Re: 2 Disks Jumped Out While Reshaping RAID5

[Neil Brown <neilb@suse.de>]

On Sunday September 6, majedb@gmail.com wrote:
> I forgot to mention that I'm running mdadm 2.6.7.1 (latest in Ubuntu
> Server repositories).

You will need at least 2.6.8 to be able to assemble arrays which are
in the middle of a reshape.  I would suggest 2.6.9 or 3.0.

> 
> I tried forcing the assembly, but as mentioned, I just got an error:
> root@Adam:/var/www# mdadm -Af /dev/md0
> mdadm: superblock on /dev/sdh1 doesn't match others - assembly aborted

This incorrect message is fixed in 2.6.8 an later.

> 
> I know I could've pasted whatever I wrote here, but it seemed
> redundant. I'll keep your hint in mind for the next time, if any
> (hopefully not).

Redundant?  How is that relevant?
If you want help, your goal should be to make it as easy as possible
for people to help you.  Having all information in one email message
is easy.  Having to use a browser to get some of it makes it hard.
Having to register on the website to down load an attachment makes it
nearly impossible.

> 
> This may be of an interest to you:
> root@Adam:/var/www# mdadm -E /dev/sd[a-h]1 | grep Reshap
> sda1  Reshape pos'n : 3357066496 (3201.55 GiB 3437.64 GB)
> sdb1  Reshape pos'n : 3357066496 (3201.55 GiB 3437.64 GB)
> sdc1  Reshape pos'n : 3357066496 (3201.55 GiB 3437.64 GB)
> sdd1  Reshape pos'n : 3357066496 (3201.55 GiB 3437.64 GB)
> sde1  Reshape pos'n : 3357066496 (3201.55 GiB 3437.64 GB)
> sdf1   Reshape pos'n : 3357066496 (3201.55 GiB 3437.64 GB)
> sdg1  Reshape pos'n : 2554257664 (2435.93 GiB 2615.56 GB)
> sdh1  Reshape pos'n : 3357066496 (3201.55 GiB 3437.64 GB)

If you can post that again but without the "grep" I might be able to
be more helpful.  (i.e. the complete output of "mdadm -E /dev/sd[a-h]1").

NeilBrown



> 
> Note that sdd1 was the spare.
> 
> The UUIDs are all the same and the superblock is all similar except
> for the reshaping position of sdg1.
> 
> I didn't try to recreate the array as I've never faced this issue
> before, so I don't know what kind of repercussions it may have.
> 
> What I do know, that at the worst case scenario, I can recreate the
> array out of 7 disks (all but sdg1), but lose about 2.3TB worth of
> data :(
> 
> On Sun, Sep 6, 2009 at 12:32 AM, NeilBrown<neilb@suse.de> wrote:
> > On Sun, September 6, 2009 6:22 am, Majed B. wrote:
> >> Hello all,
> >>
> >> I have posted my problem already here:
> >> http://ubuntuforums.org/showthread.php?p=7900571#post7900571
> >> It also has file attachments of the output of mdadm -E /dev/sd[a-h]1
> >
> > It seems that you need to log in to read the attachements... so I haven't.
> >
> >>
> >> I appreciate any help on this.
> >
> > Hopefully you just need to add "--force" to the assemble command
> > and it would all just work.  However I haven't tested that on an array
> > that is in the process of a reshape so I cannot promise.
> > I might try to reproduce your situation and with some scratch drives
> > and check that mdadm -Af does the right thing, but it won't be a day
> > or so, and as I cannot see the --examine output I might get the
> > situation a bit wrong ... (hint hint: it is always best to post
> > full information rather than pointers to it, unless said information is
> > really really big).
> >
> > NeilBrown
> 
> -- 
>        Majed B.
> --
> 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
--
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