How to use ddrescue to clone disk drive (better than ghost for failing drives)

See also DataRecoveryCleanupFiles

1. Identify disks

Run this to identify the disks connected to your linux PC

lsblk | grep ^sd; echo; lsblk|grep ^sd |sed -e 's/ .*$//'| xargs -i hdparm -I /dev/{} | grep -v '^ *$' | grep 'number\|device size\|/dev' -i | grep -v '1024[*]1024'

2. Clone

If you know that about 100GB are used in a 400GB disk set USED_SIZE to 125GB (real use +25% overhead). If not sure set a random number, it doesn't hurt.

#-------SET THESE FIRST----------
SRC=/dev/sdX
DST=/dev/sdY   #<--- you can also write the disk to a file e.g. /mnt/temp/disk.raw
USED_SIZE=125GiB
A_NAME=RN01234
#      ^--It's best to use the RN number but you can type anything you like
#         If you have to rescue more than one disks from the same RN add -1,2,...
#------------------------------------


QUICK="-f -n -u -r1"
THOROUGH="-f -d -r3"
PARANOID="-f -d -r-1"
# How hard to try
#------------------------------------
# -d direct access to disk (slower)
# -u only scan forwards (faster)
# -n skips scraping phase (faster)
# -r how many times to retry (-1 for ever)
# -f force writting to disk/partition(instead of file)

# which part of disk to work on
#------------------------------------
FIRST_MB="-i0 -s1MiB"
FIRST_FIFTY_MB="-i1MiB -s50MiB"
USED_PART="-i50MiB -s$USED_SIZE"
REST_OF_DISK="-i$USED_SIZE"
EVERYTHING=""

# standard options
#------------------------------------
STD_OPTS="$SRC $DST $A_NAME.ddrescue"

# Main part
#------------------------------------
ddrescue $QUICK    $FIRST_MB       $STD_OPTS
ddrescue $QUICK    $FIRST_FIFTY_MB $STD_OPTS
ddrescue $QUICK    $USED_PART      $STD_OPTS
ddrescue $THOROUGH $FIRST_MB       $STD_OPTS
ddrescue $THOROUGH $FIRST_FIFTY_MB $STD_OPTS
ddrescue $QUICK    $REST_OF_DISK   $STD_OPTS
ddrescue $THOROUGH $USED_PART      $STD_OPTS
ddrescue $THOROUGH $REST_OF_DISK   $STD_OPTS
ddrescue $PARANOID $EVERYTHING     $STD_OPTS
echo ; echo "ALL DONE"
echo
echo

Consider monitoring for disk errors while cloning

You can monitor for disk errors from another terminal like this:

tail -f /var/log/syslog | grep --line-buffered -w 'sd[a-z]'

If while copying from /dev/sda to /dev/sdb, /dev/sda freezes up at position 12345678.

Reboot computer and restart copy at freeze_postition + *10000*

ddrescue -f -i 12355678 $SRC $DST mapfile

Start copying backwards down to the troubled sector

ddrescue -f -R $SRC $DST mapfile

If disk begins returning read errors from position 12345678 onwards

stop ddrescue and reboot computer

ddrescue -f -n -A -O -i 12345678 $SRC $DST mapfile

If disk fails again, reboot computer and then repeat the above command as many times as needed until it succeeds.

ddrescue -d -r3 $SRC $DST mapfile

If while copying from sda, it disappears from /dev.

Reboot computer and then repeat the same command as many times as needed until it succeeds

Finaly

You can consider adjusting the timeout to read a sector (change sdX to the disk id). You may wish to make it small in order for a quick first clone with many errors and then make it large (e.g. 2sec) to try hard to copy the few failing sectors:

echo 1 > /sys/block/sdX/device/timeout