On Tue, 18 Nov 2003 12:39:07 +1100, Luke Mewburn writes:
>If I recall correctly, the "cleaning light" on DLTs is set when the
>firmware detects that the (read) error rate exceeds a certain
>threshold.

Yep, that's not what's happening though.

Last night I gave it another cleaning (looks like new ;-), removed and
reseated the scsi cable inside the box etc, put a new tape in... it
worked for about an hour - doing a backup, then just stopped
responding.  Its done something to the bus too since st1 (a DAT drive)
is also not responding.  Its plugged into the i386 again (aha 1542C)
Hmmm this looks bad:

root@zen:139# st0(aha2:0:4:0): timed out
Nov 18 01:53:08 zen /netbsd: st0(aha2:0:4:0): timed out
st0(aha2:0:4:0): timed out AGAIN
Nov 18 01:53:10 zen /netbsd: st0(aha2:0:4:0): timed out AGAIN

I pressed the unload button on the tape drive before leaving this
morning, it unloaded ok.

root@zen:141# mt -f /dev/rst0 status
st0: already open
mt: /dev/rst0: Device busy
Nov 18 10:09:32 zen /netbsd: st0: already open
root@zen:142# mt -f /dev/rst1 status
st1(aha2:0:5:0): timed out
Nov 18 10:09:50 zen /netbsd: st1(aha2:0:5:0): timed out
st1(aha2:0:5:0): timed out AGAIN
Nov 18 10:09:52 zen /netbsd: st1(aha2:0:5:0): timed out AGAIN
^C
^Z
^\
[BREAK]
Stopped at      cpu_Debugger+0x4:       leave
db> trace
cpu_Debugger(c0799ea0,c6e31900,c6e31900,0,c0799d60) at cpu_Debugger+0x4
comintr(c0779a00) at comintr+0xf4
Xintr4() at Xintr4+0x7a
--- interrupt ---
idle(c6e31900,65,c0250a9c,c6e31900) at idle+0x1b
bpendtsleep(c058b32c,118,c046eb80,65,0) at bpendtsleep
sys_select(c6e31900,c701af80,c701af78,246) at sys_select+0x300
syscall_plain(1f,1f,1f,1f,bfbfdcd8) at syscall_plain+0xa7
db> ps
 PID             PPID       PGRP        UID S   FLAGS          COMMAND    WAIT
 4917             340       4917          0 3  0x4006               mt   xscmd
 437                1        427      32768 3     0x7            taper  physio
db> 

taper is the one holding st0 open.

This is 1.6.1, I just rebooted - the DLT didn't probe, the DAT did and
is responding ok.

Perhaps it is a powersupply or similar issue?  It usually probes and
works for a while after being power cycled, but after an hour or so
just stops talking on the bus.

>If you do determine that the drive is faulty, you can usually get them
>serviced for significantly less than it costs to replace them with an
>equivalent unit.

DLT 4000's can be had on e-bay for about $150 at present, though most
of them look in worse condition than the one I have ;-)

--sjg