Re: iSCSI: draft vs. 14 typos, suggestion, questions

["Julian Satran" <Julian_Satran@il.ibm.com>]

Mallikarjun,

We decided early one that we don't want to touch the notion of a task  set  because of the mess we are going to have to handle with TST=0.
What has changed?

Julo

"Mallikarjun C." <cbm@rose.hp.com> Sent by: owner-ips@ece.cmu.edu 07/11/2002 02:38 AM Please respond to "Mallikarjun C."

To:        "Julian Satran (Actcom)" <Julian_Satran@actcom.net.il>, <pat_thaler@agilent.com>, <tomasb@s3group.cz>         cc:        <ips@ece.cmu.edu>, <ivan.pavelka@s3group.com>         Subject:        Re: iSCSI: draft vs. 14 typos, suggestion, questions

Sorry, I'm catching up late on email....

I see that the latest wording on this is -

When the session timeout (the connection state timeout for the last failed connection) happens on the target,
it takes the following actions:
- Resets or closes the TCP connections (closes the session).
- Aborts all Tasks in the task set associated with the session.

I think the second bullet is incorrect.  While the previous wording had the
issue that Pat identified (that there may be one task set for each LU the session
can get to), the new wording still has that issue and in addition, also implies that
all tasks in the task set need to be terminated.  However, a LU may maintain only one
task set for multiple initiators (if TST=0 in control mode page) in which case, the
current wording implies that all tasks from other initiators need to be terminated
as well on one session timeout - which is not intended.

I am beginning to think that the notion of "task set" is beyond iSCSI and it's best
not to refer to it here.

I suggest the following text instead -

When the session timeout (section 4.3.5) happens on the target, it takes the following actions:
- Resets or closes the TCP connections (closes the session).
- Terminates all active tasks that were allegiant to the connection(s) that constituted the session.

Thanks.
--
Mallikarjun

Mallikarjun Chadalapaka
Networked Storage Architecture
Network Storage Solutions
Hewlett-Packard MS 5668
Roseville CA 95747
cbm@rose.hp.com

----- Original Message -----
From: "Julian Satran (Actcom)" <Julian_Satran@actcom.net.il>
To: <pat_thaler@agilent.com>; <tomasb@s3group.cz>
Cc: <ips@ece.cmu.edu>; <ivan.pavelka@s3group.com>
Sent: Thursday, July 04, 2002 9:08 PM
Subject: Re: iSCSI: draft vs. 14 typos, suggestion, questions


> I will do the text change (I hope it makes it easier for at least one reader
> is not worse for anybody else).
> The reason why we choose the wording was that SAM associates the task set
> with an initiator by the session is wording is equivalent.
>
> Julo
> ----- Original Message -----
> From: <pat_thaler@agilent.com>
> To: <Julian_Satran@il.ibm.com>; <tomasb@s3group.cz>
> Cc: <ips@ece.cmu.edu>; <ivan.pavelka@s3group.com>
> Sent: Wednesday, July 03, 2002 9:38 PM
> Subject: RE: iSCSI: draft vs. 14 typos, suggestion, questions
>
>
> > See comment below. Pat
> > -----Original Message-----
> > From: Julian Satran [mailto:Julian_Satran@il.ibm.com]
> > Sent: Wednesday, July 03, 2002 9:20 AM
> > To: Tomá? Bartu?ek
> > Cc: ips@ece.cmu.edu; ivan.pavelka@s3group.com
> > Subject: Re: iSCSI: draft vs. 14 typos, suggestion, questions
> >
> >
> >
> > *** Section 6.11 : QUESTION
> > Draft says:
> >   When the session timeout (the connection state timeout for the last
> >   failed connection) happens on the target, it takes the following
> >   actions:
> >
> >     - Resets or closes the TCP connections (closes the session).
> >     - Aborts all Tasks in the task set for the corresponding initi-
> >       ator.
> >
> > What does the "corresponding initiator" mean? We think (:-)), that
> > there is only one initiator for the session. The only possible
> > explanation we see for that paragraph is, that the target should
> > abort also other tasks of the same in __other__ sessions, but
> > why?
> > +++ your interpretation is correct - the statement means the initiator
> that "owned" the session.
> > Are you suggesting other wording?
> > ++++
> > <PAT> "Aborts all task sets associated with the session"
> > (Task set was changed to plural because task set is defined as an I-T-L
> nexus and there may be muliple ones in the session. <PAT>
> >
>
>