>  > The question is, does DLT (or other technology) acknowledge the write
>  > before or after verifying the tape.  If it's before the verification,
>  > then a situation could arise where the verification fails and there is
>  > not enough tape left in the cartridge to rewrite the buffer.  I'll bet
>  > that is the case.
>
> ...doesn't this hypothesis imply that DLT drives are quite broken
> by design???

Let me rephrase my statement, it was very unclear.

The thing that I'll bet is that case is that the DLT unit will
acknowledge data written to the buffer prior to writing it to the
physical tape.

I did not mean to imply that I think it common that it actually runs
into problems by doing so.

>  > My understanding is that DLT has a "warning" mark some distance from the
>  > end (EOT vs EOM).  The drive should attempt to write all buffered data
>  > after reaching the warning mark and then return EOT to the host.  If
>  > true, then this particular issue is probably not a big deal for DLT
>  > users.  I have no idea if other technology deals with it in similar ways
>  > or not.
>
> ...tf this understanding is true, then wouldn't the above
> hypothesis would be false??
>
> Apologies for being especially ornery today.

No problems.

I think that the separation of EOT and EOM gives it enough of a safety
margin so that acknowleged writes are *very* likely to be correctly
written.  Of course I can think up a case where the media available
between EOT and EOM is so damaged that a successful write cannot
occur...

> Has anyone read the last saves on some DLT tapes lately?
> =:)

I've certainly had problems with networker (5.5 I think), DLT 7000s, and
a Solaris server that didn't seem to notice they were overwriting the
beginning of the tape.  Never did track down whose fault that was.  But
obviously problems can occur.  If AMV doesn't take too long, then I'm
happy to enable it as another check layer.

If you're cloning all or most of your savesets, I wouldn't bother and
would rely on that to act as a media verification step.

> Perhaps it's worth pointing out that if we second-guess
> vendor's specs, software's functionality, and the like, then
> there would be no time to get any real work done!

I've looked over the DLT 7000 product manual.  It talks about buffers
and EOT/EOM, but I can't see anything in there that appears to be a
guarantee that acknowledged data will be written.

Again, I think it very unlikely, but posssible that this could happen.

If someone has other knowledge or knows what I'm overlooking in the
manual, then I'd be happy to be incorrect about this.  I also have no
idea if this is affected/controlled by the tape driver.



--
Darren Dunham                                           ddunham AT taos DOT com
Senior Technical Consultant         TAOS            http://www.taos.com/
Got some Dr Pepper?                           San Francisco, CA bay area
         < This line left intentionally blank to confuse you. >

--
Note: To sign off this list, send a "signoff networker" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=