FCP-2 resource recovery question

Binford, Charles cbinford at lsil.com
Wed Jul 21 07:06:28 PDT 1999


* From the T10 Reflector (t10 at symbios.com), posted by:
* "Binford, Charles" <charles.binford at lsil.com>
*
Bob,  the "topology factor in the definition" is clear (in my opinion) for
PLDA, but a bit misleading for FLA.  Specifically the phrase "for a specific
SCSI Initiator" leads my train of thought away from FAN (which gets all
public initiators in one ELS).  I think a slight wording change would clear
that up.  I'd suggest something along the following:

"RR_TOV is the minimum time a SCSI Target shall wait for address
authentication from each SCSI Initiator in a private loop, or address
authentication from the FL_Port in a public loop following the completion of
the Loop Initialization Protocol (i.e., the receipt of CLS while in the
OPEN-INIT state)." 

Two items in the above suggested words I like to emphasize.  First calling
out the individual authentication (e.g ADISC/PDISC) verses the group
authentication (FAN).  I don't care if ADISC/PDISC / FAN is used or generic
terms like I used, I just think it is clearer to call the two styles of
authentication out separately.  Second, I use the term 'address
authentication' instead of 'exchange authentication'.  I don't feel real
strong on this one, but I do think it is a more accurate term of what we are
doing.

Next, in an N_Port to F_Port environment (direct switch attach) an
online-offline-online transition forces a reLogin with the fabric as
described in PH 23.5.3.  However I have always interpreted this section to
imply that if the reLogin with the fabric results in all addresses being the
same (N_Port D_ID, fabric's PortName/NodeName) then "address authentication'
has occurred and SCSI exchanges may resume.  Do we need words for this
environment also?  Does everyone agree with my interpretation?

Finally, whether or not it is loop or direct switch attach, the link may be
down for an extended period of time.  We currently have words saying the
timer starts *after* the link is back up for loop, I'd assume the same
applies to direct switch attach?  In other words, don't time the offline
period??

Charles Binford
LSI Logic Storage Systems
(316) 636-8566

> -----Original Message-----
> From:	Bob Snively [SMTP:Bob.Snively at EBay.Sun.COM]
> Sent:	Tuesday, July 20, 1999 5:52 PM
> To:	fc at nsco.network.com; t10 at aztec.co.lsil.com
> Subject:	RE: FCP-2 resource recovery question
> 
> * From the T10 Reflector (t10 at symbios.com), posted by:
> * Bob Snively <Bob.Snively at EBay.Sun.COM>
> *
> I have always hoped that the text was pretty obvious.
> 
> 	RR_TOV is the minimum time a SCSI Target shall wait for a specific 
> 	SCSI Initiator to perform Exchange Authentication following the 
> 	completion of the Loop Initialization Protocol (i.e., the receipt of
> CLS 
> 	while in the OPEN-INIT state). RR_TOV is also the minimum time a
> SCSI 
> 	Target shall wait for a SCSI Initiator response following transfer
> of 
> 	Sequence Initiative from the SCSI Target to the SCSI Initiator
> (e.g., 
> 	following transmission of the FCP_XFER_RDY during a write command).
> 
> This provides both clear indications as to when the timer should start
> (independent of topology in the second case), and when it should end.
> 
> In the first case, there is a topology factor in the definition of
> exchange authentication, which means PDISC/ADISC for PLDA environments and
> FAN for FLA environments.  For fabric attach point-to-point environments,
> the question gets a bit more interesting, but not much.  If a link
> reset protocol is required, I have always assumed that it will be finished
> in a short time compared with RR_TOV.  Thus a recovery for exchange
> authentication purposes is indistinguishable from the recovery waiting for
> transfer of sequence initiative.  For recoveries requiring a new login
> (such as online-offline-online transitions), it would seem to me that
> the RR_TOV would not be so relevant, since exchange resources were already
> cleared at both ends.
> 
> What other factors have I missed?
> 
> Do we want to specify this in tabular format, as Charles does below,
> as part of the discovery annex or in the body of the document?
> 
> Bob
> 
> 
> 
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com





More information about the T10 mailing list