ddrescue

ddrescue [options] infile outfile [mapfile]

ddrescue is a data recovery tool that copies data from one block device to another while handling read errors intelligently. Unlike dd, which stops on the first read error, ddrescue skips bad sectors and continues copying good data, then returns to problematic areas for repeated retry attempts.The mapfile (called logfile in older versions) tracks which blocks have been successfully copied, which failed, and which remain untried. This allows operations to be interrupted and resumed without starting over, making it practical to run multiple passes over days or weeks. The first pass typically uses -n to quickly copy all readable data, skipping errors. Subsequent passes focus on bad sectors with different strategies.ddrescue employs sophisticated algorithms to maximize data recovery from failing drives. It can read in reverse, try different block sizes, and make multiple attempts with delays between retries, as sometimes failing drives become temporarily more cooperative. The tool is essential for recovering data from physically damaged media, handling situations where traditional copying tools would fail completely. Always work with a spare drive for recovered data rather than attempting in-place recovery.

-f, --force

Overwrite output device

Skip scraping phase (faster first pass)

Maximum retry passes for bad sectors

Use direct I/O

Read input in reverse

Always use a mapfile to enable resumption. The -f flag is required when writing to a device. For best results, do a quick first pass with -n, then retry passes for bad sectors. Do not confuse GNU ddrescue with dd_rescue, a different tool with similar goals.

GNU ddrescue was written by Antonio Diaz Diaz, with the first release in 2004. It was created to address dd's inability to continue past read errors and has become the standard free tool for rescuing data from failing drives.

dd(1), testdisk(8), photorec(1)

Homepage · Documentation