Bug 240952 - cannot boot root on raid with 2.6.21-1.3175.fc7
Summary: cannot boot root on raid with 2.6.21-1.3175.fc7
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: Fedora
Classification: Fedora
Component: anaconda
Version: rawhide
Hardware: All
OS: Linux
medium
high
Target Milestone: ---
Assignee: Anaconda Maintenance Team
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2007-05-23 11:43 UTC by cje
Modified: 2007-11-30 22:12 UTC (History)
1 user (show)

Fixed In Version: F7
Clone Of:
Environment:
Last Closed: 2007-06-04 15:56:44 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
mdadm.conf as written by anaconda (152 bytes, application/octet-stream)
2007-05-23 23:52 UTC, Jon Burgess
no flags Details
partition layout table as it appears in anaconda (14.63 KB, application/x-abiword)
2007-05-31 21:13 UTC, cje
no flags Details

Description cje 2007-05-23 11:43:19 UTC
Description of problem:
booting fails when root is on a raid array.  error is:
mdadm: failed to RUN_ARRAY /dev/md1: Invalid argument

followed shortly by a Kernel panic (obviously).

it works fine with 2.6.21-1.3163


Version-Release number of selected component (if applicable):
2.6.21-1.3175.fc7

How reproducible:
always

Comment 1 Jon Burgess 2007-05-23 23:52:06 UTC
Created attachment 155309 [details]
mdadm.conf as written by anaconda

This looks like a bug I just experienced on upgrading to 3175. I think it is
due to the recent mkinitrd update which now uses the /etc/mdadm.conf file
written by anaconda to perform the raid start. 

On my machine I found the file generated by anaconda (see attachment) had no
entry for /dev/md1. md1 is the LVM PV containing / etc so caused a mount
failure just like you are seeing.

What does your /etc/mdadm.conf look like?

I managed to fix my setup by doing:

# mdadm --query --detail --brief --scan > /etc/mdadm.conf

Then rebuilt the initrd by removing and re-installing the '3175 kernel and all
was fine.

Comment 2 Jon Burgess 2007-05-24 00:03:26 UTC
I should mention that anaconda had problems with the raid setup during the fc6
-> f7t4 upgrade so that may have contributed to the bad mdadm.conf file. See bug
238926 for further details.

Comment 3 cje 2007-05-24 08:40:01 UTC
thanks!  the mdadm.conf file certainly was the problem.  the md1 line had the
'level' and 'uuid' copied from the md0 line when they should have both been
different.  using mdadm --query as suggested fixed the problem.

i'll mark this as duplicate but add a note to the open bug because this
mdadm.conf was created by anaconda on a fresh install of F7T4, not an upgrade.

*** This bug has been marked as a duplicate of 151653 ***

Comment 4 Will Woods 2007-05-27 20:08:17 UTC
Okay, I think we've determined that this bug is not actually a dup of 151653.
This bug concerns anaconda possibly creating incorrect mdadm.conf files.

What was the disk layout / RAID setup you were trying to create? Does a fresh
install of rawhide or F7rc2 do the right thing?

Comment 5 cje 2007-05-30 23:10:37 UTC
hi.  sorry, been away for a while.  thanks for looking into this.

here's the contents of /etc/mdadm.conf (corrected) and /proc/mdstat and the
output of mount:

[~]$ cat /etc/mdadm.conf 
ARRAY /dev/md1 level=raid0 num-devices=2 UUID=1616e478:4a8b6b93:9d351b4e:3624de5b
ARRAY /dev/md0 level=raid1 num-devices=2 UUID=61bc5504:64623059:dadcf744:9253fda9
[~]$ cat /proc/mdstat 
Personalities : [raid0] [raid6] [raid5] [raid4] [raid1] 
md0 : active raid1 sda1[0] sdb1[1]
      104320 blocks [2/2] [UU]
      
md1 : active raid0 sda2[0] sdb2[1]
      77946880 blocks 256k chunks
      
unused devices: <none>
[cje@bed ~]$ mount
/dev/mapper/VolGroup00-LogVol01 on / type ext3 (rw)
proc on /proc type proc (rw)
sysfs on /sys type sysfs (rw)
devpts on /dev/pts type devpts (rw,gid=5,mode=620)
tmpfs on /dev/shm type tmpfs (rw)
/dev/md0 on /boot type ext3 (rw)
none on /proc/sys/fs/binfmt_misc type binfmt_misc (rw)
sunrpc on /var/lib/nfs/rpc_pipefs type rpc_pipefs (rw)

hope that makes the config clear.  i've made a volume group on md1 but don't
know how to show that.

i'm downloading RC2 now and will try it as soon as i can but bittorrent says
it'll be a while yet.  hopefully give it a go tomorrow evening. (around 1900 UTC
31 May) .. if it speeds up a bit.

Comment 6 cje 2007-05-31 21:13:41 UTC
Created attachment 155843 [details]
partition layout table as it appears in anaconda

i'm trying out RC2 now while i download F7 (it did speed up .. but then F7 was
released a few hours later!)

on another system with a similar layout i tried upgrading from FC4 to F7RC2
today and it booted just fine (lots of problems after that, just not _this_
problem).

Comment 7 cje 2007-05-31 21:55:20 UTC
F7RC2 seems to be okay.  i've done a clean install with the above config and
it's booted fine.  it's produced a rather different mdadm.conf:


# mdadm.conf written out by anaconda
DEVICE partitions
MAILADDR root
ARRAY /dev/md0 super-minor=0
ARRAY /dev/md1 super-minor=1

i'll do a final test with F7 tomorrow.  in today's test i used a VNC install and
kept the old partition layout (just marked /, /boot and swap to be formetted) -
tomorrow i'll do a regular install and change the partition sizes a bit to make
sure it's definitely doing a clean setup.

Comment 8 cje 2007-06-04 15:56:44 UTC
did the test with F7, non-VNC install and a modified layout and all worked well.


Note You need to log in before you can comment on or make changes to this bug.