Re: Proposed Connection Recovery Additions for Draft 03

[julian_satran@il.ibm.com]

Mark,

Excellent job.  It described mostly what I had in mind.

With your permission I will use even your wording whenever possible.

I thought I had the AckStatRN (MaxStatRN) but I could not piggyback it on
the right messages
- the headers are too short.

Also if you think about a layered implementation - Costa's proposal to use
a separate
PDU makes sense as those are generated by iSCSI alone and convey values of
iSCSI registers.

I would like to hear what everybody has to say about it. The alternative
looks like increasing the
header (48?)

I would also like to clarify for implementors that device-recovery is to be
based on Initiator Tag as it
it is the only identifier unchanged for the life of the task while the two
numberings are only for
the delivery mechanism (iSCSI).

Regards,
Julo