I recently started playing with software RAID on my 2.2.13 system. After
ditching a faulty scsi disk things were going swimmingly. However, in
between some kernel compiles, I've managed to break the software raid
system...
I'm running 2.2.13 with devfs (although the problem also manifests itself
if I disable devfs with the "devfs=nomount" kernel option).
The symptom is that attempts to use the md devices (e.g., via
raidstart) result in the error:
/dev/md/0: Invalid argument
The raidtab is fairly simple:
raiddev /dev/md/0
raid-level 0
nr-raid-disks 2
persistent-superblock 1
chunk-size 4
device /dev/scsi/host1/bus0/target2/lun0/part1
raid-disk 0
device /dev/scsi/host1/bus0/target3/lun0/part1
raid-disk 1
I've checked all the obvious reaons for this problem:
# cat /proc/mdstat
Personalities : [2 raid0]
read_ahead not set
md0 : inactive
md1 : inactive
md2 : inactive
md3 : inactive
# lsmod
Module Size Used by
[...]
raid0 1760 0 (unused)
Running mkraid yields:
DESTROYING the contents of /dev/md/0 in 5 seconds, Ctrl-C if unsure!
handling MD device /dev/md/0
analyzing super-block
disk 0: /dev/scsi/host1/bus0/target2/lun0/part1, 4441941kB, raid superblock at
4441856kB
disk 1: /dev/scsi/host1/bus0/target3/lun0/part1, 4441941kB, raid superblock at
4441856kB
mkraid: aborted
I've even tried compiling the raid0 modules into the kernel, rather than as
a module, but with the same results.
Am I missing something *really* obvious here? Or is there a way to extract
a more useful error message out of raidstart? For what it's worth, an
strace shows raidstart failing on an ioctl:
open("/dev/md/0", O_RDWR) = 4
ioctl(4, 0x931, 0x811) = -1 EINVAL (Invalid argument)
Thanks for your help,
-- Lars
--
Lars Kellogg-Stedman <[EMAIL PROTECTED]> --> http://www.larsshack.org/
