Conference decwet::networker

NetWorker defines per device type what the block size should be
for reading and writing for that type of device.  Currently, ALL
tape devices are read and written in 32K blocks.  In the near future
we will be writing larger blocks for some devices which are more efficient
with larger blocks.

This message tells us that NetWorker attempted a 32K read, and only got
10K from record on the tape.  That, in combination with the volume name 
being "" tells me that the tape either had not been written by NetWorker
as you thought it had (maybe a fresh tape got miced into the stacker?)
or else the tape is physically corrupted.

NetWorker's reaction is to drop the assumption that it knows where things are
on tape, and to move around by record.  It will still get to the data
if it is there, but positioning will be very slow.

My bet is that somehow a non-NetWorker tape got into the stacker.

Kevin

    Kevin,
    	I guarantee you that no other tapes got into the stacker. I can run
    the backup & when I try to read the data back in it fails. Sometime it
    enters "Unknown" ... although the tapes had been labelled Alpha1.001 &
    Alpha1.002. 
    
    	If I run an inventory on the tapes in the stacker. Only those tapes
    that networker has written to are rejected with the aforementioned
    error. These are then listed as being none-networker tapes. The
    remaining tapes in the rack are registered as valid labelled, but
    unused tapes.
    
    	If there's any info you think I should copy onto this conference
    I'll go back out to site & download it all.
    
    Brendan.

what is the error when you label one of those tapes.

Can you supply us an example of the messages seen when you label a new tape
 as you did when you started this process??

I wonder if these tapes are Compact3 or 2 (or 1)??

Can you try doing 
	mt -f /dev/nrmt1h rewind
	mt -f /dev/nrmt1h weof 4
	mt -f /dev/nrmt1h rewind

 to reset the tape and then try lebelling (relabel) using NetWorker.

Let us know what happens - thanks!

	1.  Check the system error log for any hardware errors.
	
	2.  Make sure that the SCSI interface you are using is supported by
	    the HSZ for tape passthrough.  Not all are.

	3.  Make sure that you have the current version of HSOF for the
	    HSZ40.

	It sounds like the HSZ is not transfering the full 32k byte blocks
	to the tape.  This is not likely to be an application problem.
	Very few people are using NetWorker and tape drives behind
	an HSZ so we are not as experianced at we might like with
	this combination.

    Dear All, 
    
    Thanks very much for your help. I'm terribly sorry for doubting you.
    After thinking about your comments on swapped tapes, it dawned on me
    that if they hadn't been swapped...which they weren't...someone must be
    writing to them. Before visiting the customer site I checked this out,
    running an NSR backup, tarring to the tape with 10k bock size, & trying
    to restore from networker. Sure enough I got the exact same
    errors....the "unknown" bit is because Networker thinks it has a valid
    NSR labelled tape in the drive. After the first verification this is
    dropped.
    
    I visited the customer & found a cron job writing data to the tape
    every night. This has now been stopped & the customer warned
    strenuously that the Stacker should really be dedicated to networker,
    if they want to avoid problems.
      

Thanks for letting us know the outcome!