Message-ID: <AEC4671C8179D61194DE0002B328BDD26A7DF0@ATLOPS>
From: Eddy Quicksall <eddy_quicksall@ivivity.com>
To: wrstuden@wasabisystems.com, Eddy Quicksall <eddy_quicksall@ivivity.com>
Cc: ips@ece.cmu.edu
Subject: RE: iSCSI: clearing SCSI objects
Date: Tue, 6 May 2003 18:52:39 -0400 
MIME-Version: 1.0
Content-Type: text/plain
Sender: owner-ips@ece.cmu.edu
Precedence: bulk

Perfect ... thanks. I read that section a few times but must have had
blinders on when I hit that paragraph.

Eddy

-----Original Message-----
From: wrstuden@wasabisystems.com [mailto:wrstuden@wasabisystems.com] 
Sent: Tuesday, May 06, 2003 6:33 PM
To: Eddy Quicksall
Cc: ips@ece.cmu.edu
Subject: RE: iSCSI: clearing SCSI objects

On Mon, 5 May 2003, Eddy Quicksall wrote:

> If you have persistent reserves, you will lose them.
>
> Keeping them around in the absence of persistent reserves is for full
> compliance and it is easy to do.

Ok, I re-read section 6.3.4, and see an easy out. You don't need the LRU
list you were talking about. ;-)

Make use of the paragraph that starts, "IF the logical unit does not
retain state..." and you're fine. Just send the UA/POWER ON OCCURRED and
you're set.

If there is a persistent reservation, you are retaining state so you have
to remember if you logged out due to nexus loss. But that is just one bit
on top of all the other weight of a reservation, and also you can note the
resource shortage when a future initiator attempts to make a new
registration (or reservation).

So you don't need the LRU list.

Take care,

Bill