Minutes o f SCSI Protocol Study Group

James Smart USG smart at zk3.dec.com
Mon Sep 29 07:58:05 PDT 1997


* From the T10 (formerly SCSI) Reflector (t10 at symbios.com), posted by:
* James Smart USG <smart at zk3.dec.com>
*

George,

Not having a lot of background on these discussions, I want to make sure
that the following is not impacted by what you describe below (actually
to make sure the situation is not made worse than it already is today).

We are finding that multiple input ports to a SCSI lun need to preserve
and present this lun as a single task queue, independent of the input
port that the CDB was received on. When I read you note below, I'm
viewing it in the light that the Initiator may or may not be on the
same scsi bus.

In the scenarios in which we are concerned, there are 1 or more hosts
attached to a multi-ported device via one or more adapters (not necessarily
on the same scsi bus), and ios are concurrently issued over the different
ports. In these scenarios, device recovery and interpretation of SENSE data
is much more easily and correctly handled if SCSI Bus Resets, BDRs, and like
conditions where a LUN's task set is affected is also reflected to the
LUN's task set on the alternate port(s). This is especially true under
high-availability cases where io sent on a failed adapter need to be
terminated to ensure no data corruption.

We would like to see the ability for a host or set of hosts to :

  a) Issue commands concurrently to the storage device through
     any number of initiators (which may be on the same or different
     scsi busses)

  b) Issue a single "CLEAR TASK SET", or "TARGET RESET" issued
     on one of the ports, also clears all related tasks on the LUN(s)
     from the other ports.

  c) Since we are still in a parallel scsi world, we would like to see
     a single SCSI Bus Reset, or Bus Device Reset issued on one port
     affect the LUN task sets on all ports.


James Smart


----- Begin Forwarded Message

Date: Thu, 25 Sep 1997 09:24:06 -0400
From: George Penokie <gop at us.ibm.com>
To: <t10 at symbios.com>
Subject: Minutes o f SCSI Protocol Study Group

* From the T10 (formerly SCSI) Reflector (t10 at symbios.com), posted by:
* George Penokie <gop at us.ibm.com>
*
---------------------- Forwarded by George Penokie/Rochester/IBM on 09/25/97
08:20 AM ---------------------------


George Penokie
09/23/97 05:41 PM

To: owner-t10 at symbios.com @ internet @ IBMLMS01
cc:
From: George Penokie/Rochester/IBM @ IBMUS
Subject: Minutes o f SCSI Protocol Study Group

The minutes of the SCSI Protocol Study Group state the following:

'Ralph Weber offered an alternate proposal that would  define a task set per
initiator.  This has the beauty of maintaining the task set rules.  The
group seemed to favor this direction so Gene agreed to modify his proposal
in this manner for the next meeting. He will add a 3-bit field to the
Control Mode page and define a value for task set per logical unit and a
task set per initiator.'

The statement 'task set per initiator' in the above paragraph is not correct.
It should be 'task set per initiator/logical unit'. The definition of 'task set
per initiator' is that there is one task set per initiator and all the LUNs
within the target are also within the task set boundaries. So if you have
two initiators and two logical units you would have two queues.

The definition of 'task set per initiator/logical unit' is that there is one
task set per I_T_L nexus. So if you have two initiators and two logical units
there would be four queues. This is, I believe the behavior you want.

This is all described better in 92-141r8.

Bye for now,
George Penokie


*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com


----- End of Forwarded Message





James Smart                                            email: smart at zk3.dec.com
Alpha UNIX Systems - Digital Equipment Corp.           Phone: 603-881-2472
110 Spit Brook Road,    Nashua, NH  03060              Fax  : 603-881-2257
Disclaimer: My statements and opinions reflect my own thinking and are not
   necessarily reflective of any employer or association.

*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com




More information about the T10 mailing list