Re: Failed JBOD RAID on old NAS, how to diagnose/resurrect?
From: Song Liu <song@kernel.org>
Date: 2020-03-10 21:20:02
On Sun, Mar 8, 2020 at 7:15 AM Chris Green [off-list ref] wrote:
Well I've got it working again but I'm very confused as to *why* it
failed the way it did.
A 'cat /proc/mdstat' produced:-
Personalities : [linear] [raid0] [raid1]
md4 : active raid1 sda4[0]
973522816 blocks [2/1] [U_]
md1 : active raid1 sdb2[0] sda2[1]
256960 blocks [2/2] [UU]
md3 : active raid1 sdb3[0] sda3[1]
987904 blocks [2/2] [UU]
md2 : active raid1 sdb4[0]
973522816 blocks [2/1] [U_]
md0 : active raid1 sdb1[0] sda1[1]
1959808 blocks [2/2] [UU]
So md2 and md4 (the main parts of the two 1Tb disk drives) seemed to
be OK from the RAID point of view. But I noticed that the block
device for /dev/md4 didn't exist:-
~ # ls -l /dev/md*
brw-r----- 1 root root 9, 0 Sep 29 2011 /dev/md0
brw-r----- 1 root root 9, 1 Sep 29 2011 /dev/md1
brw-r----- 1 root root 9, 10 Sep 29 2011 /dev/md10
brw-r----- 1 root root 9, 11 Sep 29 2011 /dev/md11
brw-r----- 1 root root 9, 12 Sep 29 2011 /dev/md12
brw-r----- 1 root root 9, 13 Sep 29 2011 /dev/md13
brw-r----- 1 root root 9, 14 Sep 29 2011 /dev/md14
brw-r----- 1 root root 9, 15 Sep 29 2011 /dev/md15
brw-r----- 1 root root 9, 16 Sep 29 2011 /dev/md16
brw-r----- 1 root root 9, 17 Sep 29 2011 /dev/md17
brw-r----- 1 root root 9, 18 Sep 29 2011 /dev/md18
brw-r----- 1 root root 9, 19 Sep 29 2011 /dev/md19
brw-r----- 1 root root 9, 2 Sep 29 2011 /dev/md2
brw-r----- 1 root root 9, 20 Sep 29 2011 /dev/md20
brw-r----- 1 root root 9, 21 Sep 29 2011 /dev/md21
brw-r----- 1 root root 9, 22 Sep 29 2011 /dev/md22
brw-r----- 1 root root 9, 23 Sep 29 2011 /dev/md23
brw-r----- 1 root root 9, 24 Sep 29 2011 /dev/md24
brw-r----- 1 root root 9, 25 Sep 29 2011 /dev/md25
brw-r----- 1 root root 9, 26 Sep 29 2011 /dev/md26
brw-r----- 1 root root 9, 27 Sep 29 2011 /dev/md27
brw-r----- 1 root root 9, 28 Sep 29 2011 /dev/md28
brw-r----- 1 root root 9, 29 Sep 29 2011 /dev/md29
brw-r----- 1 root root 9, 3 Sep 29 2011 /dev/md3
brw-r----- 1 root root 9, 5 Sep 29 2011 /dev/md5
brw-r----- 1 root root 9, 6 Sep 29 2011 /dev/md6
brw-r----- 1 root root 9, 7 Sep 29 2011 /dev/md7
brw-r----- 1 root root 9, 8 Sep 29 2011 /dev/md8
brw-r----- 1 root root 9, 9 Sep 29 2011 /dev/md9
The fix was simply to use 'mknod' to create the missing /dev/md4, now
I can mount the drive and see the data.
What I don't understand is where /dev/md4 went, how would it have got
deleted? I have yet to reboot the system to see if /dev/md4
disappears again but if it does it's not a big problem to create it
again.
Should the RAID block devices get created as part of the RAID start
up? Maybe there's something gone awry there.Do you have proper /etc/md.conf? Thanks, Song