No Volumes No Shares - Superblock Issue?

[NigelBrownless]

Code: Select all

dd if=/dev/md4 bs=1 skip=17071982430 count=512 of=17071982430.img

Gives an Error. Even correcting it to:

Code: Select all

dd if=/dev/md4 bs=1M skip=17071982430 count=512 of=17071982430.img

It comes back with....

Code: Select all

dd: invalid number '17071982430'

[NigelBrownless]

In the meantime to see if we have superblocks at each 4096 MB we can do so dd copies

Code: Select all

dd if=/dev/md4 bs=1M skip=40960 count=1 of=40961.img
dd if=/dev/md4 bs=1M skip=81920 count=1 of=81921.img
dd if=/dev/md4 bs=1M skip=409600 count=1 of=409601.img


Outputs in the attached file...

[fvdw]

Code: Select all

dd if=/dev/md4 bs=1 skip=17071982430 count=512 of=17071982430.img

Gives an Error. Even correcting it to:

Code: Select all

dd if=/dev/md4 bs=1M skip=17071982430 count=512 of=17071982430.img

It comes back with....

Code: Select all

dd: invalid number '17071982430'

The first probably fails because the integer value of skip is too large for dd command
The second is invalid as you try skip 17071982430 blocks of 1 MB which far more the size of the array.

Anyhow lets leave that for what is for the moment, the deviation but also slow down of the scan might be caused by problems with reading the disk as result of bad sectors. Lets check the blocks you copied to see if superblocks are present

[fvdw]

I looked at the three files in SB4096.rar
All three contain the superblock with exactly same content as the primairy superblock.
So indeed it seems we have an AG size of 4096 MB (4GB)
We copied only 1 MB at each position and the content of the three files are exactly the same for this 1 MB. That makes me a little surprised.
Now XFS stores besides the superblock also inode and log data, maybe it uses some space on each AG but I cannot find exact info on that.
At a certain point you would expect differences on locations were data (files )are stored.

I would like check some larger parts and compare those before adapting the primairy superblock.

Code: Select all

dd if=/dev/md4 bs=1M skip=40960 count=20 of=40961-20.img
dd if=/dev/md4 bs=1M skip=409600 count=20 of=409601-20.img

[fvdw]

The idea is to adapt the primairy superblock to AG size of 4096 MB and number of AG's to 465

Lets also check that we have a superblock at beginning of AG 465

Code: Select all

dd if=/dev/md4 bs=1M skip=1900544 count=1 of=ag465.img

[NigelBrownless]

Code: Select all

dd if=/dev/md4 bs=1M skip=40960 count=20 of=40961-20.img
dd if=/dev/md4 bs=1M skip=409600 count=20 of=409601-20.img

File Attached....

[NigelBrownless]

Code: Select all

dd if=/dev/md4 bs=1M skip=1900544 count=1 of=ag465.img

File Attached...

[fvdw]

The 20MB larger files have exactly same content. Apparently inodes and journal log takes more place.

ag465.img contains a super block.

So lets try to adapt the primairy superblock. Attached an adapted primairy super block with AG size of 4096 MB and number of AG's to 465

To write to the array

Code: Select all

dd if=md4-512-new.img bs=1 count=512 of=/dev/md4

Then try to run

Code: Select all

mdadm --verbose --assemble --force /dev/md4 /dev/sda2 /dev/sdb2
mkdir /mountpoint
mount -o ro /dev/md4 /mountpoint

and see what happens

[NigelBrownless]

Apologies for delayed response, been away from computer for weekend!
I uploaded the file and tried to run...

Code: Select all

dd if=md4-512-new.img bs=1 count=512 of=/dev/md4

and got the following output...

Code: Select all

dd: writing '/dev/md4': No space left on device
1+0 records in
0+0 records out
0 bytes (0B) copied, 0.001030 seconds, 0B/s

So then I thought I need to Assemble the array first, so I did...

Code: Select all

mdadm --verbose --assemble --force /dev/md4 /dev/sda2 /dev/sdb2

which was successful...

Code: Select all

mdadm: looking for devices for /dev/md4
mdadm: /dev/sda2 is identified as a member of /dev/md4, slot 0.
mdadm: /dev/sdb2 is identified as a member of /dev/md4, slot 1.
mdadm: added /dev/sdb2 to /dev/md4 as 1
mdadm: added /dev/sda2 to /dev/md4 as 0
mdadm: /dev/md4 has been started with 2 drives.

So I then created a mount directory, Successfully and then tried the DD again...

Code: Select all

dd if=md4-512-new.img bs=1 count=512 of=/dev/md4

and got the following output...

Code: Select all

512+0 records in
512+0 records out
512 bytes (512B) copied, 0.188305 seconds, 2.7KB/s


Now ready to try the Mount again...

Code: Select all

mount -o ro /dev/md4 /mountpoint

And got the following output...

Code: Select all

mount: Structure needs cleaning

I think that's progress?

[fvdw]

yes, I think you should now run xfs_repair.
For that the array must be assembled but must be kept unmounted.
It is advised to do first a check to see what xfs_repair will do by using the -n option, then no changes will be made to the filesystem only a specification what xfs_repair will do.

Code: Select all

xfs_repair -n /dev/md4

After that we judge if it is ok to run xfs_repair and let it make the changes