Data corruption after raid migration (raid5 to raid6 in 2.6.31 with mdadm 3.1)

Data corruption after raid migration (raid5 to raid6 in 2.6.31 with mdadm 3.1)

[KueiHuan Chen]

Dear Neil,

  I encounter a little problem when perform the raid5 to raid6
migration. The result is data corruption if i stop the raid while
reshape and then reassemble the raid. Here is the operation oder:

Environment : linux-2.6.31 . mdadm 3.1

./mdadm -C /dev/md2 -R -amd -l5 -n4 /dev/sd[cdef]3 -z 1000000
mkfs.ext3 /dev/md2
mount /dev/md2 /volume1/
cp /bin/* /volume1/
./mdadm /dev/md2 -a /dev/sdg3
./mdadm --grow --level=6 --layout=default --backup-file=/root/bak2 /dev/md2
umount /dev/md2
./mdadm -S /dev/md2  (Stop it before reshape end)
./mdadm -A /dev/md2 --backup-file=/root/bak2 /dev/sd[cdefg]3
mount /dev/md2 /volume1  <--- This will cause an error that file
system could not be mounted.

  The interesting thing is if i don't stop the raid while reshape. The
data will not corruption. Is it means the operations not permitted
while the reshape process ?

Thanks.

Best Regards,
KueiHuan-Chen
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Data corruption after raid migration (raid5 to raid6 in 2.6.31 with mdadm 3.1)

[Michael Evans]

My copy of mdadm 3.1 didn't want me to reshape to raid6 without using
linux-2.6.32 or later (I used the latest RC: 2.6.32-rc5-git5).

Re: Data corruption after raid migration (raid5 to raid6 in 2.6.31 with mdadm 3.1)

[KueiHuan Chen]

Dear Michael,

   Did you also reduce the raid disks ? That operation need 2.6.32 and
later. Others such like level change is working at 2.6.31.

Thanks.

Best Regards,
KueiHuan-Chen



2009/11/6 Michael Evans <mjevans1983@gmail.com>:
> My copy of mdadm 3.1 didn't want me to reshape to raid6 without using
> linux-2.6.32 or later (I used the latest RC: 2.6.32-rc5-git5).
>

Re: Data corruption after raid migration (raid5 to raid6 in 2.6.31 with mdadm 3.1)

[Michael Evans]

Yeah, I did it the wrong way the first time and wound up with a raid-6
array that expected one more drive than I had.  It would be a lot
easier if I could have specified the target number of destination
stripes and then also specified the number of parity stripes.

I needed to reduce the added space back out, so that it could be added
as a parity device instead of another data-device.

On Thu, Nov 5, 2009 at 5:45 PM, KueiHuan Chen <kueihuan.chen@gmail.com> wrote:
> Dear Michael,
>
>   Did you also reduce the raid disks ? That operation need 2.6.32 and
> later. Others such like level change is working at 2.6.31.
>
> Thanks.
>
> Best Regards,
> KueiHuan-Chen
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html