[->UUIDudev] new array not assembled correctly after reboot

Bug #469574 reported by dogandp
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
mdadm (Ubuntu)
Invalid
Undecided
Unassigned

Bug Description

Binary package hint: mdadm

I'm not sure of what actually caused this bug, so I'll give some background:

I have a lot of raid arrays, run off of various partitions on several disks. Everything was working fine - when I would create a new array, upon reboot it would automatically be detected correctly. For safety, I would habitually cat /proc/mdstat > logfile to make sure if there WERE any problems I could manually reassemble it.

At some point I upgrades to 9.04, but I THINK it was before any problems started.

The trigger in my mind was trying to create md66. It created fine, but now upon reboot, a single drive from that array gets assigned to an inactive array, md_d6. In order to get md66 back (which I actually reassembled now as md6), I need to send a bunch of --stop's to md_d6 - it seems, one for each drive in md66, and then I can do an assemble and it works fine.

This problem persists every reboot.

Since then, I have created md7, and the problem also affects THAT array - and also, md10. All of the arrays need to be assembled manually upon reboot.

Is this related to having tried to create md66? or is this a coincendence masking something else? I couldn't find much info here on any of this - nor on the meaning of md_d vs md ...

:~# lsb_release -rd
Description: Ubuntu 9.04
Release: 9.04
:~# apt-cache policy mdadm
mdadm:
  Installed: 2.6.7.1-1ubuntu8
  Candidate: 2.6.7.1-1ubuntu8
  Version table:
 *** 2.6.7.1-1ubuntu8 0
        500 http://us.archive.ubuntu.com jaunty/main Packages
        100 /var/lib/dpkg/status

Revision history for this message
dogandp (ubuntu-dogandpanda) wrote :

I should elaborate - all arrays created BEFORE md66 still autoassemble perfectly.

Revision history for this message
ceg (ceg) wrote :

I think it was in 9.04 when udev switched to use mdadm --incremental.

So you reported what's discussed at http://ubuntuforums.org/showthread.php?p=8407182, and experience those
Bug #550131 initramfs missing /var/run/mdadm dir (loosing state)
Bug #136252 [->UUIDudev] mdadm.conf w/o ARRAY lines but udev/mdadm not assembling arrays. (boot fails)

Revision history for this message
ceg (ceg) wrote :

Your old arrays probably work because they got written to mdadm.conf during the update.

summary: - mdadm does not remember its actions correctly
+ [->UUIDudev] new array not assembled correctly after reboot
Changed in mdadm (Ubuntu):
status: New → Confirmed
Revision history for this message
ceg (ceg) wrote :

workaround:

# start your arrays manualy
# /usr/share/mdadm/mkconf force-generate /etc/mdadm/mdadm.conf
# update-initramfs -k all -u

Revision history for this message
ceg (ceg) wrote :

The workaround generates a mdadm.conf describing active arrays and updates the initramfs for all kernels, so they will contain the new file.

Revision history for this message
Surbhi Palande (csurbhi) wrote :

@dogandp, do you still see the bug?
The workaround that ceg has suggested, seems to be a standard method, when you create a MD array after installation. This is because, by default md does not update initramfs. So you need to do this manually.

A safer way is to first check the contents of /etc/mdadm/mdadm.conf and the output of /usr/share/mdadm/mkconf. mkconf only shows the output of *existing active* arrays. If you already have definitions of array which are not active right now, but you may like to see them active later, then these definitions will be overwritten by mkfconf force-generate.

Marking the bug as Invalid right now as this updating initramfs is expected in such cases. Please do reopen the bug, if you still see it inspite of updating initramfs.

Changed in mdadm (Ubuntu):
status: Confirmed → Invalid
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.