dmraid set won't activate

[Chad Bernier]

Binary package hint: dmraid

I'm not entirely sure if this is a dmraid bug or not. I finally fixed the problems of kernel modules. I have two custom kernels, 2.6.19 with the 2.6.18 dmraid45 patch, and 2.6.20 with the dmraid45 patch provided by Mandriva. I have no reason to beleive that this module is causing a problem. The raid set used to work using an older version of dmraid that i compiled because the edgy version wasn't sufficient. I'm using feisty now.

here's the problem.

root@jp-super:/home/jp# dmraid -ay -v -d
DEBUG: _find_set: searching nvidia_edjdcidj
DEBUG: _find_set: not found nvidia_edjdcidj
DEBUG: _find_set: searching nvidia_edjdcidj
DEBUG: _find_set: not found nvidia_edjdcidj
DEBUG: _find_set: searching nvidia_edjdcidj
DEBUG: _find_set: found nvidia_edjdcidj
DEBUG: _find_set: searching nvidia_edjdcidj
DEBUG: _find_set: found nvidia_edjdcidj
DEBUG: _find_set: searching nvidia_edjdcidj
DEBUG: _find_set: found nvidia_edjdcidj
DEBUG: _find_set: searching nvidia_edjdcidj
DEBUG: _find_set: found nvidia_edjdcidj
DEBUG: _find_set: searching nvidia_edjdcidj
DEBUG: _find_set: found nvidia_edjdcidj
DEBUG: _find_set: searching nvidia_edjdcidj
DEBUG: _find_set: found nvidia_edjdcidj
DEBUG: checking nvidia device "/dev/sda"
DEBUG: checking nvidia device "/dev/sdc"
DEBUG: checking nvidia device "/dev/sdd"
DEBUG: checking nvidia device "/dev/sdb"
DEBUG: set status of set "nvidia_edjdcidj" to 16
RAID set "nvidia_edjdcidj" already active
INFO: Activating raid5_ls RAID set "nvidia_edjdcidj"
ERROR: opening "/dev/.static/dev/mapper/nvidia_edjdcidj"
DEBUG: freeing devices of RAID set "nvidia_edjdcidj"
DEBUG: freeing device "nvidia_edjdcidj", path "/dev/sda"
DEBUG: freeing device "nvidia_edjdcidj", path "/dev/sdc"
DEBUG: freeing device "nvidia_edjdcidj", path "/dev/sdd"
DEBUG: freeing device "nvidia_edjdcidj", path "/dev/sdb"
root@jp-super:/home/jp# ls /dev/.static/dev/m
mapper/ midi00 midi03 midi3 mixer2 mpu401stat
mem midi01 midi1 mixer mixer3
midi0 midi02 midi2 mixer1 mpu401data
root@jp-super:/home/jp# ls /dev/.static/dev/mapper/
nvidia_edjdcidj1 nvidia_edjdcidj2
root@jp-super:/home/jp# ls /dev/mapper/
control nvidia_edjdcidj
root@jp-super:/home/jp#

as you can see, it tries to open a device that doesn't exist, although the partitions on that device exist. Then, in the directory where everything should exist, the device exists, but the partitions don't. Nothing can be mounted.

[Chad Bernier (berniercr)]

Is this related to udev? everything else seems to be working right.

[Tormod Volden (tormodvolden)]

Can you mount /dev/.static/dev/mapper/nvidia_edjdcidj1 ? I am more worried about your custom kernels. How did you make them? Can you try to rather use the standard kernel, but rebuild and replace the needed modules?

[Chad Bernier (berniercr)]

No, that is unmountable. It complains that it isn't a valid block device. I would love to try using the standard kernel with only the needed modules changed. However, that is a bit too advanced for me, so you'll have to provide instructions. If i just copied them over, it would complain about wrong versions, because for who knows what reasons, the version numbers of the source never matches the kernel image.

[Tormod Volden (tormodvolden)]

Can you give a link to that Mandriva patch you are using, or just attach it here?

[Chad Bernier (berniercr)]

certainly. just check out this bug. https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.20/+bug/97655

It has the two patches and a discussion about it.

[Tormod Volden (tormodvolden)]

I recompiled the drivers/md tree with the dmraid45 patch you referred to in bug #97655. Try the normal kernel, but replace all modules in /lib/modules/2.6.20-15-generic/kernel/drivers/md with these. (I am not sure which modules have been changed, so replace them all.) Take a backup of that directory first...

[Chad Bernier (berniercr)]

Something is wrong. I deleted all the modules in that directly after backing them up, and added your new ones. I rebooted into the current kernel as stated. I reboot the computer. It takes a long time, complains about the lack of raid45, io buffer errors, and end of device, as usual.

I manually try to load the dm-raid4-5 module to see if that works before checking to see if its in the initrd.

jp@jp-super:/lib/modules/2.6.20-15-generic/kernel/drivers/md$ modprobe dm-raid4-5
FATAL: Module dm_raid4_5 not found.

running update-modules doesn't help at all. I try manually loading the module directly incase it's just having trouble finding it.

jp@jp-super:/lib/modules/2.6.20-15-generic/kernel/drivers/md$ sudo insmod dm-raid4-5.ko
insmod: error inserting 'dm-raid4-5.ko': -1 Invalid module format

So i can't get the module loaded.

[Chad Bernier (berniercr)]

haha, sorry, tormod, but i failed to mention that I'm using x86_64. I bet you gave me x86 modules. You should have asked though :)

[Tormod Volden (tormodvolden)]

Sorry Chad, I forgot there are other animals than -i386 out there :) See http://ubuntuforums.org/showthread.php?t=279628 for how to build a subset of modules.

[Tormod Volden (tormodvolden)]

Can you give some more information on your hardware, motherboard, raid controller etc?

[Chad Bernier (berniercr)]

I can't help you with this bug anymore. The computer was not mine and had to be returned to its owner. The owner will just have to live without it for now.

It was using nvraid, but you already know that. I don't have the computer with me to look up mobo model number or anything. I did try compiling those modules, but they wouldn't load either.

[Tormod Volden (tormodvolden)]

Ok, I guess we can just close this bug then. Please reopen if you are able to reproduce the problems another time.