Re: Problems with software RAID on SATA
At 19:43 19/08/2005, you wrote:
At 17:52 19/08/2005, you wrote:
At 15:44 19/08/2005, you wrote:
Quoting Stephen Tait <email@example.com>:
As you can see, the only mention of md2 is the "md: md2 stopped" line,
whereas of course I'd be expecting a "raid1: raid set md2 active with 2
out of 2 mirrors" message. Does anyone more au fait with kernel
software RAID know why the kernel won't even attempt to start md2?
Just for fun, have you tried re-partitioning your disks to create a new
It looks like your were using the entire disks of sda and sdb?
Try making a raid 1 from something like sda2 and sdb2. Ignore the first
partition. You may have to zero the superblock prior.
I think your system is fine, if you can manually create an array, mount
it and start using it. If its not happening on boot, then there's
something little thats not making it initialize.
But who knows for sure.
Hope you get it working.
Interesting you should mention the superblock, since the reason md2
wasn't created during the install process was due to some error message I
got about the superblock that wanted me to reboot in order to re-read the
partition table, although I can't remember it exactly. And yes, sd[a|b]1
is a primary partition with the whole disc on it. I'll try chopping up
the drives a bit more and give it another whirl... ho hum, only another
220mins to go until it re-syncs... again! Yeah, I've turned sd[a|b]1 into
an as-yet unformatted 1024MB partition with the RAID array on sd[a|b]2
that comprises the rest of the drive.
Oh how I wish for another 3ware... :D
I've also just built a customised kernel with all of the relevant ATA and
SCSI options compiled in rather than as modules; if either of these two
approaches works I'll post details back to the list, if they don't work
I'll cry :D
Thanks again for your help Mike!
Slight update; with the custom kernel the formatted RAID array will
survive a reboot, but only if I remove it's entry from fstab to stop it
from being fscked at boot (even putting it in fstab with the noauto option
led to a fsck hang), so I've temporarily hacked a script that'll mount it
after all the other services have started and that seems to work. I'm
going to try formatting it as JFS or something just to see if this is a
problem specific to ext2/3. More hair-tearing as and when!
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org with a
subject of "unsubscribe". Trouble? Contact firstname.lastname@example.org
When the partition is formatted JFS (built directly into the kernel),
everything seems to work fine - the RAID array survives the reboot when
entered into fstab, fscks fine, everything. So as far as I can tell there's
an issue with software RAID and ext3 volumes - is it worth opening some
sort of a bug report on this? I haven't seen this on any other systems I
have that use software RAID, but then I usually stick to JFS (or Reiser at
a pinch) for all my own machines.