ddrescue

Take an image of a device, creating a log file

$ sudo ddrescue [/dev/sdb] [path/to/image.dd] [path/to/log.txt]

$ sudo ddrescue [[-f|--force]] [[-n|--no-scrape]] [/dev/sdX] [/dev/sdY] [path/to/log.txt]

ddrescue [options] infile outfile [logfile]

-a, --ask
    ask before overwriting existing data in outfile

-b, --block-size=bytes
    set data block size to bytes; default 1024

-B, --binary-prefixes
    use binary (KiB, MiB, GiB) instead of decimal prefixes

-c, --cluster-size=n
    read forward n blocks then backward 1 block

-C, --compare
    compare trial copies to final copies in third pass

-d, --direct
    use direct disk access for infile and outfile

-D, --retrim
    trim areas of outfile before writing

-e, --skip-errors
    continue after all read errors (same as --force)

-f, --force
    overwrite outfile and ignore its size

-F, --fill-mode
    fill with filltype: zero, zeroes, zeros, nul, nuls, nul-fill, nulls, 0x00

-g, --max-error-granularity=n
    exit if average bad sector length is n blocks or more

-G, --generate-badblocks
    output bad blocks list to stdout

-h, --help
    print this help

-i, --ignore-ipcs
    do not abort on SIGINT and SIGPIPE if IPCS is full

-I, --iterative
    do multiple passes over bad areas until all are rescued

-J, --jpeg
    try to repair damaged JPEG files

-K, --keep-size
    do not truncate outfile

-L, --color-level=n
    control use of ANSI color codes; 0=none, 1=err, 2=warn, 3=all

-m, --match-blocksizes
    match block sizes in mapfile

-M, --do-mapfile-checksums
    compute and check mapfile checksums

-n, --no-scrape
    skip the scraping phase

-N, --no-trim
    do not trim trailing zeros

-o, --overwrite
    overwrite existing data; same as --force

-O, --no-overwrite-errors
    continue after write errors

-P, --compare-process=n
    compare n blocks in parallel

-Q, --use-pcache-size=bytes
    set persistent cache size

-r, --retry-passes=n
    do n scraping passes; default 0

-R, --reverse
    read blocks in reverse order

-s, --size=bytes
    copy only bytes starting from 0

-S, --sparse
    use sparse writes for outfile

-t, --truncate
    truncate outfile to zero length

-T, --timeout=n
    wait n seconds for read requests

-u, --unbuffered-input
    use unbuffered input

-U, --unbuffered-output
    use unbuffered output

-v, --verbose
    be more verbose

-V, --version
    print version info

-w, --wait-retry=n
    wait n ms before retrying read

-W, --no-warnings
    suppress warnings

-x, --ignore-checksums
    do not check mapfile checksums

-X, --extend-outfile
    extend outfile if it is shorter than infile

-y, --retry-mapfile
    retry reading mapfile on error

-Z, --debug
    set debug level (0-10)

ddrescue is a powerful GNU utility for recovering data from damaged storage media like hard drives, SSDs, CDs, or DVDs. Unlike dd, which stops or retries indefinitely on errors, ddrescue intelligently skips bad sectors, marks them in a log file, and retries in later passes using sophisticated algorithms. It performs multiple passes: first a direct copy of good data, then scraping bad areas with reduced speed, and finally intensive retrying.

It supports huge files (terabytes), direct I/O for speed, cluster reads, and splitting large jobs. A mapfile (logfile) tracks progress, domains (good/bad/non-tried), and allows safe resumption after interruptions. This makes it essential for disk cloning or imaging failing drives.

Key advantages include non-contiguous copying, reverse direction scraping (-R), and options for sparse files or truncation. Always run as root for block devices and test on copies first to avoid data loss.

Requires root for raw devices; test on images first to avoid data loss; logfiles essential for resumability; can take days/weeks for large failing drives; no write support to source.

ddrescue -d -r3 /dev/sdX image.img rescue.log
Starts fast copy with direct I/O, 3 retries, log for resume.

1. Fast copy: ddrescue -n /dev/sdX image.img mapfile
2. Scrape: ddrescue -d -r3 /dev/sdX image.img mapfile
3. Intensive: ddrescue /dev/sdX image.img mapfile

Developed by Antonio Díaz Díaz since 2000; GNU ddrescue 1.0 released 2004; latest 1.11.1 (2024). Evolved from dd_rescue, now standard for data recovery, with improvements in multi-pass algorithms, speed, and large disk support.

dd(1), dd_rescue(1), ddrescueview(1), gnu-parted(8)