A persistent reservation question

Qin Zhang qzhang at ariodata.com
Tue Jun 29 09:06:33 PDT 2004


* From the T10 Reflector (t10 at t10.org), posted by:
* "Qin Zhang" <qzhang at ariodata.com>
*
Get it. Thanks.

-----Original Message-----
From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf Of Casado,
Reyes
Sent: Tuesday, June 29, 2004 2:28 AM
To: Qin Zhang; MEHROTRA, VIVEK (STSD); t10 at t10.org
Subject: RE: A persistent reservation question

* From the T10 Reflector (t10 at t10.org), posted by:
* "Casado, Reyes" <reyes.casado at hp.com>
*
Hello,

My understanding is that the Reserve command should fail with
RESERVATION CONFLICT. 

Reserve/Release commands are obsolete in SPC-3. They are described in
SPC2rev20, and I believe all rules specified there apply unless stated
differently in SPC3. I am quoting from SPC-2 section 5.5.1:

"The two methods [Reserve/Release and Persistent Reservation] are
prevented from creating conflicting and undefined interactions using
RESERVATION CONFLICT status in the following manner. If a logical unit
has executed a PERSISTENT RESERVE OUT command with the REGISTER or the
REGISTER AND IGNORE EXISTING KEY service action and is still registered
by any initiator, all RESERVE commands and all RELEASE commands
regardless of initiator shall conflict
and shall terminate with a RESERVATION CONFLICT status. If a logical
unit has been reserved by any RESERVE
command and is still reserved by any initiator, all PERSISTENT RESERVE
IN and all PERSISTENT RESERVE
OUT commands shall conflict regardless of initiator or service action
and shall terminate with a RESERVATION
CONFLICT status."

A different matter is if the device has been persistently reserved (with
PR Out commands.) Then the restriction in SPC3 section 5.6.2 applies.

Regards,
Reyes Casado
Hewlett Packard

-----Original Message-----
From: owner-t10 at t10.org [mailto:owner-t10 at t10.org]On Behalf Of Qin Zhang
Sent: Tuesday, June 29, 2004 1:43 AM
To: MEHROTRA, VIVEK (STSD); t10 at t10.org
Subject: RE: A persistent reservation question


* From the T10 Reflector (t10 at t10.org), posted by:
* "Qin Zhang" <qzhang at ariodata.com>
*
Hi Vivek,

As my understanding, the RESERVE command should success if no
reservation has been made under the condition you described. (RESERVE
command does not have key).

For SPC-3r19 section 5.6.2, the host can still issue read/write command
under the two described condition, in that sense, the RESERVE command
return GOOD status is understandable even though it is persistently
reserved.

Thanks,
Qin
ArioData Networks
-----Original Message-----
From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf Of
MEHROTRA, VIVEK (STSD)
Sent: Monday, June 28, 2004 2:14 PM
To: t10 at t10.org
Subject: RE: A persistent reservation question

* From the T10 Reflector (t10 at t10.org), posted by:
* "MEHROTRA, VIVEK (STSD)" <vivek.mehrotra at hp.com>
*
Hi Rob,

After I look in the SPC-3 section 5.6.2, it still does not answer my
question. Actually I wanted to know, if a host has registered
persistently by sending SPC-3 PERSISTENT RESERVE OUT command with
REGISTER or REGISTER AND IGNORE EXISTING KEY service action, and then
the device device receives SPC-2 RESERVE command (with the same key)
|from this host, what should the device do in this case ?

SPC-3r19 section 5.6.2 says "A RESERVE(6) or RESERVE(10) command shall
complete with GOOD status, but no reservation shall be established and
the persistent reservation shall not be changed ...". I am not clear
about what does the part "... and the persistent reservation shall not
be changed" means. Does this mean that we get a RESERVE (SPC-2) command
after the host is PERSISTENTLY RESERVED (using SPC-3 style reservation).
If this is true, then my question is different.

Please advice, on what I am thinking is correct or incorrect.

Thanks in advance

Vivek Mehrotra

-----Original Message-----
From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf Of Elliott,
Robert (Server Storage)
Sent: Thursday, June 24, 2004 7:04 PM
To: t10 at t10.org
Subject: RE: A persistent reservation question


SPC-2, which defined both commands, requires that RESERVATION CONFLICT
status be returned.

SPC-3, which considers RESERVE/RELEASE obsolete, allows them to be
treated as NOOPs in certain cases (see section 5.6.2 of revision 19, all
the revisions of 03-232, 02-483, and 02-231).

-- 
Rob Elliott, elliott at hp.com 
Hewlett-Packard Industry Standard Server Storage Advanced Technology 
https://ecardfile.com/id/RobElliott 


-----Original Message-----
From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf Of
MEHROTRA, VIVEK (STSD)
Sent: Thursday, June 24, 2004 3:23 PM
To: t10 at t10.org
Subject: A persistent reservation question


Hi Folks, 
What is the best approach to handle the scenario, when a device is
registered for persistent reservation using SCSI 3 style PERSISTENT
RESERVE OUT command (expecting a same style command with RESERVE service
action), but receives a old style RESERVE command from the initiator.
I didn't see SCSI specs talking about this anywhere ! 
Thanks in Advance 
Vivek Mehrotra 
*
* 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
*
* 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 mailing list