unhooked disks, ran "rmsf". Rebooted_2x Still get lbolt error

[Steve Post]

hpux K class server running 11.0.

I removed some unused disks from the server on path 8/0 and 8/4. I made sure they were unused. I unplugged the scsi cable. The scsi card should be self terminating. I rebooted without the disks.
It comes up fine. But the devices are still expected.
I ran "rmsf -H 8/0" on the boxes to remove the disks from where they were.
Rebooted again.
Still get lbolt errors.
I run ioscan -f and do NOT see the unhooked devices.
Everything looks fine.
But when I run ioscan, or dmesg. I get lbolt errors.

SCSI: Unexpected Disconnect -- lbolt: 33941, dev: cb01e002, io_id: 100002c

You know, it's been about a day without the disks plugged in. I already rebooted twice. You would think the machine would EXPECT the disconnect by now.

Any idea what I missed?
Of course I don't seem to find the step by step instructions for a obviously routine sysadmin task (ie removing hardware).

[Steven E. Protter]

Shalom,

Perhaps the disks are configured into a volume group.

strings /etc/lvmtab

mv /etc/lvmtab /etc/lvmtab.old
vgscan -a

You'll still have to boot to clear the lbolt.

SEP

[Tim Nelson]

Are you sure you have the correct device(s).

cb01e002 would be /dev/dsk/c1t14d0


do an lssf and the device file. If a valid HW address comes back then the HW is still there, otherwise you have gremlins :)

[Steve Post]

I moved lvmtab, and ran vgscan -a. It didn't help. It was worth a try.

I found lots of bogus device files still in /dev/dsk/ and /dev/rdsk. I got rid of all of them, then ran "insf -e" to bring back any devices that should be there. (Of course I didn't get rid of the vg00 disks or the CDROM).

I rebooted. I still got errors.
Jun 25 14:04:51 ob2b vmunix: scb->cdb: 12 00 00 00 80 00
Jun 25 14:04:52 ob2b vmunix: SCSI: Resetting SCSI -- lbolt: 33245, bus: 1
Jun 25 14:04:52 ob2b vmunix: SCSI: Reset detected -- lbolt: 33245, bus: 1
Jun 25 14:04:57 ob2b vmunix: SCSI: Unexpected Disconnect -- lbolt: 33745, dev: c
b010002, io_id: 100001f
Jun 25 14:04:57 ob2b vmunix: SCSI:
to..............
Jun 25 14:04:59 ob2b vmunix: SCSI: Unexpected Disconnect -- lbolt: 33864, dev: c
b01f002, io_id: 100002d

As they say in the 1930's..... No Soap.

[OldSchool · Accepted solution]

it's now talking about

c1t0d0 and c1t15d0

where *were* the disks you pulled? I think I'd be looking real hard at the cable / terminator issues.

[Steve Post]

I didn't have c1t15d0. I had c1t0d0.

Yep. You got it. Terminators. I was going to check it....eventually.

I ran "ioscan". I got lbolt errors.

I threw terminators on the cards of every spot I pulled a cable from. I ran ioscan and didn't get any more lbolt errors. I rebooted, and no more problem.

I knew it was something silly.

The cause is always the last thing you look at. But of course it is. Why would you keep looking if you already have your solution?