Question

2

After rescue with `dd` - what to do?

rated 0 times [ 2] [ 0] / answers: 1 / hits: 1249 / 3 Years ago, fri, august 20, 2021, 10:45:59

I have an external that very suddenly conked out on me. It might not be in horrible shape, although I can't be sure.

I just tried to recover it by using ddrescue, straight to a big partition.

sudo dd if=/dev/sdb of=/dev/sda3 bs=4096 conv=notrunc,noerror,sync

It ran for a couple hours and said that 160 bg were copied. I did not make an image, just copied sdb over to sda3.

The disc light was light constantly that whole time, although i didn't hear any noise.

So then after it runs I find that the partition I saved to has lost its label and claims to be empty. Here is the info, sda3 is the partition I saved to:

sudo fdisk -lu



Disk /dev/sda: 2000.4 GB, 2000398934016 bytes

255 heads, 63 sectors/track, 243201 cylinders, total 3907029168 sectors

Units = sectors of 1 * 512 = 512 bytes

Sector size (logical/physical): 512 bytes / 4096 bytes

I/O size (minimum/optimal): 4096 bytes / 4096 bytes

Disk identifier: 0x0005901e



   Device Boot      Start         End      Blocks   Id  System

/dev/sda1            2048     4294655     2146304   82  Linux swap / Solaris

/dev/sda2   *     4294656  1172654030   584179687+  83  Linux

/dev/sda3      1172654080  3907028991  1367187456   83  Linux

Note: sector size is 4096 (not 512)

In Gparted, I am told the superblock is bad on this partition. It does offer the option to assign a new UUID to the partition, which I haven't tried yet.

So I don't know what to do now, was the transfer made at all? Is the data on this partition?

I think the bad drive was mounted when I did the transfer, I wasn't warned about it and so didn't notice. I think I can get another copy of the disc but I do want to be careful here. Do I have anything worth working with, and can I get into this drive somehow? I understand that the sector size has changed due to the dd, and also that the error on the external is now on the partition.

But how to access this? Should I have made an image? ANY advice would be appreciated.

Answers

Only authorized users can answer the question. Please sign in first, or register a free account.

fectlyole

Add To Favorites

Follow

Total Points: 64

Total Questions: 110

Total Answers: 110

Location: Nicaragua

Member since Thu, Feb 3, 2022

2 Years ago

answered 3 Years ago scrubuz · Accepted Answer

Check the result of dd with echo $? right after termination. It has to be 0! Check the output of dd for obvious error. I don't see the size of /dev/sdb in the question.

Specifying a device file without partition number doesn't copy the partition, but the whole device including the partition table. Either you know which partition to rescue, i.e. specify /dev/sdaX as if argument of dd or make and image of the whole device (better idea), mount it under a loop device, e.g. findout next free loop device with losetup -f, use it with losetup /dev/loopY /path/to/image.img and make the kernel recognize the partitions with kpartx /dev/loopYp? where ? should be the number of the partition you wanted to rescue the way described in you question.

Depending on how severely broken the disk it, i.e. whether further reads might damage it more, and how much space you have on any storage device, make an image of the original or of the copy (the result would be the copy of a copy, but this might prevent damage to the original by reading).

Speed up things: When you use the copy of /dev/sdb on /dev/sda3 (only use it if you have to), you are copying a lot of useless data if /dev/sdb is only 160 GB (the difference between the end of /dev/sdb and the end of /dev/sda3). If you want to skip that, tell dd to only copy the length of /dev/sdb with bs and count argument (man dd is mandatory reading in this case).

Maybe there's a way to tell the kernel with some tweaks that /dev/sda3/ contains a partition table, but I'd consider this rather tricky and experimental.