David> Anyone have a procedure complete with (bpstart_notify and
David> bpend_notify) scripts?

David> I have 10 vobs to backup.

There are several issues that need to be addressed here, and
unfortunately I seem to be hitting a wall here at my site due the
large number of VOBs which are spread across multiple filesystems and
multiple hosts.

You also need to think about how long you want your VOBs to be
locked.  In our case here, we need to minimize how long I lock any one
of 309+ VOBs spread across five different servers.  In your case, with
only 10 VOBs, it's possibly not as big an issue.

In any case, the basic procedure of:

  - get list of VOBS
    - lock each VOB
  - start backup
  - WAIT UNTIL COMPLETION
  - unlock each VOB

works just fine for small data sets, and where you are not doing
development in a 24x7 environment.  At our site, we can't wait until
backups complete, so we have a different procedure currently, which
I'm working on changing from Legato centric to work with NetBackup 4.5
as well.

What we do is on a per-filesystem basis (and the root of all my
problems with NetBackup!!!).  

  - figure out which filesystem we're backing up
  - find VOBs on that filesystem
    - lock each VOB

  - Determine type of filesystem and volume manager used:
    if VxFS then
       make a snapshot volume 10% of existing volume
       make a VxFS snapshot onto the new volume
       mount it to filesystem_backup
    endif

    if UFS then
      if VxVM then
         break off a plex
         make a new volume with this plex
         fsck it
         mount it to filesystem_backup
      endif
      if Solstice DiskSuite then
         break off a sub-mirror
         make a new volume with this sub-mirror
         fsck it
         mount it to filesystem_backup
      endif
    endif

  - unlock each VOB (Key point, this only locks the VOBS for about 5
                     minutes total)

  - backup filesystem_backup

  - unmount filesystem_backup
  - cleanup snapshot volumes and/or filesystems


As you can see, it's a more involved process, but it works well for us
since it lets our developers work while we do our backups.  With 310+
VOBs to backup, and over 600gb of data, we needed this type of setup.

Does anyone have any suggestions on how we can make the above work
with NetBackup?  The use of bpstart_notify and bpend_notify scripts
just doesn't work since you can't get a per-filesystem info out of
them.

And I don't want to have to create a policy per-server-per-filesystem
to get the info I need.  

Thanks,
John
   John Stoffel - Senior Unix Systems Administrator - Lucent Technologies
         stoffel AT lucent DOT com - http://www.lucent.com - 978-952-7548