View Full Version : [ubuntu] RAID Array Not Recognized After Reboot

November 25th, 2008, 03:51 AM
I am having a problem with my RAID1 array and I am hoping I could find some help here. I have two 250 GB disks (and I am NOT using the array for root or home partitions). I have gone through all the steps that (I believe) are needed, however, the array doesn't automatically start work when I reboot. After reboot, if I type the two following commands, everything works like a charm. So, my question is...what am I missing?

Here are the two commands that, once typed after reboot, everything is splendid:

sudo mdadm --assemble /dev/md0
sudo mount -t ext3 /dev/md0 /media/raid1/

The two disks that make up the array are SATA, 250 GB and /dev/sdb and /dev/sdc.

Because the two commands above, I am fairly convinced that the RAID array is ok...it just doesn't start on reboot.

Any help is greatly appreciated!

Thanks, Thad

November 25th, 2008, 04:05 PM
Did it ever start on re-boot?

Do you have a line in /etc/fstab to mount the array?

November 25th, 2008, 05:00 PM
No, it never restarted after a reboot. Here is the line in my /etc/fstab file:

/dev/md0 /media/RAID1 ext3 defaults 0 0

Any thoughts? What else can I verify. This seemingly simple problem is driving me crazy.

November 25th, 2008, 05:07 PM
You did make the mountpoint at /media/RAID1 using mkdir?

If so here's what I'd do in this order: delete /etc/mdadm/mdadm.conf, remove mdadm using apt-get, re-boot. Re-install mdadm using apt-get and then run this:

sudo mdadm --assemble --scan

That will auto-create a new mdadm.conf file and all should be mountable now. See what happens!

November 25th, 2008, 05:46 PM
Thanks Frank. Yes, the directory is there for the mount point. I will try the instructions you provided when I get home tonight...thanks so much for the help.

November 26th, 2008, 02:56 AM
Ok, I did what you suggested, but still no luck. After I reboot, I type the following command:

sudo mdadm --detail /dev/md0

and I get the following message:

mdadm: md device /dev/md0 does not appear to be active.

So, I go ahead and type the following and then everything works fine.

sudo mdadm --assemble /dev/md0

sudo mount -t ext3 /dev/md0 /media/raid1/

Any additional thoughts? In my prior message I had /etc/fstab pointing to /media/RAID1. I have since changed fstab to point to /media/raid1.

Thanks in advance for any help. This is driving me crazy.

November 26th, 2008, 07:01 PM
Okay, what does the mdadm.conf file show?

And also:

sudo mdadm -D /dev/md0

after you have assembled and mounted it manually?


df -m

December 1st, 2008, 01:16 AM
Thanks for sticking with me on this one. Here is the info you asked for:

1) mdadm.conf says the following:

# Please refer to mdadm.conf(5) for information about this file
# by default, scan all partitions (/proc/partitions) for MD superblocks
# alternatively, specify devices to scan, using wildcards if desired.
DEVICE partitions
# auto-create devices with Debian standard permissions
CREATE owner=root group=disk mode=0660 auto=yes
# automatically tag new arrays as belonging to the local system
HOMEHOST <system>
# instruct the monitoring daemon where to send mail alerts
# definitions of existing MD arrays
ARRAY /dev/md0 level=raid1 num-devices=2 UUID=a8cb78ec:cef973a4:a95107c4:8449513d
#This file was auto-generated on Tue, 25 Nov 2008 19:47:31 -0600
# by mkconf

2) sudo mdadm -D /dev/md0 says:

/dev/md0: Version : 00.90.03
Creation Time : Tue Nov 25 17:59:58 2008
Raid Level : raid1
Array Size : 244195904 (232.88 GiB 250.06 GB)
Used Dev Size : 244195904 (232.88 GiB 250.06 GB)
Raid Devices : 2
Total Devices : 2
Preferred Minor : 0
Persistence : Superblock is persistent
Update Time : Sun Nov 30 17:23:24 2008
State : clean
Active Devices : 2
Working Devices : 2
Failed Devices : 0
Spare Devices : 0
UUID : a8cb78ec:cef973a4:a95107c4:8449513d (local to host gollum)
Events : 0.6
Number Major Minor RaidDevice State
0 8 17 0 active sync /dev/sdb1
1 8 33 1 active sync /dev/sdc1

3) df -m says:

Filesystem 1M-blocks Used Available Use% Mounted on
/dev/sda2 12297 4207 7471 37% /var
run 886 1 886 1% /var/run

varlock 886 0 886 0% /var/lock
udev 886 1 886 1% /dev
devshm 886 1 886 1% /dev/shm
lrm 886 39 848 5% /lib/modules/2.6.24-19-generic/volatile
/dev/sda3 67179 41344 22449 65% /home
/dev/sdd1 151423 108930 40967 73% /media/160GB
gvfs-fuse-daemon 12297 4207 7471 37% /home/thad/.gvfs
/dev/md0 236594 125133 109076 54% /media/raid1

4) And, finally, I noticed on boot I get the following message:

Mounting local Filesystems
EXT3-fs: unable to read superblock
mount: wrong fs type, bad option, bad superblock on /dev/md0
missing codepage or helper program, or other error

Thanks again for your help. Looking forward to hearing back.

