>I've seen some sites do an "upd vol acc=readwrite whereacc=readonly"
>automatically every day, just to compensate for TSM's annoying habit of
>marking tapes bad whemever there's a drive failure.

They are treating a symptom, but that may be a realistic recourse for them,
given perhaps a dusty or otherwise contaminated environment they can't do
anything about, which is making for dirt on the tapes.  Or the tapes may be
from a defective batch, with shedding oxide.  Or the drive transport may not
exactly keep the tape aligned with the heads.  (I know of one site which had
its tape robot in a parking garage, and discovered the hard way that air
filters should not be ignored over time.)  One would prefer to address the
root problem, but the company may not support that kind of diligence.

With tapes like 3590 having a Volume Control Region, another approach is to
move the data off the volume and reinitialize it to reset error counters, and
keep records for gauging recurrence, as another customer recently posted.
Having your customer engineer perform a periodic wet cleaning of the tape
transport can do wonders.  But tapes are tapes, and inevitably wear out over
time and repeated use.  Optical platters are conceptually an attractive
alternative serial medium; but tape technology has thus far been making
greater strides in terms of capacity, and so remains dominant...though their
historic problems remain as well.

  Richard Sims, BU