16. DDRescue: data recovery from damaged disk

A bad sector is a sector on a computer’s disk drive that is either inaccessible or unwritable due to permanent damage, such as physical damage to the disk surface. Flash memory may also have “bad sectors” (even if technically there is no sector in flash memory) due to permanent damage like failed flash memory transistors.

Instead of working directly on the damaged disk, it’s recommended to create a copy and to work on the clone. Two possibilities: create a disk image (a file) or overwrite a new/empty disk.

When cloning a disk to a healthy disk, the destination disk will remain healthy. There is no way to recreate the missing content (content that was stored in the sector that now failed to be read), so if the file that was using this sector is “recovered”, it will be damaged/corrupted.

Warning

Do not reformat a disk if you want to recover its content. Do not reuse a disk with bad sectors. Reinstalling the OS or reformating the partition will at best hide the problem for a moment.

ddrescue can be found for Linux or macOS. If your computer is using another operating system, no problem, create a Linux Live USB! (See Creating a live USB)

16.1. ddrescue on Linux

ddrescue is available on all Linux distribution.

  • CentOS: yum install ddrescue

  • Debian/Ubuntu: apt install gddrescue

  • Fedora: dnf install ddrescue

Use lsblk or testdisk -lu to identify all the disks.

16.2. ddrescue on macOS

To install ddrescue:

  • Press Command+Space and type Terminal and press enter/return key.

  • Run in Terminal app:

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install.sh)"
brew install ddrescue

Done! You can now use ddrescue. Use diskutil list to get information on all available disks and their partitioning.

16.3. DDRescue: disk to file image

It’s the recommended method for forensic purpose. You need enough space to store the file: if you want to create a clone of a 1TB disk, you need at least 1TB free on a filesystem. Avoid FAT filesystem for the destination as they are limited to 4GB file.

In the following example, an image named image_sdb.dd will be created from the second disk /dev/sdb.

ddrescue /dev/sdb image_sdb.dd sdb.log

The log file sdb.log can be used to restart the recovery. It can take a few hours to several days to clone a disk with a lot of bad sectors.

Note

If your version of ddrescue support them, the options –verify-on-error –reopen-on-error are useful when the disk interface is unstable (commun with USB devices) and with –min-read-rate=1M, ddrescue will keep the slow areas for the end of the recovery.

16.4. DDRescue: disk to disk copy

The destination disk must be at least as big as the original one. Be careful, two disks of the same announced capacity from different vendors or sometimes from different models of the same vendor can differ slightly in size (a few 100 MB).

Ie. WD10EZRZ and WD10EZEX are two models sold by Western Digital as 1TB model, in fact the first one is 1,000,000 MB, the second one 1,000,204 MB.

Before beginning, disconnect all disks, USB device, CD/DVD reader/writer not needed: there is less chance to overwrite the wrong disk.

ddrescue --force /dev/sdb /dev/sdc sdb.log

The log file sdb.log can be used to restart the recovery.

16.5. ddrutility: restricting ddrescue to NTFS allocated data block

When a disk contains a lot of bad sectors, it may be safer to use ddrutility to limit the copy to allocated data block from an NTFS partition.

testdisk -lu /home/kmaster/data/data_for_testdisk/ntfs.dd
TestDisk 7.1-WIP, Data Recovery Utility, August 2016
Christophe GRENIER <grenier@cgsecurity.org>
http://www.cgsecurity.org
Please wait...
Disk /dev/sdb - 130 MB / 124 MiB - CHS 16 255 63 (RO)
Sector size:512


Disk /dev/sdb - 130 MB / 124 MiB - CHS 16 255 63 (RO)
     Partition                       Start        End    Size in sectors
 1 * HPFS - NTFS                   32     255487     255456 [NTFS]
     NTFS, blocksize=512

In this example, the first NTFS partition begins at sector 32 and the sector size is 512 bytes.

ddru_ntfsbitmap /dev/sdb -i $((32 * 512)) sdb1_domain
ddrescue /dev/sdb sdb.dd sdb.log -m sdb1_domain