isw: unsupported map state 0x3 on /dev/sda for RAID_Volume0
Christian Schuhegger
Christian.Schuhegger at gmx.de
Thu Jun 19 19:22:31 UTC 2008
Hello Adam,
Cetnerowski, Adam wrote:
> The problem stems from dmraid not supporting loop devices. You can
> change this behavior by either modifying the device names or the
> function that parses the device names. Either way please take a look at
> interested() in lib/device/scan.c
thank you very much for the hint, but I was not able to make it work. My
initial attempt was to just create a new hard link in /dev for the loop
device:
ln /dev/loop0 /dev/hdi
ln /dev/loop1 /dev/hdj
but I found out that dmraid is scanning the directory /sys/block to find
block devices and in there I cannot create new files/hard links.
After modifying the source of dmraid to accept loop devices it still
does not work, because in the get_device_serial() function dmraid uses
ioctls to get the serial numbers of the scanned devices and the loop
devices do not have a serial number.
I see that get_dm_test_serial() gets the serial number from a file, but
I still have to find out how this works. I've read in the document "
dmraid - device-mapper RAID tool
Supporting ATARAID devices via the generic Linux device-mapper
Heinz Mauelshagen
Red Hat Cluster and Storage Development
"
in section 3.4 Testbed that it should be possible to use this testbed
functionality for what I want, but I am not certain on what exactly I
have to do?
I still cannot understand why even with working directly with dmsetup I
cannot setup a working device map.
I think for the short term I will just create backups of the mapped
devcies, e.g.:
/dev/mapper/isw_bjebjjjbd_RAID_Volume04
But for the future perhaps it would be possible to add the functionality
to dmraid to be able to map disk images like I intended to do, e.g. if
you specify on the commandline the loop device nodes:
dmraid -r /dev/loop0 /dev/loop1
I think it is really serious that it is not possible to take the disk
images in case of a disk failure and work with the disk images in order
to rescue as much data as possible. I've found the article:
http://robmeerman.co.uk/unix/medley?s=sata
"Rescuing a RAID-0 FAKE-RAID Array"
but he worked directly on the disks. I think there is really a shortage
of information about how to rescue data from failing ataraid devices.
Thank you very much for your help,
--
Christian Schuhegger
http://www.el-chef.de/
More information about the Ataraid-list
mailing list