[SATLUG] RAID5 Recovery - Reassembly Questions #3

David Labens adlabens at swbell.net
Wed Aug 26 22:10:04 CDT 2009


If I'm understanding superblock, the superblock on all 4 disks should be the same...

Super Offest : 488391984 sectors
  &
Internal Bitmap : =234 sectors from superblock

Show to be the exact same on all 4 disks (new copy, 2 original high event count, 1 original low event count).

That's what I'm looking to see the same?

In fact, here's the printout from one - the items with "*" in the first column are different - EVERYTHING ELSE is exactly the same on ALL FOUR HDDs:





          Magic :
a92b4efc

        Version : 1.0

    Feature Map : 0x1

     Array UUID :
13782a18:85c82f51:e999ccd0:c2ca0614

           Name : 0

  Creation Time : Sat
Jan 10 13:03:52 2009

     Raid Level :
raid5

   Raid Devices : 4

 

 Avail Dev Size :
488391728 (232.88 GiB 250.06 GB)

     Array Size :
1465175040 (698.65 GiB 750.17 GB)

  Used Dev Size :
488391680 (232.88 GiB 250.06 GB)

   Super Offset :
488391984 sectors

          State :
clean

*    Device UUID :
7cd2101b:57561ce9:d9ba5317:d46c59d3

 

Internal Bitmap : -234 sectors from superblock

*    Update Time : Sat
Aug  8 20:24:05 2009

*       Checksum :
d16f35eb - correct

*         Events : 2986

 

         Layout :
left-asymmetric

     Chunk Size : 128K

 

*    Array Slot : 1 (0,
1, failed, failed, failed)

*   Array State
: uU__ 3 failed


I presume this means everything matches properly and liftoff is "GO!"
???

David Labens

San Antonio, TX

--- On Wed, 8/26/09, David Labens <adlabens at swbell.net> wrote:

From: David Labens <adlabens at swbell.net>
Subject: Re: [SATLUG] RAID5 Recovery - Reassembly Questions #2
To: "The San Antonio Linux User's Group Mailing List" <satlug at satlug.org>
Date: Wednesday, August 26, 2009, 10:00 PM

Sam, 



I've got your original updated one printed out & on the desk in
front of me (it's that important).  I'd rather have a bottle in front
of me than a frontal lobotomy, tho either would probably be more
pleasurable than the process I've endured!

David Labens

San Antonio, TX

--- On Wed, 8/26/09, Samuel Leon <satlug at net153.net> wrote:

From: Samuel Leon <satlug at net153.net>
Subject: Re: [SATLUG] RAID5 Recovery - Reassembly Questions #2
To: "The San Antonio Linux User's Group Mailing List" <satlug at satlug.org>
Date: Wednesday, August 26, 2009, 9:51 PM

David Labens wrote:
> Do I need to issue the command in any order?  Or will it figure it out?
> 
> 
> David Labens
> 
No the order of the disks to assemble does not matter.  And make sure everything is connected because once you force assemble you will have to add the drive that it left out.

I attached a copy of the guide I wrote earlier:
Sam

> Here is an updated guide for the next steps:
>
> First make sure that the raid superblocks are the same for both the old
> and new drive:
> mdadm --examine /dev/sda /dev/sdb
>
> Then shutdown, reconnect everything and bring the computer up.  Might
> also make sure that all your disks are linked to the right device names
> as needed (sda,sdb ect)
>
>
> then force the array to reassemble:
> mdadm --assemble --force /dev/md0 /dev/sde1 /dev/sda1 /dev/sdb1 /dev/sdc1
>
>
> (make sure those drive numbers are right, I copied them from an earlier
> post. Also, if you don't get some output saying that the array *has been
> started* then you need to stop right here.  The output you should get
> should be similar to:
> mdadm: forcing event count in /dev/hdd1(2) from 228 upto 232
> mdadm: clearing FAULTY flag for device 1 in /dev/md0 for /dev/hdd1
> mdadm: /dev/md0 has been started with 2 drives (out of 3).
> )
>
> That should bring it up in degraded mode with one failed drive. So then do:
> mdadm -D /dev/md0
>
> and that will list some info and at the bottom you will see what drive
> is still removed.  So add it to the array with:
>
> mdadm --add /dev/md0 /dev/sdX
>
> but if there is still more than one drive missing from the array at this
> point never use mdadm --add because something is wrong and that will
> make it worse.
>
> If added correctly, the array should start to resync.  Let it finish and
>  make sure the filesystem on /dev/md0 stays unmounted.  Once it is done
> resyncing check the filesystem (if it is ext3) with:
> e2fsck -D -f /dev/md0
>
> You will probably get some errors and have to press "y" a hundred times.
> Any files that are lost will be put in your lost+found folder and will
> be given their inode number as the file name.
>
> Then you can mount the filesystem on /dev/md0. Then you should be done.
-- _______________________________________________
SATLUG mailing list
SATLUG at satlug.org
http://alamo.satlug.org/mailman/listinfo/satlug to manage/unsubscribe
Powered by Rackspace (www.rackspace.com)
--
_______________________________________________
SATLUG mailing list
SATLUG at satlug.org
http://alamo.satlug.org/mailman/listinfo/satlug to manage/unsubscribe
Powered by Rackspace (www.rackspace.com)


More information about the SATLUG mailing list