FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Ubuntu > Ubuntu User

 
 
LinkBack Thread Tools
 
Old 11-09-2011, 12:04 PM
Marius Gedminas
 
Default 8.04 md raid5 troubleshooting

On Wed, Nov 09, 2011 at 12:24:51PM +0100, Reinhard Haller wrote:
> md0 : active raid5 sda3[3](F) sdb2[1] sdc2[2]
> 775119872 blocks level 5, 256k chunk, algorithm 2 [3/2] [_UU]
...
> fdisk says about /dev/sda:
>
> Platte /dev/sda: 400.1 GByte, 400088457216 Byte
...
> Gerät boot. Anfang Ende Blöcke Id System
> /dev/sda1 * 1 25 200781 83 Linux
> /dev/sda2 26 392 2947927+ 82 Linux Swap / Solaris
> /dev/sda3 393 48641 387560092+ fd Linux raid autodetect
...
>
> The only shortly available spare device is a 1TB SATA drive.
>
> smartctl on /dev/sda reports a healthy drive and /dev/sda1 is still available

Did you run a self-test? smartctl -t long /dev/sda and then (a couple
of hours later, it'll tell you the estimated test time) check the
results with I forget which exactly option, but smartctl -a /dev/sda
works.

> What is the suggested way to get the array back working?

Connect the spare drive, partition it, use mdadm to hot-add it, wait
while it syncs in the background.

> Does it make sense to "revert" the faulty state of /dev/sda?

I doubt it -- if the disk failed once, it'll likely fail again.

> How to partition the 1 TB drive to be able to grow the array after
> replacing all old drives with 1TB ones?

I think you'd have to replace all the current partitions that comprise
your RAID array with larger ones, one by one (waiting for full syncs),
and then use mdadm --grow to enlarge the array (and wait for a sync),
and then use resize2fs (assuming ext3 fs) to enlarge the partition.

Up-to-date backups are recommended, as per usual.

Marius Gedminas
--
Any sufficiently advanced Operating System is indistinguishable from Linux.
-- Jim Dennis
--
ubuntu-users mailing list
ubuntu-users@lists.ubuntu.com
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-users
 
Old 11-09-2011, 03:40 PM
Reinhard Haller
 
Default 8.04 md raid5 troubleshooting

Am 09.11.2011 14:04, schrieb Marius Gedminas:

What is the suggested way to get the array back working?

Connect the spare drive, partition it, use mdadm to hot-add it, wait
while it syncs in the background.


Does it make sense to create a RAID-1 with the spare drive and the
"defective" drive to clone the boot partition (assuming the defective
drive is responsive)?


Reinhard

--
ubuntu-users mailing list
ubuntu-users@lists.ubuntu.com
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-users
 
Old 11-09-2011, 06:15 PM
Marius Gedminas
 
Default 8.04 md raid5 troubleshooting

On Wed, Nov 09, 2011 at 05:40:37PM +0100, Reinhard Haller wrote:
> Am 09.11.2011 14:04, schrieb Marius Gedminas:
> >>What is the suggested way to get the array back working?
> >Connect the spare drive, partition it, use mdadm to hot-add it, wait
> >while it syncs in the background.
>
> Does it make sense to create a RAID-1 with the spare drive and the
> "defective" drive to clone the boot partition (assuming the
> defective drive is responsive)?

Most disk failures I've seen are localized: you can't read the data from
one particular bit. It so happened that the bit that couldn't be read
appeared in the middle of your RAID-5 partition (statistically likely,
since RAID takes most up of the disk's sectors). So, no, I wouldn't
expect any issues accessing the old boot partition.

RAID-1 sounds like a good idea. I'm not exactly sure how you can
convert an existing standalone partition into a RAID-1 array; you
probably need to shrink it first a bit to leave space for the RAID
metadata, which, I think, lives at the end?

Worst case, create a new RAID-1 on the new disk and clone your /boot
contents with tar, update the UUID in /etc/fstab and reinstall GRUB. I
don't think the /boot UUID is written into the initramfs so at least you
wouldn't need to regenerate those. I think.

Marius Gedminas
--
Voodoo Programming: Things programmers do that they know shouldn't work but
they try anyway, and which sometimes actually work, such as recompiling
everything.
-- Karl Lehenbauer
--
ubuntu-users mailing list
ubuntu-users@lists.ubuntu.com
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-users
 
Old 11-10-2011, 07:57 AM
Reinhard Haller
 
Default 8.04 md raid5 troubleshooting

Am 09.11.2011 20:15, schrieb Marius Gedminas:

On Wed, Nov 09, 2011 at 05:40:37PM +0100, Reinhard Haller wrote:


Does it make sense to create a RAID-1 with the spare drive and the
"defective" drive to clone the boot partition (assuming the
defective drive is responsive)?

Most disk failures I've seen are localized: you can't read the data from
one particular bit. It so happened that the bit that couldn't be read
appeared in the middle of your RAID-5 partition (statistically likely,
since RAID takes most up of the disk's sectors). So, no, I wouldn't
expect any issues accessing the old boot partition.
In the meantime the extended selftest finished without problems (the
only significant difference between the defective drive and the others
is a reallocated sector count of 1).


RAID-1 sounds like a good idea. I'm not exactly sure how you can
convert an existing standalone partition into a RAID-1 array; you
probably need to shrink it first a bit to leave space for the RAID
metadata, which, I think, lives at the end?

Hope to find an Howto

Thanks
Reinhard


--
ubuntu-users mailing list
ubuntu-users@lists.ubuntu.com
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-users
 
Old 11-10-2011, 12:26 PM
"compdoc"
 
Default 8.04 md raid5 troubleshooting

>In the meantime the extended selftest finished without problems (the
>only significant difference between the defective drive and the others
>is a reallocated sector count of 1).


Reallocated sectors are why drives drop out of arrays. I've had that problem
with the consumer desktop drives I use in my arrays. I think enterprise
class drives have a feature that overcomes that problem.

Once a drive starts to develop bad sectors it can't be trusted in an array
because if it drops out again, and then another drive drops out for any
reason, you've lost the array. At least that's true with simple arrays up to
and including raid 5.

Replace the drive with a new one - don't use refurbished drives. Once you
have a set of trusted/proven drives, the array can run for years.





--
ubuntu-users mailing list
ubuntu-users@lists.ubuntu.com
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-users
 

Thread Tools




All times are GMT. The time now is 02:51 PM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright ©2007 - 2008, www.linux-archive.org