I've got home mounted on a LVM partition over two RAID 1 arrays.
Unfortunately one of the disks in the second array has dropped out. I
Just to check I understand,
- you have 4 physical hard disks (2 x 2TB and 2x ??).
- you have two RAID1 arrays comprising of 2 physical hard disks each.
- one of the arrays is healthy and the other is degraded
- both RAID1 arrays have been made a member of the same LVM VG (volume
group).
- your have a LV (logical volume) named "home" which belongs to this VG.
At the moment it looks like if the remaining disk in the degraded RAID1
array fails you will lose your home LV. Backup the contents of your home
LV immediately if you have not already done so!
If the healthy RAID1 array has enough free space to hold the entire home
LV I would remove the degraded RAID1 array from the volume group for the
time being.
http://tldp.org/HOWTO/LVM-HOWTO/removeadisk.html
can still fdisk -l it, but when I ran dd if=/dev/sdd1 of=/dev/null it
got about 132M through a 2Tb drive before crashing out. This doesn't
bode well.
From memory, someone please correct me if I am wrong, dd will fail at
the first unreadable sector unless it is told to skip on error
(conv=noerror?).
dd failing is not necessary fatal, but its never reassuring, bear in
mind the disk will not remap the sector until a _write_ to that sector
fails.
You seem to be using dd just to test you can read the disk
(of=/dev/null) but I will mention this anyway: I tend to use ddrescue,
from the gddrescue package, for making images of failing hard disks as
it keeps a log of which sectors it has failed to read (and you can use
this log to repeatedly retry reading the failed sectors, sometimes it
works!)
I can't really afford a new drive, but was wondering if I should just
reformat, try and check it for errors and then try and add it back to
the array? I was hoping that doing that might allow it to exclude any
bad blocks and get me something working for a while whilst I save up.
keep the "failed" disk in a safe place until you have a backup (or at
the very least you are not relying on the degraded raid1 array), you
might still be able to read something useful from it if the remaining
"healthy" disk goes up in smoke.
Once your data is definitely safe, I would
1) run a long SMART test of the disk using smartmontools,
2) if the smart test did not fail, I would run badblocks against the
"failed" disk. Badblocks, doing write-mode test (-w), will overwrite the
entire disk (causing failed sectors to be remapped) and then read the
entire disk back to ensure that the "pattern" that was written can still
be read. This could take a long time on a 2TB disk!
3) run another long SMART test, badblocks might cause enough sectors to
be remapped that SMART will now consider the drive to be failing (even
if the badblocks test passes).
If 1, 2 & 3 pass you probably don't need to replace the disk
If anyone has any comments, suggestions or queries, I'd be glad to hear
them,
Thanks,
James
If the disks are of a similar size, you might be better off with RAID6
over RAID1 as this will allow *any* 2 disks to fail instead of *up to* 2
disks to fail before your array becomes unreadable.
In theory you could create 4 partitions of equal size over the 4 disks
and create a RAID6, and partition off the remaining space of the 2
larger disks for a RAID1.
You might also want to consider if you need raid at all! you could use
half the disks to store incremental backups of the other disks (RAID is
not a backup etc).
I hope this helps somewhat, I feel I might just be telling you what you
already know!
--
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/54de220a.9000...@xact.uk.com
