on this -- and if anyone knows of a mirror of the FAQ, I'd like the link since I can't get the original site to load.

Hal

The implication of this is that hde, hdg, hdh no longer have valid raid superblocks. That is a bit of a worry.

There should be one per drive. Something has happened to those drives.

Is there any chance they are on a different controller which maybe sees them as a different size, or something weird like that?

Maybe trying re-arranging where the drives are plugged in and see what "mdadm --examine /dev/xxx" says in each configuration. Somehow you need to find those other superblocks before you can assemble your array.

NeilBrown