OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
 
From: Moy, John (John.MoySYCAMORENET.COM)
Date: Tue Mar 27 2001 - 12:53:21 CST

  • Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

    You know all your interfaces. The neighbors that
    you want are all the ones attached to those interfaces.
    You *could* figure out *exactly* who they were by
    listening for hellos for a while, but I wouldn't
    bother. Just send the grace-LSAs out each interface
    n times, where n= the interface's Dead/Hello interval.

    John

    -----Original Message-----
    From: Alex Zinin [mailto:azinincisco.com]
    Sent: Tuesday, March 27, 2001 1:34 PM
    To: Moy, John
    Cc: OSPFDISCUSS.MICROSOFT.COM
    Subject: Re: draft-ietf-ospf-hitless-restart-00.txt,
    draft-ietf-idr-restar t-00.txt

    John,

    > They are *all* the potential neighbors.

    And how do you know them all when you have just
    started?

    Alex.

    > You know who *they* are. After some number of
    > retransmissions, you assume that they are not
    > there.

    > John

    > -----Original Message-----
    > From: Alex Zinin [mailto:azinincisco.com]
    > Sent: Tuesday, March 27, 2001 1:25 PM
    > To: Moy, John
    > Cc: OSPFDISCUSS.MICROSOFT.COM
    > Subject: Re: draft-ietf-ospf-hitless-restart-00.txt,
    > draft-ietf-idr-restar t-00.txt

    > John:

    >>> How can you guarantee that the neighbors do receive the Grace
    >>> LSA before the first Hello from the restarting OSPF process and that
    >>> they receive it _before_ the Hello?

    >> You retransmit the grace-LSA until they acknowledge it, and

    > This question has already been discussed (quote below).
    > The restarting router does not know who "they" are.

    > Alex.

    >> Thursday, January 04, 2001, 12:28:30 PM, Moy, John wrote:

    >>> From: Acee Lindem [mailto:aceeREDBACK.COM]

    >>> John,
    >>
    >>> Wouldn't a router also have to remember it's full
    >>> neighbors (independent of whether or not cryptographic
    >>> authentication is in use) in order to reliably
    >>> flood the link local LSA? If not, aren't some slight
    >>> modifications to link local flooding required?
    >>
    >> Acee-
    >>
    >> No, you don't have to remember. If you are sending
    >> the grace-LSA *after* the restart (something I
    >> wouldn't personally recommend, but I know it's
    >> something that Alex wants to support), you're not really
    >> flooding it since you (the restarting router) no
    >> longer have a full adjacency anyway, so just
    >> sending it out all interfaces suffices. If you want
    >> to be extra sure that the (old) full neighbors get it,
    >> you can send it more than once... (but I wouldn't
    >> bother personally).

    >> John
    >>

    >> then you send your first hello. If your implementation can't
    >> do this for some reason, you modify your implementation.

    >> John