Hope many of you have seen this advisory from Veritas today.
http://seer.support.veritas.com/docs/273806.htm

The advisory is confusing too, on one hand they say " in NetBackup 
5.0, the size of a single stream backup job has been extended to 2
exabytes (2000 TB)." Then they say "This issue can occur on all
versions of NetBackup when restoring from Network Data Management
Protocol (NDMP) images that are 1 terabyte in size. Backups may exit
with a Status 0 (successful); however, restores of these images may
fail."

They basically recommend breaking up NDMP jobs to less than a TB each.
This is ridiculous. We have a few NetApp Nearlines,  that are used as
disk backups for the primary filers (snapvault/snapmirror) and have
large volumes (2 to 4 TB) with hundreds of qtrees. We use Netbackup to
run weekly off-site tape backups. The current file list just has the
list of volumes on the filers.
Now to fix this issue that's specific to Veritas, they recommend
manually adding qtrees to the netbackup policy. This would be
nightmare as qtrees get added and removed frequently. And the last
time I tried, it didn't support wild cards in the file list for NDMP
backups. What a pain. All the truck load of tapes we have been sending
off-site for the past few months (after upgrading to  5.x) are now
suspect. Hope I don't have restore from any of them. And the boss
wants all policies to be updated to reflect this recommendation. Why
doesn't he hold Veritas accountable? How can software companies get
away with this kind of shoddy products? May be they get their
inspiration from Microsoft...

How are others managing large NDMP backups?  Not using  Veritas for
it?  Veritas and NDMP  never got along well for a long time and I
don't think it will for a long time  either. They have been promising
SSO support for NDMP since 4.5. Go figure

Enough of my rants, back to editing the policies, I have a long night
ahead of me  - sigh :(

-G

PS: now you know why my last name is Rage :)