TAS bit or other mechanisms
Elliott, Robert (Server Storage)
Elliott at hp.com
Wed Feb 6 09:06:11 PST 2008
* From the T10 Reflector (t10 at t10.org), posted by:
* "Elliott, Robert (Server Storage)" <Elliott at hp.com>
Serial Attached SCSI (SAS) has several tools to deal with that problem:
1. I_T NEXUS RESET task management function. If initiators are polite enough
to send this instead of hard reset, they won't mess up other initiators.
Some alternatives are
a) a series of ABORT TASKs
b) an ABORT TASK SET
c) a LOGICAL UNIT RESET (if the logical unit is only being used by one
but they may not be powerful enough to solve whatever problem the reset is
trying to fix.
For Fibre Channel, some implementations applied creative license to the
standards and interpreted the obsolete TARGET RESET like I_T NEXUS RESET,
only affecting the initiator that send the reset.
2. Broadcast (Asynchronous Event). Provide notice to an initiator that a
unit attention has occurred (e.g. because of a hard reset).
In protocols that lack asynchronous event notification, QUERY TASK SET and
QUERY TASK can be used to confirm that commands that seem to have timed out
have really disappeared.
3. QUERY UNIT ATTENTION task management function. Determine which unit
attention condition caused Broadcast (Asynchronous Event) or what event
caused all the commands to disappear (per QUERY TASK SET) without disrupting
normal command traffic and possibly losing an unrelated unit attention
If an initiator determines that a reset occurred (29h/nnh additional sense
code), it knows its commands have been aborted and need not wait for them to
time out. It can skip its normal command timeout error handling procedure of
issuing another hard reset, avoiding the ping-pong reset problem.
Pertinent proposals include: 01-015, 06-024, 06-044, 07-067, 07-144, 07-224
Rob Elliott, elliott at hp.com
Hewlett-Packard Industry Standard Server Storage Advanced Technology
> -----Original Message-----
> From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf
> Of Alan Gess
> Sent: Tuesday, February 05, 2008 4:49 PM
> To: t10 at t10.org
> Subject: TAS bit or other mechanisms
> * From the T10 Reflector (t10 at t10.org), posted by:
> * "Alan Gess" <agess at istor.com>
> One issue that I have come across over the years with SCSI is
> the "ping
> pong" of hard resets with a device that has multiple
> initiators issuing
> tasks to it. This occurs when one initiator reset the devices while
> another initiator has tasks outstanding to the same device. I am sure
> you all know about this issue.
> I see a TAS (Task Abort Status) bit was added to help resolve
> this issue
> but after testing four different manufactures of drives, only one
> supports it.
> My question: Is there another mechanism in place to prevent
> this or does
> anyone have any ideas on how to overcome this issue?
> Alan Gess
> iStor Networks, Inc.
> agess at istor.com
> * For T10 Reflector information, send a message with
> * 'info t10' (no quotes) in the message body to majordomo at t10.org
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org
More information about the T10