Thread (16 messages) 16 messages, 4 authors, 2011-04-24

Re: Server down-fail​ed RAID5-asking for some assistance

From: John Valarti <hidden>
Date: 2011-04-24 16:04:55

On Sun, Apr 24, 2011 at 6:29 AM, NeilBrown [off-list ref] wrote:
Ahh, I see it.  This is a bug in there:  ->used isn't set to zero after 'dv'
is allocated.  This was fixed in 3.0.  I don't remember that bug...

I cannot see any easy way to work around that bug.
..
 on the CentOS 5.5 rescue media. I think it's
quoted
time to try something more recent: John, could you try SystemRescueCD
from http://www.sysresccd.org/ and run
   mdadm -Evvs
and if that shows your RAID5 members again,
   mdadm -Afvv /dev/md1
Getting a newer mdadm is definitely a good idea.

Safest to explicitly list the devices that you want
    mdadm -Afvv /dev/md1 /dev/sd[abc]2


NeilBrown
OK, I have Fedora 14 install media handy, so I booted from that.
Once at a shell:
   mdadm --version:   3.12

mdadm -S /dev/md1
mdadm -S /dev/md1 /dev/sd[abc]2
WORKED!

cat/proc/mdstat

Personalities : [raid0] [raid1] [raid6] [raid5] [raid4] [raid10] [linear]
md1 : active raid5 sdc2[0] sdb2[2] sda2[1]
      734925312 blocks level 5, 256k chunk, algorithm 2 [4/3] [UUU_]

unused devices: <none>

Rebooted the system, and it sees my RAID and my OS again.
As I write it is busy running journals and fsck
So far the only dubious part seems to be /tmp. No worry about that.

So, noe the next important part:
What to do next?
Attach another disk bigger than the RAID and copy everything to it?

Assuming yes, then what?
Speculating a bit here:

Add a new good disk and rebuild?
After that, remove the other disk that failed and we just forced back,
and rebuild again?
Then work my way through the other 2 old disks and rebuild 2 more time?

If yes, I could use some command line syntax to make sure I do it the
right way..

If "no" I am all ears as to what to do next.

Oh, and btw:
Thank you
Happy Easter.

--
John V
In a much better mood today.
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help