No Volumes No Shares - Superblock Issue?

[NigelBrownless]

When prompted, how many Blocks should I enter for the Scan to read?
The Scan finishes quickly when using Large numbers?

In fact the following stops straight away, I guess the to=0?

Code: Select all

 ./scanmd3 /dev/sda2
Read 1 M blocks to read: 512000
to=0


Should I scan the Assembled md4 or one of the disks sda2 or sdb2?

[fvdw]

Ah think I need to adapt the c program to be able to use 32bits integers, sorry for that

In the meantime you could make two files to check super4097 and 8193

Code: Select all

dd if=/dev/md4 bs=1M skip=4096 count=2 of=4097.img
dd if=/dev/md4 bs=1M skip=8192 count=2 of=8193.img

Then we can see if those are superblocks

[NigelBrownless]

4097 & 8193 attached.

[fvdw]

I will look at those
Attached a new version of scanmd3 allowing bigger integers

[NigelBrownless]

New C Scanner with Larger Integer support started at 15:40 UK.

[fvdw]

super4097 and 8193 contain both a superblock but the content is equal to the primary superblock stored in block 1 :scratch
If a superblock is present at start of each AG then the AG size as mentioned in the superblock and number of superblocks is wrong
The superblock seem to be present in start sector of each 4096 MB block then AG size seems to be 4096 MB and so the file system should contain 465 AG's
The AG should then have size of 1.048.576 (0x100000) blocks of 4096 bytes and not 121.968.988 blocks (0x0745195c) as specified in the superblock which is 476.441,359375 MB

Lets see if the c scanner finds a superblock block every 4096 MB

[fvdw]

this is what xfs_repair reported as you psted

Code: Select all

Phase 1 - find and verify superblock...

fatal error -- couldn't allocate block map, size = 121968988

Seems we are getting somewhere. This confirms that AG size in superblock is not correct.

[NigelBrownless]

Attached is the Output.txt file following the completion of the Scan.
It seemed to slow up over time, but much quicker than the previous Script version.

[fvdw]

mmm...you scand 512MB so you would expect 121 superblocks, only 100 entries found.
Furthermore not the output as expected, the bytes number do not match with those from the scan based on dd. :scratch
for the first two backup superblocks You would expect first backup superblock at byte 4.294.967.297 and the second at byte 8.589.934.593
Also it has an entry found at byte 499584974848 (which is the number of bytes of the AG size as mentioned in the primary superblock.
But we scanned that area previously and did find a superblock there. :scratch
Maybe something wrong with the byte count in the C program need to review that.
Lets check one entry

Code: Select all

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

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

Hope you still want to continue with this, for me it is a learning excersize to increase my XFS knowledge

[NigelBrownless]

Hope you still want to continue with this, for me it is a learning excersize to increase my XFS knowledge

Yes, absolutely. I am also learning but also still hold on to the hope that we may get some data back.

Apologies for the recent slow feedback.....Been extra busy at Work. Should be able to generate those Outputs you've requested tomorrow (Wednesday).

Once again, thanks for your assistance.