Re: iscsi rev 7.93 : reject pdu.

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

Santosh,

Since that is my handiwork, :-), let me answer that.  This 
also brings up a good opportunity to explain something that
hasn't been discussed on ips.  

As captured by the change log, format errors are now mandated 
to cause session failures.

The rationale is that format errors basically mean serious, 
reproduceable coding problems and ought to be escalated to 
the highest level - besides the fact that they may have 
messed up the framing sync.  It was also felt that diagnostics 
on Format errors outlived their purpose beyond the early 
development stage of the protocol, since we do not expect
to see these errors in production implementations.  This change 
is generally part of the protocol simplification efforts, 
and was proposed by Mark (Bakke).  Removing the "first bad byte" 
field in Reject is merely a side-effect of the above change, 
since this field explicitly was meant to reject format errors.

Let me also just comment on the rationale behind leaving the
protocol error handling as it is.  While "protocol errors" are 
problems, they may not be the serious reproduceable menace that 
format errors are, neither do they mess up framing.

Thanks.
-- 
Mallikarjun 


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


Santosh Rao wrote:
> 
> Julian,
> 
> Could you please clarify why the "first bad byte" field in the reject
> pdu has been removed ? I can find nothing in the change log that states
> this change, nor can I remember any discussion on ips asking for this
> change.
> 
> Regards,
> Santosh