[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

    Re: FMarker, RFMarkInt and SFMarkInt negotiation

    What would the IT and TI add? The information is already available in the PDU (IT and IT PDUS are different).


    Martins Krikis <>
    Sent by:

    20-02-02 00:50

            Subject:        FMarker, RFMarkInt and SFMarkInt negotiation


    Dear list members,

    I seem to be having problems (or at least mild
    surprises) reading A.3.1 to A.3.3 of draft 10.

    Would you mind answering a few questions and hearing
    out some beginner's ideas? Many thanks in advance.
    Is it true that RFMarkInt when used by initiator
    means the intervals for the T->I direction, while
    this same key used by target refers to the I->T
    direction? Similarly (well, dually) for SFMarkInt?

    That is, if initiator starts the negotiation and sends
     FMarker=receive; RFMarkInt=17,51
    the target should answer with something like
     FMarker=send; SFMarkInt=34
    (the numbers used are just examples).

    I'm not sure that RFMarker is properly negotiated
    if the answer for it comes with the key "SFMarker"...

    The fact that FMarker=receive by originator
    is the same as FMarker=send by responder isn't
    pretty either, but at least the given example
    leaves no doubt. Or does it? What if send/receive
    is always with respect to the target? It would
    contradict the general principle that most everything
    in iSCSI is named with respect to initiator, but
    no example eliminates this possibility...

    May I suggest that we use something like ITFMarkInt
    and TIFMarkInt to unambiguously show which
    direction is being talked about?

    I would also love to tinker with FMarker values
    a bit, ideally get rid of it (thus getting rid of
    a clear parameter dependency), and use ITFMarkInt=0
    and TIFMarker=0 to denote that markers are not in
    use in the respective direction.

    Any comments will be greatly appreciated.

     Martins Krikis

    P.S. I am not really a fan of features that are
        useless for software implementations and that
        break protocol layering, but would like to at
        least understand how they are to be negotiated...

    Do You Yahoo!?
    Yahoo! Sports - Coverage of the 2002 Olympic Games


Last updated: Fri Feb 22 13:18:03 2002
8848 messages in chronological order