Need help with Syslog message "SCSI: Unexpected Disconnect"

[Scott Frye_1]

Received the bellow message in syslog. I'm not sure where to begin to detemine what caused this. Can anyone help?

Thanks

Scott

Jan 3 14:52:15 h0306n1 vmunix: SCSI: Unexpected Disconnect -- lbolt: 228560593, dev: cb062002, io_id: 60003c3
Jan 3 14:52:15 h0306n1 vmunix: SCSI: Unexpected Disconnect -- lbolt: 228560594, dev: cb062002, io_id: 60003c3

[Steven E. Protter]

Few things usually cause this:

1) Cable disconnect.
2) Drive failure.
3) Swapping out a hot swap drive.

The good news is with items 1 and 3 you can simply reconnect and next time you boot the problem will go away.

If a drive has failed you must replace it.

You can try and use xstm mstm or cstm to run an exercize on disk devices and see which of the above problems exists.

SEP

[Eric Antunes]

Hi Scott,

Check also this post about lbolt:

http://forums1.itrc.hp.com/service/forums/parseCurl.do?CURL=%2Fcm%2FQuestionAnswer%2F1%2C%2C0x9d086d96588ad4118fef0090279cd0f9%2C00.html&admit=716493758+1104847801286+28353475

Regards,

Eric Antunes

[steven Burgess_2]

Hi Scott

Looks like a disk issue, my suggestion is you take a peek at /dev/dsk/c6t2d0

diskinfo /dev/rdsk/c6t2d0

get the lv information to check for any stale extents

for lvol in $(pvdisplay -v /dev/dsk/c2t0d0 | sed -n '/LV Name/,/---/p
' | egrep -v '(LV|Phy|^$)'| awk '{print $1}')
do
lvdisplay -v $lvol | grep -i stale
done

HTH

Steve





HTH

Steve

[Scott Frye_1]

Steve,

I do not have a disk c6t2d0

[Eric Antunes]

Hi,

Execute the following to see witch device is bad:

ioscan -fn|grep SCSI

[Paul Torp]

look in "mstm"

find the scsi device..

format the log and watch it..

u should see the eroors there..

-paul

[Scott Frye_1]

I definitely see something in mstm:
0/8/0/0.1.0 SCSI Disk (HPDVD-ROM) Information Warning

now, can I tie this back to my syslog message?

[Zeev Schultz]

I don't think its about cd/dvd-rom on path 0/8 issue.It has SCSI id = 1 and your dev is translated like this:

dev: cb062002
cb = device major number, equals 203 in dec,
lsdev | grep 203 gives -- sctl, meaning SCSI controller driver.
06 = run 'ioscan -fnCext_bus', see in the second column for number 6.
20 = SCSI id 2 , lun id 0.
For example it could be cXt2d0 , replace X with
number you find from ioscan -fnCext_bus.
As to message itself it could be many things,
from SCSI cable,termination,adapter issues to patches and PV timeouts (usually when connected to slow old disks and complex disk arrays...)

[Scott Frye_1]

Zeev,

I ran the commands you provided (THANKS!) and got:
ext_bus 6 1/2/0/0 c720 CLAIMED INTERFACE SCSI C895 Ultra2 Wide LVD

So am I to assume it is something with this? I do not have a device called c6t2d0...

Thanks again

Scott

[Zeev Schultz]

Well in that case it could be either some device connected to this SCSI adapter and not
seen in the ioscan output (I guess you tried to run 'ioscan -fn' for full list?) or SCSI adapter issue (fixed probably by patches) but the last is less obvious since HP-UX messages are of "Unexpected Disconnect" kind...

Meaning,something was/is connected there but not probably seen by ioscan. You can also scroll to the beginning of /var/adm/syslog/syslog.log - it does an 'ioscan ' like listing when server boots.Or even OLDsyslog.log to see what was connected there before.

And if nothing looks connected there I'd first go and see with my own eyes that it's indeed so...

[Scott Frye_1]

OK, I believe with this output that my problem is with c6t7d0. Would I be correct?

ioa 1 1 sba CLAIMED BUS_NEXUS System Bus Adapter (803)
ba 8 1/0 lba CLAIMED BUS_NEXUS Local PCI Bus Adapter (782)
ba 9 1/2 lba CLAIMED BUS_NEXUS Local PCI Bus Adapter (782)
ext_bus 6 1/2/0/0 c720 CLAIMED INTERFACE SCSI C895 Ultra2 Wide LVD
target 13 1/2/0/0.7 tgt CLAIMED DEVICE
ctl 6 1/2/0/0.7.0 sctl CLAIMED DEVICE Initiator
/dev/rscsi/c6t7d0

[Stephen Keane]

Might not be relevent to your situation, but I've seen similar SCSI error messages when a slower devices (such as a tape drive/ATL) is connected to the same SCSI bus as fast devices (such as disks).

[Scott Frye_1]

Might be. If my guess is correct and the HW path of 1/2/0/0 is the problem, I have a LTO tape device SCSI attached.

[Uwe Zessin]

It is still a bit strange, because a SCSI disconnect itself isn't an error. A SCSI target (disk, tape, ...) receives a request from the initiator (computer) and then can disconnect to release (free up) the bus for other operations. When it has e.g. received the data from the medium, it requests the bus, addresses the initiator and sends the data to it.

[Scott Frye_1]

I believe I found my problem HW. It looked like HW path 1/2/0/0 was the problem. This path is associated with an LTO tape device connected via SCSI.

Thanks for the replies!