Skip to contentSkip to conversation

BUGREPORT: mdadm v2.0-devel - Examine and zero-superblock misbehaving

DISCUSSION Tyler 2 participants 3 messages May 3, 2005 MD RAID

lore
Share & tools
lore

Tyler

author
#0 lore raw

On a site note, mdadm --zero-superblock doesn't seem to work correctly with mdadm v2.0-devel, it reports no error when you use it, but you can mdadm -E /dev/hdX after using it, and there is still a superblock, if you use mdadm v1.9.0-2, it works correctly, and the examine shows nothing afterwards. Examples follow after the first problem. Mdadm v2.0 also seems to misreport the findings using the examine function.. "381 spares." Using kernel 2.6.12-rc3-mm2.

First i run zero superblock with mdadm v2.0-devel from current dir:

root@localhost:~/dev/mdadm-2.0-devel-1# ./mdadm --zero-superblock /dev/hdc

Then examine the superblock using mdadm v2.0-devel from current dir:

root@localhost:~/dev/mdadm-2.0-devel-1# ./mdadm -E /dev/hdc
/dev/hdc:
         Magic : a92b4efc
       Version : 01.00
    Array UUID : f4b2dc026e:a3cc3e95:de0563f5:199250
          Name :
 Creation Time : Tue May  3 11:16:22 2005
    Raid Level : raid5
  Raid Devices : 3

Device Size : 390721952 (186.31 GiB 200.05 GB)
Super Offset : 390721952 sectors
State : active
Device UUID : f4b2dc026e:a3cc3e95:de0563f5:199250 Update Time : Tue May 3 11:16:22 2005
Checksum : d4cd5d80 - correct
Events : 0

Layout : -unknown-
Chunk Size : 64K

Array State : Uuu 381 spares

Then i try using mdadm v1.9.0 (from system path):

root@localhost:~/dev/mdadm-2.0-devel-1# mdadm --version
mdadm - v1.9.0 - 04 February 2005
root@localhost:~/dev/mdadm-2.0-devel-1# mdadm -E /dev/hdc
mdadm: No super block found on /dev/hdc (Expected magic a92b4efc, got 00000000)

Then i try erasing the superblock again with mdadm v1.9.0 from system path:

root@pink:~/dev/mdadm-2.0-devel-1# mdadm --version
mdadm - v1.9.0 - 04 February 2005
root@pink:~/dev/mdadm-2.0-devel-1# mdadm --zero-superblock /dev/hdc
mdadm: /dev/hdc does not appear to have an MD superblock.

Then examine it again with mdadm v2.0-devel from current dir:

root@pink:~/dev/mdadm-2.0-devel-1# ./mdadm -E /dev/hdc
/dev/hdc:
         Magic : a92b4efc
       Version : 01.00
    Array UUID : f4b2dc026e:a3cc3e95:de0563f5:199250
          Name :
 Creation Time : Tue May  3 11:16:22 2005
    Raid Level : raid5
  Raid Devices : 3

Device Size : 390721952 (186.31 GiB 200.05 GB)
Super Offset : 390721952 sectors
State : active
Device UUID : f4b2dc026e:a3cc3e95:de0563f5:199250 Update Time : Tue May 3 11:16:22 2005
Checksum : d4cd5d80 - correct
Events : 0

Layout : -unknown-
Chunk Size : 64K

Array State : Uuu 381 spares

#1 lore raw
Quoted 10 lines · On Tuesday May 3, [email protected] wrote:

On a site note, mdadm --zero-superblock doesn't seem to work correctly with mdadm v2.0-devel, it reports no error when you use it, but you can mdadm -E /dev/hdX after using it, and there is still a superblock, if you use mdadm v1.9.0-2, it works correctly, and the examine shows nothing afterwards. Examples follow after the first problem. Mdadm v2.0 also seems to misreport the findings using the examine function.. "381 spares." Using kernel 2.6.12-rc3-mm2.

