Re: iSCSI:regarding connection shutdown

["Mallikarjun C." <cbm@rose.hp.com>]

Sastry,

Thanks for the close review.

As I mentioned in an earlier response to Nandakumar 
on this subject, the "graceful connection shutdown"
is referring to a graceful TCP connection shutdown.

With that said, the usage of "and" in the quoted sentence
was meant to underline the point that a connection shutdown
must be initiated only after the Logout dialogue is cleanly
complete - a mere `no active tasks` is not sufficient to
drop the connection - since iSCSI connection and session 
structures would hang around (until timeout, waiting for
session continuation) if a clean logout is not done.  This
was sought to be actively discouraged.

On thinking more about it, I think perhaps this sentence
says more than is necessary.  It's completely legitimate
for initiators to Logout("close") a connection to implicitly 
abort all active tasks.  If one merely ensures that the 
connection is *not* in the FFP to start a TCP connection 
close, that is completely adequate (since the active tasks 
if any, were automatically taken care of in the Logout process).  
So, I would actually suggest wording like -

 "A graceful connection shutdown MUST be initiated only
  when the connection is not in full-feature phase."  


We already have wording there that takes care of outstanding
tasks -"Connection termination with outstanding commands
may require recovery actions."

Also, I noticed that the section doesn't consistently use 
"termination" - we need to address that aspect as well.

Please comment if something isn't clear.

Thanks.
-- 
Mallikarjun 


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


Subrahmanya Sastry K V wrote:
> 
> Hi all,
> 
>  Section 1.2.6 of the iscsi draft 07 says :
> "Graceful connection shutdowns MUST only occur when there are no
> outstanding tasks that have allegiance to the connection or when the
> connection is not in full-feature phase."
> 
> But section 2.2.6 of the iscsi draft 08 and 09 say :
> "Graceful connection shutdowns MUST only occur when there are no
> outstanding tasks that have allegiance to the connection and when the
> connection is not in full-feature phase."
> 
> The sentence mentioned in draft 07 sounds clear as it talks about the
> two possibilities of graceful connection shutdown.
> But the same sentence in draft 08 and 09 suggests a different meaning
> because of 'and'. Is there any reason why this has been changed to
> 'and' ?
> 
> I'm asking this because this change has not been recorded in the
> change log. Please clarify .
> 
> Regards,
> Sastry
> 
> --
> Subrahmanya Sastry K V
> Member Technical Staff
> HCLTech, Chennai, INDIA
> http://san.hcltech.com
>