QIC-24 problem
Operator
root at fuzzy.ucsc.edu
Mon Oct 8 07:30:00 AEST 1990
Looking for advice with 1/4" cart tape problem.
System: SUN 3/50 8MB OS 4.1 (fuzzy.ucsc.edu)
In 3rd party shoebox: WANGTEK 5099ES (60MB QIC-24 1/4" cart tape)
terminators removed [first device]
WREN IV 94171-376 (300 MB disk)
terminators in [last device]
Symptoms: Tape fails intermittantly during tar tv short tars seem to go ok
and read back ok on large old tar tapes, tar tv usually gets read error
part way through large (30MB+) new tars seem to write ok but errors during
tar tv to check tape - read error reported.
error messages in /usr/adm/messages
st0: tape syncronization lost
st0: tape rewinding
in SUNDIAG, does ok on several (e.g. 7 )passes of short (1000 blk) and
quick (382 blk) test but usually eventually dies on "Big read I/O error,
sense key=0x17" and tape light stays on. I can get the tape test started
again by stopping sundiag, removing and reinserting tape and starting
test. The read errors typically happen during 1st of 4 File tests.
In test to EOT, a variable number of blocks are written: 128773, 129403,
129655, 130033. Typically tape will fail on readback at random points in
tape: e.g. 32257, 96769 blocks in. On the 130033 write it suceeded in
reading back same number. Occasionaly get write errors, e.g. "Big write
failed on /dev/nrst0 block 129277 I/O error, sense key 0x4"
2 new 3M DC 600A tapes were used in tests. SUNDIAG recognizes st0 as a
WANGTEK 150 and does not offer usual QIC-11/QIC-24 choice that is offered
for other QIC-24 tape drives.
Questions:
Might the problem be due to the fact the the OS thinks this is a QIC-150?
If so, is there a fix (e.g. to streg.h)? Do I need to get rid of this
mdel of QIC-24 and get something else? (I need the QIC-24 rather than
QIC-150 to communicate with a 3/110 not on the net). Is it probably just
a flakey drive that should be swapped? Can someone point to doc
explaining sense key meanings?
Michael M. Cohen, Experimental Psychology, UCSC 408-459-2655
root at fuzzy.ucsc.edu
More information about the Comp.sys.sun
mailing list