... or so it would seem. My 4G IBM DCAS was only about a year old when it
went south and took my cvs repository with it. I got an IBM DDRS 4G drive
as replacement which appears to be a different drive hardware (7,200
instead of 5,400 rpm) with the same controller (at least storage.ibm.com
points to the DCAS technical manual for DDRS details).

This DDRS tends to spin down every now and then "out of the blue".
Sometimes I see comments in the syslog, sometimes the machine simply hangs.
I cannot tell whether the drive first spins down and up again or whether
some rude action from the driver forces the drive into a cold reset.

Examples:

The occasional

Jun 12 23:08:19 espresso /netbsd: -->scsi_show: scsi_main is not running
Jun 12 23:08:19 espresso /netbsd: Bus signals (00/00):
Jun 12 23:08:19 espresso /netbsd: Dma status (08): PMATCH
Jun 12 23:08:19 espresso /netbsd: busy:0, spl:2204
Jun 12 23:08:19 espresso /netbsd:       0       pdma_ready0
Jun 12 23:08:19 espresso /netbsd:       1       pdma_ready4
Jun 12 23:08:20 espresso /netbsd:       2       reselect1
Jun 12 23:08:20 espresso /netbsd:       3       tpio1
Jun 12 23:08:20 espresso /netbsd:       4       tpio2
Jun 12 23:08:20 espresso /netbsd:       5       reselect2
Jun 12 23:08:20 espresso /netbsd:       6       scsi_main1
Jun 12 23:08:20 espresso /netbsd:       7       info_transf1
Jun 12 23:08:20 espresso /netbsd:       8       tpio1
Jun 12 23:08:20 espresso /netbsd:       9       tpio2
Jun 12 23:08:20 espresso /netbsd:       10      info_transf6
Jun 12 23:08:20 espresso /netbsd:       11      info_transf1
Jun 12 23:08:20 espresso /netbsd:       12      tpio1
Jun 12 23:08:20 espresso /netbsd:       13      tpio2
Jun 12 23:08:20 espresso /netbsd:       14      hmessage1
Jun 12 23:08:20 espresso /netbsd:       15      linkcheck
Jun 12 23:08:20 espresso /netbsd:       16      cautos1
Jun 12 23:08:20 espresso /netbsd:       17      cautos3
Jun 12 23:08:20 espresso /netbsd:       18      hmessage5
Jun 12 23:08:20 espresso /netbsd:       19      scsi_main2
Jun 12 23:08:20 espresso /netbsd:       20      scsi_main4
Jun 12 23:08:20 espresso /netbsd:       21      scsi_main5
Jun 12 23:08:20 espresso /netbsd:       22      irq
Jun 12 23:08:20 espresso /netbsd:       23      pdma_ready0
Jun 12 23:08:20 espresso /netbsd:       24      pdma_ready4
Jun 12 23:08:20 espresso /netbsd: ncrscsi0: Spurious interrupt.

and

Jun 27 23:39:26 espresso /netbsd: ncrscsi0: Resetting SCSI-bus (Timeout waiting
for phase-change)
Jun 27 23:39:26 espresso /netbsd: ncrscsi0: Resetting SCSI-bus (Timeout waiting
for phase-change)
Jun 27 23:39:26 espresso /netbsd: sd1(ncrscsi0:2:0): Sense Error Code 0x0
Jul  3 03:24:28 espresso /netbsd: ncrscsi0: Resetting SCSI-bus (Timeout waiting
for phase-change)
Jul  3 03:24:28 espresso /netbsd: sd1(ncrscsi0:2:0): Sense Error Code 0x0

and

Jul  6 23:08:20 espresso /netbsd: sd1(ncrscsi0:2:0): Sense Error Code 0x36
at bl
ock no. 14353920 (decimal)
Jul  6 23:08:20 espresso /netbsd: ncrscsi0: Resetting SCSI-bus (Timeout waiting
for phase-change)
Jul  6 23:08:21 espresso /netbsd: sd1(ncrscsi0:2:0): Sense Error Code 0x36
at bl
ock no. 1536 (decimal)
Jul  6 23:09:19 espresso /netbsd: zstty0: 4 silo overflows, 0 ibuf floods
Jul  6 23:10:19 espresso /netbsd: zstty0: 1 silo overflow, 0 ibuf floods
Jul  7 03:23:47 espresso /netbsd: ncrscsi0: Resetting SCSI-bus (Timeout waiting
for phase-change)
Jul  7 03:23:47 espresso /netbsd: sd1(ncrscsi0:2:0): Sense Error Code 0x6d
Jul  7 03:23:48 espresso /netbsd: ncrscsi0: Resetting SCSI-bus (Timeout waiting
for phase-change)
Jul  7 03:23:48 espresso /netbsd: sd1(ncrscsi0:2:0): Sense Error Code 0x36
at bl
ock no. 1536 (decimal)

[The machine was wedged here during a sup run]


-- "espresso" is a headless IIci with two disks running vanilla 1.4. The
Quantum Empire runs flawless, and always has.

Jun 24 22:21:56 espresso /netbsd: ncrscsi0 at obio0
Jun 24 22:21:56 espresso /netbsd: scsibus0 at ncrscsi0: 8 targets, 8 luns
per ta
rget
Jun 24 22:21:56 espresso /netbsd: sd0 at scsibus0 targ 0 lun 0: <QUANTUM,
EMPIRE
_1080S, 1220> SCSI2 0/direct fixed
Jun 24 22:21:56 espresso /netbsd: sd0: 1029MB, 2874 cyl, 8 head, 91 sec,
512 byt
es/sect x 2109376 sectors
Jun 24 22:21:56 espresso /netbsd: sd1 at scsibus0 targ 2 lun 0: <IBM,
DDRS-34560
, S97B> SCSI2 0/direct fixed
Jun 24 22:21:56 espresso /netbsd: sd1: 4357MB, 8387 cyl, 5 head, 212 sec,
512 by
tes/sect x 8925000 sectors


-- I am currently running an sbc kernel. Unfortunately, the sbc driver does
not do disconnect/reconnect afaik, so tape operations are a pain.

If anybody has ideas what switches to toggle on ncrscsi (Allen?), I'd be
happy to try them out.

	hauke



--
"It's never straight up and down"     (DEVO)