View previous topic :: View next topic |
Author |
Message |
octomancer n00b
Joined: 29 Dec 2006 Posts: 6
|
Posted: Fri Jun 08, 2018 11:48 am Post subject: mdadm still claiming old RAID member drive on boot |
|
|
Hi,
Not sure this is the right place to ask this, apologies if not.
I had a software (mdadm) RAID5 composed of a bunch of 2TB drives in my Gentoo server. It was created a long time ago, the member disks' partition tables were MSDOS ones, a reasonable choice at the time. The partition type was set to "fd", the mdadm auto-assemble partition type. These partitions were created with fdisk. To replace this RAID5, I created a new RAID6 on 4TB drives. I had to use GPT partition tables on those due to the 4TB size. I used parted to create these partitions. Once I'd copied all the data over, I decided that I wanted to use one of the old 2TB drives as a general purpose data drive in the same server. Because I had been using parted, I used it to create a GPT partition on this old 2TB drive. Then I used mkfs.ext4 to create a filesystem. So far so good.
But now, when I boot my server, mdadm "claims" the old 2TB drive, as shown:
Code: | marv ~ # cat /proc/mdstat
Personalities : [raid6] [raid5] [raid4]
md126 : inactive sdb[4](S)
1953514496 blocks
md127 : active raid6 sdg1[5] sdf1[4] sde1[6] sdc1[0] sdd1[1]
11720656896 blocks super 1.2 level 6, 512k chunk, algorithm 2 [5/5] [UUUUU]
bitmap: 0/30 pages [0KB], 65536KB chunk
unused devices: <none> |
This prevents the filesystem on /dev/sdb1 getting mounted at boot time.
Nothing bad is happening, I can stop the bogus raid device with
Code: | mdadm -S /dev/md126 |
and then mount it. But I want this to happen automatically at boot time (it has a correct entry in fstab).
So my question is, what do I have to do to /dev/sdb or /dev/sdb1 to stop mdadm seeing it as a RAID member at boot? I don't know what the issue is, is it the presence of a RAID superblock, is mdadm still seeing the old partition type of "fd", or is it something completely different?
I suspect that I could probably achieve the effect I want by putting something like this in /etc/mdadm.conf
Code: | ARRAY <ignore> devices=/dev/sdb* |
but I'd rather frob /dev/sdb or /dev/sdb1 so I don't need that
Any info greatly appreciated
Cheers,
Rich |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9679 Location: almost Mile High in the USA
|
Posted: Fri Jun 08, 2018 12:32 pm Post subject: |
|
|
Yes, if md detects a superblock on any disk on the system when autodetecting, it will assume that it needs to eventually assemble it and snap it up... I believe this happens regardless of the partition type, it still seems to check them all.
You should wipe the superblock with something like
# mdadm --zero-superblock /dev/sdb
if you don't want it ever showing up again. Backup before using this command unless you want to trash the contents of the drive anyway. _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
octomancer n00b
Joined: 29 Dec 2006 Posts: 6
|
Posted: Fri Jun 08, 2018 12:51 pm Post subject: |
|
|
Thanks for the quick reply.
At the moment, the RAID superblock and the ext4 filesystem are happily co-existing on that drive (/dev/sdb). Unless zeroing the superblock is going to write zeros somewhere other than the superblock, I can't see why it would trash the contents of the drive. I have a spare SATA port in the server, so I'll put another of the old 2TB drives back in it (and zero the superblock before doing anything else and change the partition type to something other than fd with fdisk before creating the GPT partition) and backup the data to it before I try zeroing the superblock on /dev/sdb. But really, if it the ext4 filesystem is happy with a valid superblock there, it should be happy without one. Ooooh, I get do science! I will report back with the results. Hopefully I don't create a monster. |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9679 Location: almost Mile High in the USA
|
Posted: Fri Jun 08, 2018 12:58 pm Post subject: |
|
|
Well,... you were warned. It shouldn't cause trouble, but if it does, wouldn't it be nice if you had a backup? _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
NeddySeagoon Administrator
Joined: 05 Jul 2003 Posts: 54234 Location: 56N 3W
|
Posted: Fri Jun 08, 2018 6:51 pm Post subject: |
|
|
octomancer,
It sounds like you have a version 0.9 raid superblock on that drive. will tell.
That puts the raid superblock at the end of the volume.
The raid superblock is now inside the new filesystem. Zeroing the superblock may or may not be harmless to the filesystem metadata.
A raid version 0.9 superblock has space for 28 volumes, so much of it won't be used. If mkfs.ext4 overwrote an unused part of the superblock, mdadm may not notice.
If zeroing the superblock inside the filesystem trashes filesystem metadata ... well, that's what backups are for. _________________ Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
Back to top |
|
|
octomancer n00b
Joined: 29 Dec 2006 Posts: 6
|
Posted: Fri Jun 15, 2018 1:38 pm Post subject: |
|
|
Hi
You're right Neddy.
Code: |
# mdadm -E /dev/sdb
/dev/sdb:
Magic : a92b4efc
Version : 0.90.00
UUID : 03a399e7:13b3a40d:8ff0046e:0760338e
Creation Time : Fri Apr 30 23:14:40 2010
Raid Level : raid5
Used Dev Size : 1953134400 (1862.65 GiB 2000.01 GB)
Array Size : 9765672000 (9313.27 GiB 10000.05 GB)
Raid Devices : 6
Total Devices : 5
Preferred Minor : 127
...
|
I didn't actually get around to doing anything with this, I did indeed find that
Code: | ARRAY <ignore> devices=/dev/sdb* |
in mdadm.conf gets around the problem. I'll sort it out this weekend, replace the current drive with one with a properly zeroed superblock.
Cheers both
R |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|