It turns out that I am just fixing an APAR that MAY be related
to this problem. The APAR number is IC13430. What is happening
is that at times a SCSI command to a drive is timing out
because drive microcode is doing long-running recovery trying
to either load or read a tape. The APAR will raise the timeout
value the device driver uses so that you don't see a timeout
(CC=205), but will see an I/O error instead, if drive recovery
fails.
On AIX, the timeout means the SCSI bus the device is on gets
reset. That can put other hardware on the bus in a strange
state or the drive itself in a strange state.
Dick Johnson - ADSM Server Device Support, IBM SSD San Jose
----------------------------- Referenced Item ---------------------------
ANR8302E I/O error on drive DRIVE1 (/dev/mt0) (OP=REW, CC=205, KEY=FF,
ANR8302E I/O error on drive DRIVE1 (/dev/mt0) (OP=REW, CC=205, KEY=FF,
ASC=FF, ASCQ=FF).
ANR8355E I/O error reading label for volume ARC015 in drive DRIVE1
(/dev/mt0).
|