I think I have the answer to this, but would like someone to verify since
I don't speak "hardware driver-ese".

On this TSM Linux server, the 3590 drives it uses (FC via QLOGIC), are
constantly "trashing" the tapes it writes to. Almost every tape is marked
as "READONLY".

Upon investigation, I found a correlation. In the activity log, I see a
message about "end of file" being reached on the tape, followed by an
error

OP=FLUSH, Error Number=5, CC=0, KEY=40, ASC=00, ASCQ=00.

The IBMtape driver is at 1.4.11.  The only fixes since then (latest is
1.4.14) that might be close (this is where the "hardware driver-ese" comes
in) is:

"Fixed Write Filemark 0 handling at the early warning zone" ?

Your thoughts/suggestions on this ?