Subject: Re: more st wierdness
To: None <rkboni@concentric.net>
From: Matthew Jacob <mjacob@ns.feral.com>
List: current-users
Date: 11/07/1997 20:43:08
>From rkboni@doppelganger.old-oak.net  Fri Nov  7 20:36:42 1997
>Seeing all these messages about st.c reminds me of a problem I have on my 
>machine... Basically, 9/10 tape dumps die unexpectedly (it looks like the
>tape drive may have taken offense to something the SCSI code said and tries
>to reset itself??).

9 of 10? Really. Hmm.. Badness.

>
>My machine info: (abbreviated)
>
>NetBSD 1.3_ALPHA (DOPPELGANGER) #0: Tue Oct 28 14:16:16 EST 1997
>cpu0: family 5 model 2 step 6
>cpu0: Intel Pentium (P54C) (586-class)
>real mem  = 33157120
>avail mem = 29179904
>using 430 buffers containing 1761280 bytes of memory
>mainbus0 (root)
>isa0 at pcib0
>aic0 at isa0 port 0x340-0x35f irq 11	[this is a Compaq AIC-6260 card]
>scsibus0 at aic0: 8 targets
>sd0 at scsibus0 targ 0 lun 0: <FUJITSU, M1606S-512, 6226> SCSI2 0/direct fixed
>sd0: 1041MB, 3457 cyl, 6 head, 102 sec, 512 bytes/sect x 2131992 sectors
>st0 at scsibus0 targ 3 lun 0: <EXABYTE, EXB-85058SQANXR1, 06S1> SCSI2 1/sequential removable
>st0: density code 0x0, 1024-byte blocks, write-enabled

Okay, you've got a fairly standard 8505- it's coming up in fixed block
mode. Yick.

>
>The problem: 
>I try and dump some partition to the tape, after some time all the drive 
>lights on the exabyte go on, the tape rewinds after some thought, and dump
>is either convinced that the tape's ended or sits frozen in D+ state. I
>*know* the tape's not done (even a 500Meg partition should fit easily
>on a compressed 112m 8mm tape), since all the partitions I've tried have
>been fairly small (vs. tape size).

It won't even have to be compressed. In 8200 mode you'll get, what is it- 1GB,
in 8500 mode you'll get 5GB. Compression should close to double that.

>
>I just tried this, and dump actually got 95% done (which is more than I
>usually get), before the "tape lights all go on, tape rewinds, dump asks
>for another tape" happened.  Here's the messages from the tail of the 
>kernel message buffer:
>
>st0(aic0:3:0):  Check Condition on opcode 0x0
>    SENSE KEY:  Unit Attention
>     ASC/ASCQ:  Power On, Reset, or Bus Device Reset Occurred
>
>st0(aic0:3:0):  Check Condition on opcode 0x0
>    SENSE KEY:  Not Ready
>     ASC/ASCQ:  Logical Unit Is in Process Of Becoming Ready
>
>[...above repeated five or six times...]
>
>Occasionally, I've also seen "st0(aic0:3:0) timed out" and a few days
>ago I saw some really scary looking message from aic0:3:0 (something
>about some unexpected SCSI transition), which unfortunately I can't find
>in my logs.
>
>Any ideas appreciated.

The above check conditions have to do (likely) with the AIC driver
either resetting the SCSI bus or resetting the device. The tape lights
all going on is from a reset condition. What are your dump arguments?
Large blocksize? What's the cabling setup?

-matt