First i run zero superblock with mdadm v2.0-devel from current dir:

Quoted 8 lines
root@localhost:~/dev/mdadm-2.0-devel-1# ./mdadm --zero-superblock /dev/hdc

Then examine the superblock using mdadm v2.0-devel from current dir:

root@localhost:~/dev/mdadm-2.0-devel-1# ./mdadm -E /dev/hdc
/dev/hdc:
          Magic : a92b4efc
        Version : 01.00

What has happened here is that "mdadm --zero-superblock" found a version 0.90 superblock and zeroed it. There was still a version 1 superblock (which lives in a different location). and -E found that. If you run --zero-superblock serveral times it should zero everything, though normally you would not expect to have multiple superblocks.

Quoted 3 lines

Array State : Uuu 381 spares

Not sure about this. I'll have to spend a while reminding my self what is meant to happen, and what actually does...

NeilBrown

author
#2 lore raw

Hi Neil,

I tried removing the superblock several times to test the theory, but it seems the removal isn't working.. maybe because of the semi-invalid information of 380 spare drives? I'm not sure?

root@localhost:~/dev/mdadm-2.0-devel-1# ./mdadm --zero-superblock /dev/hdc
root@localhost:~/dev/mdadm-2.0-devel-1# ./mdadm --zero-superblock /dev/hdc
root@localhost:~/dev/mdadm-2.0-devel-1# ./mdadm --zero-superblock /dev/hdc
root@localhost:~/dev/mdadm-2.0-devel-1# ./mdadm -E /dev/hdc
/dev/hdc:
         Magic : a92b4efc
       Version : 01.00
    Array UUID : f4b2dc026e:a3cc3e95:de0563f5:199250
          Name :
 Creation Time : Tue May  3 11:16:22 2005
    Raid Level : raid5
  Raid Devices : 3

Device Size : 390721952 (186.31 GiB 200.05 GB)
Super Offset : 390721952 sectors
State : active
Device UUID : f4b2dc026e:a3cc3e95:de0563f5:199250 Update Time : Tue May 3 11:16:22 2005
Checksum : d4cd5d80 - correct
Events : 0

Layout : -unknown-
Chunk Size : 64K

Array State : Uuu 381 spares

root@localhost:~/dev/mdadm-2.0-devel-1# ./mdadm -E /dev/hdd
root@localhost:~/dev/mdadm-2.0-devel-1# ./mdadm --zero-superblock /dev/hdd
mdadm: Unrecognised md component device - /dev/hdd
Quoted 39 lines · Neil Brown
Quoted 4 lines · On Tuesday May 3, [email protected] wrote:

On a site note, mdadm --zero-superblock doesn't seem to work correctly with mdadm v2.0-devel, it reports no error when you use it, but you can mdadm -E /dev/hdX after using it, and there is still a superblock, if you use mdadm v1.9.0-2, it works correctly, and the examine shows nothing afterwards. Examples follow after the first problem. Mdadm v2.0 also seems to misreport the findings using the examine function.. "381 spares." Using kernel 2.6.12-rc3-mm2.

First i run zero superblock with mdadm v2.0-devel from current dir:

Quoted 10 lines
root@localhost:~/dev/mdadm-2.0-devel-1# ./mdadm --zero-superblock /dev/hdc

Then examine the superblock using mdadm v2.0-devel from current dir:

root@localhost:~/dev/mdadm-2.0-devel-1# ./mdadm -E /dev/hdc
/dev/hdc:
         Magic : a92b4efc
       Version : 01.00

What has happened here is that "mdadm --zero-superblock" found a version 0.90 superblock and zeroed it. There was still a version 1 superblock (which lives in a different location). and -E found that. If you run --zero-superblock serveral times it should zero everything, though normally you would not expect to have multiple superblocks.

Quoted 4 lines

Array State : Uuu 381 spares

Not sure about this. I'll have to spend a while reminding my self what is meant to happen, and what actually does...

NeilBrown