[olug] Software RAID / accidentally raidhotadd to wrong md

lug at robotz.com lug at robotz.com
Fri Apr 2 21:05:36 UTC 2004


Yes they did sync, unfortunately.

I found a way to correct the superblock entries.  It was risky and 
what I ended up with I has mixed feelings about.

The ‘raidreconf’ utility will work without the raid md4 running. 
Using that I was able to (over a period of 14 hours) correct the 
superblock entries.  Afterwards, everything that belonged in md4 
realized their membership in md4.  

However, there were many file system errors afterwards.  I had to 
use fsck manually to correct them, a process that took a substantial 
amount of my time at the keyboard.   During the process I felt somewhat 
hopeful as I could actually see filenames I recognized come up.

After going through md4 with fsck twice and correcting an entry in 
/etc/fstab I was able to mount “home” again for the first time since 
Sunday when I broke it.  Somehow ‘raidreconf’ converted md4 from 
ext3 to ext2.  That was the entry I had to correct in /etc/fstab 
to get md4 to mount to /home.

Ok, here is my new most undesirable dilemma.  There is 130 GB of 
data, which might be everything as that number sounds about right.
But it is all sitting in /home/lost+found with number sequences 
as every filename and every directory name?!

Sample:
drwxr-xr-x    2 adam     1005         4096 Mar 19  2001 #10371116/
drwxr-xr-x    2 ractive  ractive      4096 Mar  9 20:27 #10420285/
drwx--x--x   23 lazygirl users        4096 Oct  5 16:34 #11534342/
drwxrwx---    2 root     secure       4096 Jan  3 00:06 #163841/
---x--x---    1 1750509120 801725972    49152 Sep 13  1938 #180198*
-r-S--x-w-    1 4261505341 2257411334    49152 Dec 13  1901 #180212*
-r--------    1 root     4218490     49152 Nov  7  1905 #180215
b--xr-sr-T    1 27182    29810    110, 120 Dec 23  2024 #195907

Apparently, all of the file name information has been lost.  The 
prospect of manually opening 130 GB of data file by file in a hex 
editor to determine what it ‘was’ is pretty much beyond the scope 
of reasonable or possible time allocation by one human.

I was wondering if there was a way to restore them back to their 
filenames and directory structures?  (Some kind of utility as there 
was with MSDOS and checkdisk files) or something built into Linux 
for this situation.

ty



At Friday, 2 April 2004, you wrote:

