I'm learning.

I looked in /opt/openv/netbackup/db/media/errors and found a bunch of read
errors.

I parsed the file with grep/cut/sort/uniq and came up with around 22 different
tapes that present read errors, all since May 14th, a few days before we
made our original system changes.

So, it seems that netbackup is doing the right thing and marking the drives
as down, when it is seeing the read errors.

So, now we are:
- increasing the logging levels 
- checking the storagetek array (9710) for hardware problems.
- going to try new tapes

It's hard to believe that 20+ tapes are all bad, and it's also not a coincidence
that both arrays were having problems.  Could there be something at the Sun 
level causing read errors?  In my experience, read errors are either bad
tapes or bad heads.

To answer a couple of other questions I received, we've run robtest OK, we don't
have a separate media server, and we're reinventoried the robot (actually 
reinstalled 4.3 completely yesterday).

I'm pointing to hardware problems at this point, how about you?