Conference turris::digital_unix

    
    An additional problem I have discovered on actually trying to
    pipe the output of 'exp', is that it does not work.
    
    I tried to pipe the output of 'exp' to 'tar', but possibly because 
    of a wrong parameter given by me, it creates a local file.
    
    For example, I tried -
    $exp userid=scott/tiger file=- | tar cf /dev/rmt0h -
    
    It creates a file called "-.dmp". I tried quoting that parameter, 
    for e.g-
    'file=-' or file=(-) or file=\-.
    
    From the Server Utility User's Guide, I am not able to find an 
    option to pipe out the output of 'exp'.

    Thanks in advance.

    Nandan

    If I rmember right, tape compability between HP and us is covered
    in stkhlm::magtape.
    
    i think your tar problem in .-1 is due to the -f usage.
    if you still have problems, try to use
    dd if=/dev/rmt0[lmh] conv=swap | tar xvf -
    
    Pit

    
    Pit,
    
    Thanks for the pointer. I will look at that notes conf.
    
    However, please note that part of the problem is in Oracle's 'exp'
    command. Besides, I am able to retrieve part of the data, so I guess
    the data may not need the 'swap' flag.
    
    Nandan

The official XPG4 tar archive format doesn't mention
how to form a multi-volume header. In my past experience
I've found that it may be necessary to backwards 
engineer to figure out how the other guy does it.
We have a header that comes up short (is followed by
less blocks than it claimed). Then (I believe) on the
new volume no new header is used, rather we continue
writing the blocks that we started to write when we got
ENOSPC.
Under Ultrix we added a header to precede this
continuation. The advantage was that the second volume
could be seen as a tar tape if you mounted it. However
it could also confuse you, into thinking that the 
continuation section of the file was the whole file.
No telling how the other OS's do it...
It might be worth feeding the first blocks of an HP
continuation tape to tar
	dd if=hp_tape count=2 | tar -t -v -f -

if it's not a header, tar will complain right away

If HP stores it differently, you should still be able
to work it out, using dd as a feeder to tar. If you need
to, dd can also skip records. So for example, DUNIX could
read an ultrix tarset by adding "skip" count to the dd
command when you start on the second volume. You might
want to dd to disk, the files before and after the volume
change, so you can study the problem more carefully.

good luck,
Charlie Katz

    
    Charlie,
    
    Thanks for the info.
    
    Nandan