All disks in the array suddenly non-fresh??

Hi,

I'm very new to raid so my question is probably very basic.

This machine I have, is set up with three partitions on three different disks (/dev/hda3, /dev/hde1, /dev/hdg1) forming a raid-5 array, /dev/md0.
At some point, the machine became unresponsive, and on a reboot, I get the following:

...
md: md driver 0.90.0 MAX_MD_DEVS=256, MD_SB_DISKS=27
...

md: Autodetecting RAID arrays.
md: autorun ...
md: considering hdg1 ...
md:  adding hdg1 ...
md:  adding hde1 ...
md:  adding hda3 ...
md: created md0
md: bind<hda3>
md: bind<hde1>
md: bind<hdg1>
md: running: <hdg1><hde1><hda3>
md: kicking non-fresh hdg1 from array!
md: unbind<hdg1>
md: export_rdev(hdg1)
md: kicking non-fresh hde1 from array!
md: unbind<hde1>
md: export_rdev(hde1)
md: personality 4 is not loaded!

md :do_md_run() returned -22
md: md0 stopped.
md: unbind<hda3>
md: export_rdev(hda3)
md: ... autorun DONE.

I am using classic (non-mdadm) raid tools. Of course everything like /proc/mdstat, shows I have no md0, nor can I start md0. What would be the next logical thing to investigate, to get out of this situation?

Boris

Hi Boris,

This kind of problem is a result of unclean shutdown. There are 2 problems here...

(1) MD superblock on hdg1 and hde1 are out-of-date.

(2) raid5 module is not loaded. Make sure that raid5 module is loaded.

I recommend that you install and use mdadm to fix this problem.

* Examine md superblock:
mdadm --examine /dev/hda3
to record Layout, Chunk Size, disks order (disk0, disk1, disk2)
* Recreate the raid5 array with 1 "missing" disk:
mdadm -C /dev/md0 -l 5 -c [ChunkSize] -p [Layout] -n 3 [disk0] [disk1]
missing
* mount (readonly) /dev/md0
* verify the filesystem
* if everything looks good:
mdadm --add /dev/md0 [disk2]
* umount and mount /dev/md0

If you cannot boot to a console to do the above, you can boot from Ubuntu or gentoo livecd.

--
Regards,
Mike T.