Ladder Diagrams for Error Recovery update, T10/00-137r4

Robert Snively rsnively at Brocade.COM
Thu Jun 1 10:27:19 PDT 2000


* From the T10 Reflector (t10 at t10.org), posted by:
* Robert Snively <rsnively at Brocade.COM>
*

My interpretation of the intent of the working group as expressed in the
May meeting is in T10/00-150r4.  As I understood it:

11.6 Compaq CommentID:6 (Technical)

Title: Remove paragraph

Comment: For out of order, the error detection and recovery procedures 
are different. Even for in-order, recovery is different-i.e., 
reuse/non-reuse of SEQ_CNT for Classes 3 and 2 respectively for data.

Response:

This requires further study. If true, this requirement is 
undesirable, and would be sufficient to reinstate the prohibition 
on out-of-order delivery.

Carl Zeitler has prepared a series of ladder diagrams, presented in 
T10/00-173r3 and subsequent revisions, that tell how FCP-2 recovery 
operates with out-of-order delivery. In the discussion
on May 15, 2000, the committee agreed that the diagrams should be 
carried as an informative annex. A note should be included in the text 
that indicates that FCP-2 requires in-order delivery to use the 
documented recovery mechanisms. However, with the assumption that
continually increasing sequence count and fully implemented recovery 
qualifier behavior are both available, the recovery mechanisms indicated 
in the annex created by Carl may optionally be performed whether the 
transfer is in order or out of order.  Additional work will be done in 
future revisions of FCP to formalize the definition of out-of-order
recovery.

Carl's proposal to respond to multiple failures and failures of the 
retry process with an ABTS-LS was accepted.

There was a consensus that the recovery process should be the same for 
out-of-order and in-order cases. Detection may be improved and made more 
timely by exploiting the acknowledged classes of service.


>  There were concerns expressed at the May meeting that 
>  out-of-order was
>  exposed and risky and that it should be put in FCP-3 or in a 
>  separate Annex.
>  I would urge those who harbor those feelings to scrutinize 
>  these diagrams to
>  be able to express more succinctly their objections
>  to having them in a single Annex.  
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org




More information about the T10 mailing list