* linear raid space utilization
@ 2006-01-21 7:12 Andy Gajetzki
2006-01-21 8:48 ` David Greaves
2006-01-21 10:39 ` Neil Brown
0 siblings, 2 replies; 4+ messages in thread
From: Andy Gajetzki @ 2006-01-21 7:12 UTC (permalink / raw)
To: linux-raid
Hi there, I recently had a disk go bad in a linear RAID built with
mdadm. The particular disk that failed was the last device of the
RAID. I am curious about how devices are utilized in a linear RAID.
Would the md be filled sequentially from device 1 upto 5? In other
words, what are the odds that the last disk was "empty" when it went
south?
Thanks for your time,
Andy
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: linear raid space utilization
2006-01-21 7:12 linear raid space utilization Andy Gajetzki
@ 2006-01-21 8:48 ` David Greaves
2006-01-21 15:31 ` Andy Gajetzki
2006-01-21 10:39 ` Neil Brown
1 sibling, 1 reply; 4+ messages in thread
From: David Greaves @ 2006-01-21 8:48 UTC (permalink / raw)
To: Andy Gajetzki; +Cc: linux-raid
Andy Gajetzki wrote:
>Hi there, I recently had a disk go bad in a linear RAID built with
>mdadm. The particular disk that failed was the last device of the
>RAID. I am curious about how devices are utilized in a linear RAID.
>Would the md be filled sequentially from device 1 upto 5? In other
>words, what are the odds that the last disk was "empty" when it went
>south?
>
>Thanks for your time,
>
>
Linear raid does provide sequential blocks on disks 1...n
So as long as the filesystem you used was something like 'dd
of=/dev/md0' then you'd be OK <grin>
However if you made a 'proper' filesystem on it then the disk usage
would vary depending on whether it was ext2,ext3,xfs,jfs etc etc
If the bad disk is totally gone then you're at the mercy of fs deign and
fsck - otherwise try and use ddrescue to get an image of it (and the
other disks). Make a 'safe' image and a 'working' image of the bad disk
before attempting repair on 'working' images.
(Of course you need lots of disk space so you may need new disks -
depends how valuable your data is)
HTH
David
--
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: linear raid space utilization
2006-01-21 8:48 ` David Greaves
@ 2006-01-21 15:31 ` Andy Gajetzki
0 siblings, 0 replies; 4+ messages in thread
From: Andy Gajetzki @ 2006-01-21 15:31 UTC (permalink / raw)
To: David Greaves; +Cc: linux-raid, neilb
Thanks for the help. I used ddrescue to grab /dev/hdb and am now using
e2retrieve. I am somewhat optimistic that there will be at least
partial recovery ;). Have a great remainder of the weekend,
Andy Gajetzki
On 1/21/06, David Greaves <david@dgreaves.com> wrote:
> Andy Gajetzki wrote:
>
> >Hi there, I recently had a disk go bad in a linear RAID built with
> >mdadm. The particular disk that failed was the last device of the
> >RAID. I am curious about how devices are utilized in a linear RAID.
> >Would the md be filled sequentially from device 1 upto 5? In other
> >words, what are the odds that the last disk was "empty" when it went
> >south?
> >
> >Thanks for your time,
> >
> >
> Linear raid does provide sequential blocks on disks 1...n
>
> So as long as the filesystem you used was something like 'dd
> of=/dev/md0' then you'd be OK <grin>
>
> However if you made a 'proper' filesystem on it then the disk usage
> would vary depending on whether it was ext2,ext3,xfs,jfs etc etc
>
> If the bad disk is totally gone then you're at the mercy of fs deign and
> fsck - otherwise try and use ddrescue to get an image of it (and the
> other disks). Make a 'safe' image and a 'working' image of the bad disk
> before attempting repair on 'working' images.
> (Of course you need lots of disk space so you may need new disks -
> depends how valuable your data is)
>
> HTH
>
> David
>
> --
>
>
--
Andrew Gajetzki | (403) 330-6552 | Fax: (780) 675-9402 | www.gajetzki.com
---------------------------------------------------------|------------------
XML Architect - Internet Scientific Publications LLC. | www.ispub.com
Editor in Chief - Lethbridge Undergrad. Research Journal | www.lurj.org
Lead Developer - phirOSS Open Source Software | www.phiross.org
Network Analyst - GSchools Inc. | www.gschools.com
---------------------------------------------------------|------------------
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: linear raid space utilization
2006-01-21 7:12 linear raid space utilization Andy Gajetzki
2006-01-21 8:48 ` David Greaves
@ 2006-01-21 10:39 ` Neil Brown
1 sibling, 0 replies; 4+ messages in thread
From: Neil Brown @ 2006-01-21 10:39 UTC (permalink / raw)
To: Andy Gajetzki; +Cc: linux-raid
On Saturday January 21, andy.gajetzki@gmail.com wrote:
> Hi there, I recently had a disk go bad in a linear RAID built with
> mdadm. The particular disk that failed was the last device of the
> RAID. I am curious about how devices are utilized in a linear RAID.
> Would the md be filled sequentially from device 1 upto 5? In other
> words, what are the odds that the last disk was "empty" when it went
> south?
How the space is used depends on the filesystem you had created on the
device.
Many filesystems tend to spread data out so there is plenty of room to
insert new data 'close' to related data.
So if your filesystem was more than about 10% full, there if a very
good chance that a significant fraction was in that last device, and
even if it was less full than that, their is at least a fair chance
that you have lost a reasonable amount of data.
NeilBrown
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2006-01-21 15:31 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-01-21 7:12 linear raid space utilization Andy Gajetzki
2006-01-21 8:48 ` David Greaves
2006-01-21 15:31 ` Andy Gajetzki
2006-01-21 10:39 ` Neil Brown
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).