Linux RAID-5 Problem w/ raidtools-0.90??

Linux RAID-5 Problem w/ raidtools-0.90??

Post by Matthew Coen » Thu, 24 Jun 1999 04:00:00



I have a Dual-P3-450 Box running RedHat 6.0 and kernel version 2.2.9.
The drives I have are a 2.1 Gb UDMA (only used as a Boot / System Drive)

4 IBM 9.1 GB U2W SCSI Drives w/ 2Mb Cache

I have successfully created all of the partitions across all of the 4
drives, and have properly made the /etc/raidtab file

A sample of this, for my /dev/md0 device is....

raiddev /dev/md0
    raid-level            5
    nr-raid-disks      4
    nr-spare-disks   0
    chunk-size          4
    parity algorithm    left-symmetric
device    /dev/sda1
raid-disk    0
device    /dev/sdb1
raid-disk    1
device    /dev/sdc1
raid-disk    2
device    /dev/sdd1
raid-disk    3

This format worked perfectly under kernel version 2.2.5, athough I found the
raid to be a little flaky.

Under the 2.2.9 kernel ver like I mentioned above, when I do a
"mkraid --really-force /dev/md0", it appears to view the drives correctly,
but the returns "mkraid aborted" for no apparent reason.

I have heard that the raid-tools are kernel-raid-patch specific, and that
some of the ioctl numbers have changed throughout the history.. But so far I
have been unable to track down any version that will work with my setup...

If anyone has any information which could help it would be immensely
appreciated, as I am currently pulling my hair out struggling to comprehend
why nothing is being reported to the syslog and mkraid isn't returning and
useable troubleshooting info...

Please help...

Matthew Coene
-Systems Administrator.

 
 
 

Linux RAID-5 Problem w/ raidtools-0.90??

Post by Cokey de Perci » Fri, 25 Jun 1999 04:00:00



> I have a Dual-P3-450 Box running RedHat 6.0 and kernel version 2.2.9.
> The drives I have are a 2.1 Gb UDMA (only used as a Boot / System Drive)

> 4 IBM 9.1 GB U2W SCSI Drives w/ 2Mb Cache

> I have successfully created all of the partitions across all of the 4
> drives, and have properly made the /etc/raidtab file

> A sample of this, for my /dev/md0 device is....

> raiddev /dev/md0
>     raid-level            5
>     nr-raid-disks      4
>     nr-spare-disks   0
>     chunk-size          4
>     parity algorithm    left-symmetric
> device    /dev/sda1
> raid-disk    0
> device    /dev/sdb1
> raid-disk    1
> device    /dev/sdc1
> raid-disk    2
> device    /dev/sdd1
> raid-disk    3

> This format worked perfectly under kernel version 2.2.5, athough I found the
> raid to be a little flaky.

> Under the 2.2.9 kernel ver like I mentioned above, when I do a
> "mkraid --really-force /dev/md0", it appears to view the drives correctly,
> but the returns "mkraid aborted" for no apparent reason.

> I have heard that the raid-tools are kernel-raid-patch specific, and that
> some of the ioctl numbers have changed throughout the history.. But so far I
> have been unable to track down any version that will work with my setup...

> If anyone has any information which could help it would be immensely
> appreciated, as I am currently pulling my hair out struggling to comprehend
> why nothing is being reported to the syslog and mkraid isn't returning and
> useable troubleshooting info...

Yup, there seems to be a problem with software raid starting with
kernel 2.2.6 and up AFAIK.  I've got 2.2.10 at work on a machine
with a working raid0 (under 2.2.5) and it no longer works.  If I
reboot in 2.2.5 everything works just fine.  I'm not sure what the
problem/answer is, but it seems that the raidtools and kernel are
out of sync.... :-(

Best

Cokey

--
------------------------------------------------------------------
Cokey de Percin, DBA            Email: