Alan - Looks like your multi-process storage pool backup encountered a file
       that spanned volumes.  The TSM server code might do better in
scheduling the sequence of tape mounts to try to avoid such a conflict;
but timing over the long period that such backups take can thwart schemes
to deal with the situation, particularly as other events intervene (delays
due to tape drive availability problems, a higher-priority task taking
drives, etc.).

What would help would be a new server option, SPANVOLUMES, which we could
set No to prevent the complications that come with multi-process tasks,
where we'd be willing to live with the additional time to abandon the
incomplete file written at the end of a volume to start writing it afresh
on a new volume.

  Richard Sims, BU