SCSI Reset - lbolt - which bus?

[Kim Kendall]

I've been getting scsi resets

scb->cdb: 12 00 00 00 80 00
SCSI: Resetting SCSI -- lbolt: 441347943, bus: 2
SCSI: Reset detected -- lbolt: 441347943, bus: 2

Trying to make sure of the bus (2)

I do an ioscan of the ext_bus - see attachment

Is the bus the 0/0/2/1 scsi bus (instance 2)???

[Brian M Rawlings · Accepted solution]

Hello. I have put > < marks around the Instance number that matches up to your question. "Bus" in a SCSI error refers to the instance number found in "ioscan". See below:

#ioscan -kfnC ext_bus

Class I H/W Path Driver S/W State H/W Type Description
===============================================================================
ext_bus 0 0/0/1/0 c720 CLAIMED INTERFACE SCSI C895 Fa
ext_bus 1 0/0/2/0 c720 CLAIMED INTERFACE SCSI C875 Ul
ext_bus >2< 0/0/2/1 c720 CLAIMED INTERFACE SCSI C875 Ul
ext_bus 16 0/2/0/0.2.23.0.0 fcparray CLAIMED INTERFACE FCP Array In
ext_bus 14 0/2/0/0.2.23.255.0 fcpdev CLAIMED INTERFACE FCP Device I
ext_bus 17 1/0/0/0.1.23.0.0 fcparray CLAIMED INTERFACE FCP Array In
ext_bus 15 1/0/0/0.1.23.255.0 fcpdev CLAIMED INTERFACE FCP Device I
ext_bus 8 1/2/0/0.8.0.255.0 fcpdev CLAIMED INTERFACE FCP Device I
ext_bus 9 1/4/0/0.8.0.255.0 fcpdev CLAIMED INTERFACE FCP Device I

This should help you narrow it down, but you need to know what devices are connected to this bus. It appears to be an FC array (?), which rules out some common SCSI RESET issues like terminators.

You have many controllers, so you may be able to narrow it down further by swapping the controller thus identified with another (requires a reboot, but you need to get to the bottom of this).

If this is a SAN, that's about as far as I can go with it, other than suggesting that SAN components (switches & cables) shouldn't be able to generate this (maybe a flaky cable could). SCSI protocol is handled over Fibre Channel as packetized messages, generated at the drive/array controller and the HBA/driver. In between, they are just "bits in a packet", so how would they get created by mere transport mechanisms? Much more likely to be caused by one of the two ends. If you can eliminate the HBA by swapping it with another identical one, you can then concentrate on the array/drive side.

Good luck, hope it was of some use.

--bmr

[Kim Kendall]

Thx Brian.... look a little closer, it's not a FC, but scsi c875 ui. But, that's what I wanted to confirm, that it was 0/0/2/1 that was considered "bus 2".

I wasn't sure that the "bus" = "instance".