I share the same policy.  If I see any sort of error on a tape that looks
like it's jeopardizing the repository, I immediately move the data off
using 'move data' or worst case I delete the volume using discard=yes.  The
absolute worst possible thing I can think of is going to restore someone's
file and finding out that it's unavailable for some reason.  That's my
biggest nightmare, and luckily in my 15+ years of using ESMS/WDSF/ADSM/TSM
that has not happened yet.

The cost of the tape or the time needed to get rid of it is minuscule
compared to the value of a file that a user wants back really badly and
can't have back because it's not backed up reliably.

At 3/11/2004 07:41 AM Thursday, you wrote:

I have not found a useful policy. If I start getting write errors on tapes
I make them a candidate for retirement, no matter how new or old they are.