Exabyte warning - backspace from EOT can mangle tape

[George Goble]

****** Exabyte tape mangling warning - sequence to avoid --ghg 2/22/89. ***

Stan Juday, ME dept computer site specialist here, was "playing around"
with a drive we loaned the ME dept.  This drive was outfitted with a clear
plexi-glass cover.  He filled up the tape and ran it off the end (on
purpose), and got the expected hard error. He then did a backspace file
command "mt bsf N" to back the tape up.  To his horror, the tape began to
get a little slack over Guide #7, and climbed part way off the guide, thus
wrinkling it. The tape later returned to its normal position on the guide.
Guide #7 is at the "10 o'clock" position.  The drive was at firmware level
MX4$23 / SV-B017 (servo).

This is the first case of an Exabyte causing media damage since the summer
of '87 (before it was a product) here. We even had an eye witness!  I had
executed the above sequence 2 or 3 times in the past, with no problems,
but it was on older firmware revisions.  The tape was a SONY P6-120MP, the
standard 2hour (2GB) tape.

More investigation revealed, that his drive tended to have the tape go
slack around Guide #7 and the top Guide during a Backspace file operation
(tape motion is reverse, at 10X read speed, drum tension on).  The tape
did not always come off the guide.  As the backspace proceeded, the left
reel (take up reel) increased in diameter, and the tape tension increased,
so that at about the 3/4 point on the tape, the slack was not observed
anymore.  Seems to be some sort of servo problem.  I mangled a 2nd tape,
on another drive, at the 2/3 point into the tape, by initiating a bksp
file operation.

Exabyte (Jim Werder) has verified this problem to happen there also.  I
would imagine that new firmware will be tested there and here in the next
few days.  Until the servo code can be corrected, I would highly recommend
that Backspace file/record (mt bsf or bsr) operations be prohibited, if
positioned on the 2nd half of the tape.  The special case of "mt bsf 1",
being used to backspace over a filemark, can probably be allowed, since it
only moves around 2", and should not let the slack conditon develope.
Rewind/offline from the end of the tape, occurs with the drum under less
tension, and does not cause the slack condition on Guide #7.  Forward
motion operations (read and file/record space forward) have no problems in
this area.  --ghg 2/25/89

George Goble, Engineering Computer Network, Purdue U, W. Lafayette IN 47907 
(317) 494-3545  Arpa: ghg at purdue.edu  uucp: {backbone}!pur-ee!ghg