[Veritas-bu] If a save is killed, where does nbu see the EOT?
2000-12-06 16:56:33
Subject: |
[Veritas-bu] If a save is killed, where does nbu see the EOT? |
From: |
David A. Chapa david AT datastaff DOT com |
Date: |
Wed, 6 Dec 2000 13:56:33 -0800 (PST) |
This is a true statement, with caveats.
Depending on the timing between when the kills and
resubmits happened, it is likely that netbackup began
writing to a tape after some widowed image(s). This is
the exception not the rule to the operation of the
product. The basis for this type of behaviour is
simply that NetBackup's primary meaning for living is
to backup you DATA.
If some of the jobs from the previous set were still
running while the resubmits were occurring, I would
expect to see something like you described. However,
if there was ample time (and I know you are going to
ask me to define ample...but I can't) between the two
operations, I would *NOT* expect this behaviour and
would follow it up with support directly.
David Chapa
Quoting mike.marcell AT cntcorp DOT com:
> I recently upgraded from 3.2 to 3.4, and this weekend
we had several issues
> that led to the killing of many saves at various
points of completion.
> They were re-started and after completed, our
operations staff noticed that
> we used a considerable amount of additional tapes (by
additional, I mean
> more than we normally use on a weekly full).
>
> I'm curious as to how Netbackup handles this. I was
told by Veritas
> support, that NBU sees that some data has been
written to tape, and will
> begin again at the point that the last save (the
killed saves) ended, as
> opposed to going back and reusing/writing over that
space. NBU can't see
> the data that is there, so it can only be retrieved
via tar. In a nutshell,
> we lost lots of space on tapes (200+ large servers).
>
> Apparently, years ago when we purchased NBU, Veritas
had bragged to us that
> this doesn't happen this way. They indicated that NBU
doesn't have that
> index, so you basically go back to where you began
and try it again, thus
> no losing any tape space.
>
> Can anyone confirm or deny this?
>
> TIA,
>
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
=-=-=-=-=-=-=-=-=
> Mike Marcell
CNT Corp (The Arizona
> Republic)
> mike.marcell AT cntcorp DOT com Unix Systems
Administrator
> Phone: (602) 444-4958 Fax: (602)
444-8010
> Pager: (602) 519-2270 Cell:
(602) 538-5265
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
=-=-=-=-=-=-=-=-=
>
>
> _______________________________________________
> Veritas-bu maillist - Veritas-
bu AT mailman.eng.auburn DOT edu
>
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-
bu
>
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
David A. Chapa
Consulting Manager
DataStaff, Inc.
847 413 1144
|
|
|