Re: iSCSI: Question: ISID Rule?

["John Hufferd" <hufferd@us.ibm.com>]

Not exactly.

There is a CID in the equation also.  if TSID is not 0, and assuming other
things match, the CID will be used to determine if this is an implicit
logout and re-login for that connection.  If the CID is unique within the
session, then it will start a new session, else, it will log out the old
connection and create a new one.  Then of course, you need to use the Task
Management Request PDU, on the new connection to move the allegiance of the
existing tasks (one at a time) from the old connection to the new
connection.



.
.
.
John L. Hufferd
Senior Technical Staff Member (STSM)
IBM/SSG San Jose Ca
Main Office (408) 256-0403, Tie: 276-0403,  eFax: (408) 904-4688
Home Office (408) 997-6136, Cell: (408) 499-9702
Internet address: hufferd@us.ibm.com


Luben Tuikov <luben@splentec.com>@ece.cmu.edu on 03/07/2002 02:32:42 PM

Please respond to iSCSI <ips@ece.cmu.edu>

Sent by:    owner-ips@ece.cmu.edu


To:    iSCSI <ips@ece.cmu.edu>
cc:
Subject:    Re: iSCSI: Question: ISID Rule?



"KRUEGER,MARJORIE (HP-Roseville,ex1)" wrote:
>
> It's the SCSI device that is the comparison point, so you are comparing
the
> iSCSI node name of the two entities to determine if they are
"the same".

Then this implies that a session is uniquely identified by (InitiatorName,
TargetName).

Then for a new connection at the target:

 i) TSID!=0 provides implicit uniqueness,
    since a new connection will be added,
    if the session already exists.

ii) TSID=0, implies that there does NOT exist
    a session with the given TargetName, InitatorName
    and ISID given in the Login Request PDU, else close the
    connection.

Is this true?

--
Luben