NAME

       diskscan - scan a disk for failed and near failure sectors

SYNOPSIS

       diskscan [options...] block_device

DESCRIPTION

       diskscan is intended to check a disk and find any bad sectors already present and assess
       it for any possible sectors that are in the process of going bad.  The operation is all
       read-only and can cause no direct damage to the data on the disk.

       diskscan reads the entire block device and notes the time it took to read a block. When
       there is an error it is immediately noted and also when there is a higher latency to read
       a block. A histogram of the block latency times is also given to assess the health of the
       disk.

       The output of diskscan will show any serious errors or very high latency and will also
       emit an histogram at the end of the run in the form:

           I: Validating path /dev/sdg
           I: Opened disk /dev/sdg
           I: Scanning disk /dev/sdg
           Access time histogram:
                  1: 0
                 10: 0
                100: 0
                500: 120
               1000: 0
               2000: 1
               3000: 0
               4000: 0
               5000: 0
               6000: 0
               7000: 0
               8000: 0
               9000: 0
              10000: 0
              15000: 0
              20000: 0
              25000: 0
              30000: 0
           above that: 0
           I: Closed disk /dev/sdg

       This means that all I/Os in this case were between 100 and 600 msec and there were 120
       chunks being read. Current these chunks are 1MB in size.

OPTIONS

       -v, --verbose display verbose information from the workings of the scan use multiple times
       for increased verbosity.

       -f, --fix Attempt to fix areas that are nearing failure. This should only be attempted on
       an unmounted block device and never on an inuse filesystem or corruption is likely.

       -s <mode>, --scan <mode> Scan mode can be either seq or random, random reduces the chance
       that the disk will be able spend time to recover data before we try to access a sector but
       the seeks add noise to the latency measurement. Sequential test is the default and random
       test is still experimental with regard to its usefulness.

SEE ALSO

       badblocks(1), fsck(1)

AUTHOR

       Baruch Even \<baruch@ev-en.org>