>When you added the two partitions to the other raid
>array, did it go through the whole syncing process on
>both the drives?
>
>--- lug at robotz.com wrote:
>> I will type it in here....
>> 
>> /etc/raidtab
>> 
>> raiddev			/dev/md3
>> raid-level		5
>> nr-raid-disks		3
>> chuck-size		64k
>> oersustebt-superblock	1
>> nr-spare-disks		0
>> 
>> device			/dev/hda3
>> raid-disk		0
>> device			/dev/hdb3
>> raid-disk		1
>> device			/dev/hdc3
>> raid-disk		2
>> 
>> raiddev			/dev/md0
>> raid-level		1
>> nr-raid-disks		3
>> chuck-size		64k
>> oersustebt-superblock	1
>> nr-spare-disks		0
>> 
>> device			/dev/hda1
>> raid-disk		0
>> device			/dev/hdb1
>> raid-disk		1
>> device			/dev/hdc1
>> raid-disk		2
>> 
>> raiddev			/dev/md4
>> raid-level		5
>> nr-raid-disks		3
>> chuck-size		64k
>> oersustebt-superblock	1
>> nr-spare-disks		0
>> 
>> device			/dev/hda2
>> raid-disk		0
>> device			/dev/hdb2
>> raid-disk		1
>> device			/dev/hdc2
>> raid-disk		2
>> 
>> raiddev			/dev/md2
>> raid-level		5
>> nr-raid-disks		3
>> chuck-size		64k
>> oersustebt-superblock	1
>> nr-spare-disks		0
>> 
>> device			/dev/hda5
>> raid-disk		0
>> device			/dev/hdb5
>> raid-disk		1
>> device			/dev/hdc5
>> raid-disk		2
>> 
>> raiddev			/dev/md1
>> raid-level		5
>> nr-raid-disks		3
>> chuck-size		64k
>> oersustebt-superblock	1
>> nr-spare-disks		0
>> 
>> device			/dev/hda6
>> raid-disk		0
>> device			/dev/hdb6
>> raid-disk		1
>> device			/dev/hdc6
>> raid-disk		2
>> 
>> 
>> 
>> 
>> At Wednesday, 31 March 2004, you wrote:
>> 
>> >Wow, paste your raidtab if you don't mind....let me
>> >take a peek at that then read your stuff below. 
>> That
>> >will help....
>> >
>> >--- lug at robotz.com wrote:
>> >> Issue summary:  Had problems with 1 of a 3 drive
>> >> software raid array.
>> >> Harddrive is not physically bad, but the array
>> >> failed today on boot.
>> >> Used “raidhotadd” to get all of the raid5
>> partitions
>> >> (md0 – md3) 
>> >> up and running with the exception of the most
>> >> important partition 
>> >> (md4).
>> >> 
>> >> I accidentally tried to “raidhotadd” 2 of the 3
>> >> drives from md4 into 
>> >> md2.  Now hdb2 and hdc2 think they are part of
>> md2
>> >> when they actually 
>> >> should be part of md4.  This is not reflected in
>> >> /etc/raidtab but 
>> >> must be stored somewhere I cannot find.
>> >> (superblock?)
>> >> 
>> >> This leaves only 1 drive in md4 and therefore md4
>> >> will not start 
>> >> even degraded.  md4 is a RAID 5 partition, which
>> is
>> >> supposed to include 
>> >> hda2, hdb2, and hdc2.  Since RAID wont start on
>> md4,
>> >> I cannot simply 
>> >> use “raidhotremove” and “raidhotadd” to correct
>> this
>> >> mistake (assuming 
>> >> that would correct the mistake).
>> >> 
>> >> I used “raidsetfaulty” to deactivate hdb2 to stop
>> >> interfering with 
>> >> the start of md2.  (because hdb2 belongs in md4
>> and
>> >> raidhotremove 
>> >> refuses to get hdb2 out of md2 without
>> deactivating
>> >> the partition)
>> >> 
>> >> lsraid reports md0 – md3 correctly.  However,
>> typing
>> >> “lsraid –A –d 
>> >> /dev/hdb2” or “lsraid –A –d /dev/hdc2” shows them
>> as
>> >> a member of 
>> >> md2 still.  They should be in md4 instead.
>> >> 
>> >> My question #1:  Did my mistake hose the data in
>> >> both hdb2 and hdc2 
>> >> making it impossible to recover md4 (my most
>> >> critical partition containing 
>> >> /home) or is hdb2 and hdc2 merely confused about
>> >> where they belong 
>> >> but still within them is my critical data making
>> >> recovery of md4 
>> >> possible?
>> >> 
>> >> My question #2: If recovery is possible, should I
>> >> then “mkraid /dev/md4 
>> >> –f” and hope for the best or is there a safer way
>> to
>> >> _save_ my data? 
>> >> Or perhaps there is a way to make hdb2 and hdc2
>> >> realize again that 
>> >> they are part of md4.   
>> >> 
>> >> Keep in mind that since md4 won’t start I can’t
>> >> reverse the mistake 
>> >> with raidhotadd.
>> >> 
>> >> If I failed to provide any details please let me
>> >> know and I will 
>> >> elaborate further.
>> >> 
>> >> Thank you.
>> >> 
>> >> 
>> >> 
>> >> 
>> >> 
>> >> 
>> >> 
>> >> _______________________________________________
>> >> OLUG mailing list
>> >> OLUG at olug.org
>> >> http://lists.olug.org/mailman/listinfo/olug
>> >
>> >=====
>> >Terry
>> >
>> >__________________________________
>> >Do you Yahoo!?
>> >Yahoo! Finance Tax Center - File online. File on
>> time.
>> >http://taxes.yahoo.com/filing.html
>> >_______________________________________________
>> >OLUG mailing list
>> >OLUG at olug.org
>> >http://lists.olug.org/mailman/listinfo/olug
>> >
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> _______________________________________________
>> OLUG mailing list
>> OLUG at olug.org
>> http://lists.olug.org/mailman/listinfo/olug
>
>=====
>Terry
>
>__________________________________
>Do you Yahoo!?
>Yahoo! Small Business $15K Web Design Giveaway 
>http://promotions.yahoo.com/design_giveaway/
>_______________________________________________
>OLUG mailing list
>OLUG at olug.org
>http://lists.olug.org/mailman/listinfo/olug
>









More information about the OLUG mailing list