[Info-vax] strange backup behavior

[B Hobbs]

On Aug 10, 5:50 am, koeh... at eisner.nospam.encompasserve.org (Bob
Koehler) wrote:
> In article <e7dbbd3a-17e4-469c-b2e3-4a450c788... at p36g2000prn.googlegroups.com>, B Hobbs <bdhobb... at acm.org> writes:
>
> > I've noticed that the more data on our backup SDLT tape, the longer it
> > takes to add additional files to the same tape.  When a file of about
> > 35 blocks (BACKUP-STUFF.COM;67) is backed up at the beginning of the
> > tape, that backup takes less than a minute.  Backing up the same file
> > after about 45 gigabytes of other backup files takes over 30 minutes.
>
>    Every time you moiunt/dismonut a tape, it rewinds, even if you don't
>    unload it.  When you add a save set to the end of a tape, BACKUP
>    has to search forward for the logical end of volume marker (two
>    EOF markers together).

I understand your first sentence and do the minimum mounts and
dismounts.  Your second sentence makes me think that BACKUP and the
tape driver may be causing the problem.

>
>    Some tape drives have the search for EOF capability in thier own
>    firmware, for others the driver has to read each block until it
>    gets an EOF.
>
>    What model SDLT?  Probably someone knows how it works.

The bezel says HP 160/320 GB SDLT, which appears to be a Quantum SDLT
320 with HP firmware.

I'm wondering if the emulator's TK50 tape driver is interacting poorly
with BACKUP and the SDLT drive.  Ideally, someone with a real VAX, an
HP SDLT, and VMS 7.1 could BACKUP a short file to an empty tape
cartridge, BACKUP several tens of gigabytes of data to the tape, then
BACKUP the same short file to tape, all with timing info.  If their
test shows the time for the third BACKUP to be close to the first
BACKUP, then my emulator becomes the prime suspect.