Arch Linux

carbs

Member

Registered: 2012-04-08

Posts: 4

Raid assembly issue after upgrade

Hi all,

I have recently upgraded the kernel on my server from 3.2.12 to 3.2.14, but after the upgrade  my "data" array no longer starts automatically causing the boot the fail and I get asked for the maintenance mode password.

On trying to start the array manually I get the following error:

mdadm: WARNING /dev/sdb and /dev/sdb1 appear to have very similar superblocks.
      If they are really different, please --zero the superblock on one
      If they are the same or overlap, please remove one from the
      DEVICE list in mdadm.conf.

Previously I was allowing udev to assign the md* device and use a volume label to mount the filesystem, so I added the arrays to mdadm.conf and rebuilt the initrd image.  But its still not starting the array, its a assembled but set as inactive.  To start it I just have to stop it and then assemble it manually and everything works, but on reboot the same thing happens.

I did also notice some odd output in dmesg when starting the arrays automatically on boot:

[   22.047793] mdadm: sending ioctl 800c0910 to a partition!
[   22.047798] mdadm: sending ioctl 800c0910 to a partition!
[   22.047958] mdadm: sending ioctl 1261 to a partition!
[   22.047961] mdadm: sending ioctl 1261 to a partition!
[   22.048393] mdadm: sending ioctl 800c0910 to a partition!
[   22.048400] mdadm: sending ioctl 800c0910 to a partition!
[   22.048537] mdadm: sending ioctl 1261 to a partition!
[   22.048541] mdadm: sending ioctl 1261 to a partition!
[   22.050415] mdadm: sending ioctl 1261 to a partition!
[   22.050422] mdadm: sending ioctl 1261 to a partition!

I did try the suggestion from the warning about and zero'ed the superblock on that disk, but then on assembly I get the same warning on another disk (sdd & sdd1)... of cause I am using a RAID5 array, so I now need to let the array rebuild before I can zero the other disk, not really what I want to be doing ;-)

If anyone has some advice I would really appreciate it, my array setup is as follows and only md5 is having an issue all other arrays are fine on boot.

/dev/md5:
        Version : 0.90
  Creation Time : Sun Feb  5 23:31:32 2012
     Raid Level : raid5
     Array Size : 5860540416 (5589.05 GiB 6001.19 GB)
  Used Dev Size : 1953513472 (1863.02 GiB 2000.40 GB)
   Raid Devices : 4
  Total Devices : 4
Preferred Minor : 5
    Persistence : Superblock is persistent

    Update Time : Sun Apr  8 10:36:43 2012
          State : clean, degraded, recovering 
 Active Devices : 3
Working Devices : 4
 Failed Devices : 0
  Spare Devices : 1

         Layout : left-symmetric
     Chunk Size : 512K

 Rebuild Status : 3% complete

           UUID : f30de55a:7b1f2ea1:01f9e43d:ac30fbff (local to host server)
         Events : 0.166

    Number   Major   Minor   RaidDevice State
       0       8        1        0      active sync   /dev/sda1
       4       8       17        1      spare rebuilding   /dev/sdb1
       2       8       33        2      active sync   /dev/sdc1
       3       8       49        3      active sync   /dev/sdd1

mcsilva

Member

From: Portugal

Registered: 2009-12-16

Posts: 95

Re: Raid assembly issue after upgrade

It happened to me also and I think I fixed it!
If you downgrade these version packages, I think it will fix the issue:

filesystem-2012.2-2
udev-181-5
mkinitcpio-0.8.5-1

Get them in http://arm.konnichi.com/core/os/i686/

Test it and send the feedback please

falconindy

Developer

From: New York, USA

Registered: 2009-10-22

Posts: 4,111

Website

Re: Raid assembly issue after upgrade

Until I get a new release out, your fallback image will work with mkinitcpio 0.8.6.

https://bugs.archlinux.org/task/29344

Last edited by falconindy (2012-04-09 01:00:28)

---

blog | ±github

carbs

Member

Registered: 2012-04-08

Posts: 4

Re: Raid assembly issue after upgrade

That did it, although I am using x86_64 but I used the same package versions and its all good.

Thanks for the tip, maybe this is a bug?

I have a hardware RAID controller now, so I'm just about to install that to avoid these kinds of problems in the future as I've had a number of software RAID issues in the past 6 months.

carbs

Member

Registered: 2012-04-08

Posts: 4

Re: Raid assembly issue after upgrade

Until I get a new release out, your fallback image will work with mkinitcpio 0.8.6.

https://bugs.archlinux.org/task/29344

Sorry just saw this, glad to see a fix is on the way.  Keep up the good work.