Regards, Thad

December 1st, 2008, 02:43 AM
I don't know what it could be but your df file is sure different from mind, no tmpfs item:

frank@sunshine:~$ df -m
Filesystem 1M-blocks Used Available Use% Mounted on
/dev/sdb1 98552 17995 75551 20% /
tmpfs 1980 0 1980 0% /lib/init/rw
varrun 1980 1 1980 1% /var/run
varlock 1980 1 1980 1% /var/lock
udev 1980 3 1977 1% /dev
tmpfs 1980 1 1980 1% /dev/shm
lrm 1980 3 1978 1% /lib/modules/2.6.27-9-generic/volatile
/dev/md32 908514 29494 833233 4% /home/raid
/dev/sdb2 98619 58871 34778 63% /media/backup
/dev/scd1 4299 4299 0 100% /media/cdrom0
/dev/sda1 33542 18961 12891 60% /media/disk

You have this mounted, which I've not seen before:

gvfs-fuse-daemon 12297 4207 7471 37% /home/thad/.gvfs

You have a "normal" install of 8.04 or 8.10? Maybe 7.10?

All else seems okay. We know the array is good, it is just not auto mounting. It could be a "race" issue... you might wish to add a line in the initramfs-tools/init file:

/usr/share/initramfs-tools/init # to stop a race condition with md

157 maybe_break mount
158 sleep 5
159 log_begin_msg "Mounting root file system..."

sudo /usr/sbin/update-initramfs -uk all # run to rebuild the image

Make sure if you add it it is to the "mount" line. The line numbers are likely different with your system. You might start with 10 instead of 5, seconds delay.

Let us know how you are doing.

December 1st, 2008, 02:04 PM
That's it! I can't thank you enough...do you know how long I have been trying to fix that problem? Whew.

Couple of thoughts:
1) I did a normal install of v8.04, but my root partition seems strange. Instead of a 12 GB root, I ended up with a 12 GB "gvfs-fuse-daemon". Not sure how that happened or if it created the problem in this thread.

2) The entire boot process is a mystery to me. I have never worked with /usr/share/initramfs-tools/init or /usr/sbin/update-initramfs before. Any thoughts on some good reading materials to get me up to speed?

Thanks again for your help!


December 1st, 2008, 04:10 PM
Great, now tell just what was it that did it?

The man pages for tools are helpful:



Usually we find info on any subject using google search. <smile>

December 1st, 2008, 06:24 PM
Sorry...adding the code "sleep 10" right before my drives are mounted in the startup script is what did it.

Thanks again for all your help!


December 30th, 2008, 02:11 AM
I too have the same problem, that is my raid does not mount at boot. I have entered in sleep 10 into the init file and then updated it. On the first reboot the system did not mount the raid, however on the second reboot it did... Maybe because I'm using two 1TB hdd raided together I should increase the sleep to 15?
I should also note that the OS resides on its own 10GB hdd.

December 30th, 2008, 03:26 AM
Sorry...adding the code "sleep 10" right before my drives are mounted in the startup script is what did it.

Thanks again for all your help!


Great... we learn something new everyday, eh?

December 30th, 2008, 03:46 AM
The conditions for autodetection of the array are:

partitions must be marked "Linux raid autodetect" in fdisk
drivers for appropriate RAID level must be compiled into the kernel or available via other-than-filesystem means

Mounting the array that was started up is an entirely separate matter and is done normally via fstab.

Be warned that I lately encountered an obscure bug around this in that if you do a diskless book on a machine that has drives with raid autodetect partitions on it, and the kernel has the raid driver available, then the whole thing blew up by the diskless book not finding it's (NFS) root filesystem anymore. Obviously the RAID autodetection code prepared itself for being the root filesystem and wiped out the original root parameter to the kernel.

I never had to put in sleep(1) statements, but I have the drivers compiled into the kernel, so I won't have race conditions with loading the modules from initramfs.

December 30th, 2008, 04:55 PM
Driver modules are needed for hardware raid... with modern kernels md is alrady there and works well with mdadm.

I'm not sure why the warning.

December 30th, 2008, 06:21 PM
Driver modules are needed for hardware raid... with modern kernels md is alrady there and works well with mdadm.

Not when using autodetect. Autodetect happens before the filesystem containing /lib/modules is mounted. Hence you must make the modules for the appropriate RAID level available to the kernel before filesystems are mounted. You can do that either by compiling them in or by using initramfs.

December 31st, 2008, 06:47 PM
I can't say of sure but when did this stop being necessay... I remember something like 7.10 not requiring driver modules for various raid levels, as they all were built-into the kernel. I could be wrong.