Very Computer

Greetings,

On a linux system we have two DLT4000 drives which I want to use to do
tape to tape copies. Howerver when I tried:

  tcopy /dev/nst0 /dev/nst1

or

  dd if=/dev/nst0 of=/dev/nst0 ibs=10240 obs=10240

the drives wouldn't stream continuously.  It wasn't even close - they'd
whir for a second, pause (rewind slightly), whir again, etc.
Apparently both tcopy and dd read from one drive, then write to the
other.  So the drives can't keep streaming.

Is there somewhere on the net a tape utility bright enough to be able to

either multithread or at least utilize a nice large buffer to speed this
process up?

Thanks,

David Mathog

Quote:> Greetings,

> On a linux system we have two DLT4000 drives which I want to use to do
> tape to tape copies. Howerver when I tried:

>   tcopy /dev/nst0 /dev/nst1

> or

>   dd if=/dev/nst0 of=/dev/nst0 ibs=10240 obs=10240

> the drives wouldn't stream continuously.  It wasn't even close - they'd

Look into the 'buffer' program.

--

Got it, but I don't see how to make it work with tcopy.  It does work with
dd though.
Examples:

% tcopy /dev/nst0 /dev/nst1
file 0: block size 10240: 4017 records
file 0: eof after 4017 records: 41134080 bytes
file 1: block size 10240: 1201 records
file 1: eof after 1201 records: 12298240 bytes
etc.

(slow, but it does work, rewind both drives)

% tcopy -x /dev/nst0 /dev/stdout | buffer -s 10240  | tcopy /dev/stdin
/dev/nst1
or
% tcopy /dev/nst0 /dev/stdout | buffer -s 10240  | tcopy /dev/stdin
/dev/nst1
file 0: block size 4096: records 0 to 7
file 0: block size 2048: record 7
file 0: block size 4096: records 8 to 10
file 0: block size 2048: record 10
file 0: block size 4096: records 11 to 13
file 0: block size 2048: record 13
file 0: block size 4096: records 14 to 16
file 0: block size 2048: record 16
file 0: block size 4096: records 17 to 19
file 0: block size 2048: record 19
file 0: block size 4096: records 20 to 22
file 0: block size 2048: record 22
etc...
file 0: block size 2048: record 763
file 0: block size 4096: tcopy: tape op: Invalid argument
records 764 to 10111
file 0: block size 2160: record 10111
file 0: eof after 10112 records: 41134192 bytes
eot
total length: 41134192 bytes

Rewound the output tape and checked  - it had the block sizes shown.

(rewind both again)

% dd if=/dev/nst0 ibs=10240 obs=10240 | buffer -s 10240  | dd of=/dev/nst1
ibs=10240 obs=10240
4017+0 records in
4017+0 records out
0+12050 records in
4017+0 records out
% dd if=/dev/nst0 ibs=10240 obs=10240 | buffer -s 10240  | dd of=/dev/nst1
ibs=10240 obs=10240
1201+0 records in
1201+0 records out
0+3603 records in
1201+0 records out

Checked the contents on the copy with:

mt -f /dev/nst1 rewind
restore -f /dev/nst1 -t
mt -f /dev/nst1 fsf
restore -f /dev/nst1 -t

and the  restore listings looked ok.

So dd can be pushed through buffer, but apparently not tcopy.  That's too
bad because what I really want
is

tcopy -c /dev/nst0 /dev/nst1

(copy and verify) and I'd like it to run a lot faster than it does now.

Thanks,

David Mathog