On 06/06/10 18:46, Neil Bothwick wrote:
[]
>
> You can use tee to send the dd output to both the destination and md5sum.
Dang! I'd forgotten about tee; a core utility!
Yep....... using tee as you described would be the classic way to md5sum
the source; use ubiquitous dd and md5sum to quickly hash the copy.
By any chance have you played with sdd?
<quote>
sdd is a replacement for a program called 'dd'.
sdd is much faster than dd in cases where input block size (ibs) is not
equal
to the output block size (obs).
sdd does not have some of the design bugs of dd that cause the following
command to fail:
dd if=/dev/rdsk/c0t0d0s2 bs=126k | rsh otherhost 'dd ibs=4k obs=126k
of=/dev/rdsk/c0t0d0s2'
The output disk will not be equal to the input disk because the dd command
on 'otherhost' will read fragments of 'ibs' and fill them up to 4kB.
Other features not found on 'dd':
- Statistics are much better readable as from 'dd'.
Output is not # of full blocks + # of partial blocks (not
useful)
But: # of full blocks + # of bytes from partial blocks =
full amount
sparky joerg > sdd if=/dev/rdsk/c0t0d0s0 bs=126k of=/dev/null -t
sdd: Read 268 records + 22528 bytes (total of 34600960 bytes
= 33790.00k).
sdd: Wrote 268 records + 22528 bytes (total of 34600960 bytes
= 33790.00k).
sdd: Total time 10.601sec (3187 kBytes/sec)
- rmt support for if= & of=
call sdd -inull bs=63k count=1000 of=nt...@somehost:/dev/null
to to TCP speed tests.
- output file is sync'd before doing statistic report.
This enforces NFS data integrity and helps to get right numbers
when doing filesystem write perfornamance tests.
- Timing available, -time option will print transfer speed
- Timing & Statistics available at any time with SIGQUIT (^\)
- Can seek on input and output
- Fast null input
- Fast null output
- Reblocking on pipes does not fill small input blocks to
input block size
- Debug printing
- Progress printing
<>
