Re: Waiting State Question

Erblichs <erblichs@EARTHLINK.NET> Wed, 25 May 2005 08:44 UTC

Received: from cherry.ease.lsoft.com (cherry.ease.lsoft.com [209.119.0.109]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id EAA19414 for <ospf-archive@LISTS.IETF.ORG>; Wed, 25 May 2005 04:44:55 -0400 (EDT)
Received: from vms.dc.lsoft.com (209.119.0.2) by cherry.ease.lsoft.com (LSMTP for Digital Unix v1.1b) with SMTP id <9.0105A3FC@cherry.ease.lsoft.com>; Wed, 25 May 2005 4:44:53 -0400
Received: by PEACH.EASE.LSOFT.COM (LISTSERV-TCP/IP release 14.3) with spool id 72428155 for OSPF@PEACH.EASE.LSOFT.COM; Wed, 25 May 2005 04:44:51 -0400
Received: from 207.69.195.69 by WALNUT.EASE.LSOFT.COM (SMTPL release 1.0l) with TCP; Wed, 25 May 2005 04:44:51 -0400
Received: from h-68-164-85-18.snvacaid.dynamic.covad.net ([68.164.85.18] helo=earthlink.net) by pop-savannah.atl.sa.earthlink.net with esmtp (Exim 3.36 #10) id 1DarVS-0002zi-00 for OSPF@PEACH.EASE.LSOFT.COM; Wed, 25 May 2005 04:44:50 -0400
X-Sender: "Erblichs" <@smtp.earthlink.net> (Unverified)
X-Mailer: Mozilla 4.72 [en]C-gatewaynet (Win98; I)
X-Accept-Language: en
MIME-Version: 1.0
References: <20050519122021.40976.qmail@web25301.mail.ukl.yahoo.com> <1116526462.428cd77ee421b@panthermail.uwm.edu> <072701c55ca2$d08ebd40$a328fb80@Kishorepc> <1116528821.428ce0b537b88@panthermail.uwm.edu> <075301c55ca6$26ce54b0$a328fb80@Kishorepc> <428CEE56.186604C1@earthlink.net> <079301c55cb9$514775b0$a328fb80@Kishorepc> <428D4F61.10D00DA4@earthlink.net> <003f01c55d5e$d3a9d980$a328fb80@Kishorepc>
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Message-ID: <42943CD9.971A7D1C@earthlink.net>
Date: Wed, 25 May 2005 01:52:41 -0700
Reply-To: Mailing List <OSPF@PEACH.EASE.LSOFT.COM>
Sender: Mailing List <OSPF@PEACH.EASE.LSOFT.COM>
From: Erblichs <erblichs@EARTHLINK.NET>
Subject: Re: Waiting State Question
To: OSPF@PEACH.EASE.LSOFT.COM
Precedence: list
Content-Transfer-Encoding: 7bit

Kishore Rao,

	Sorry for the late response.
	Lets see. Oh...

	Adj communciation causing a split area can
	be done intentionally and un-intentionally...
	
	One might intentionally impliment a dual or more
	routing paths thru an area because..

	  One wishes to support the equiv of a VPN where
	  each group uses a diff MD5 auth.

	  OSPF does not offically support unequal cost
	  paths and thus alternate routes would be
	  unused without MPLS.

	  We wish to minimize the number of routers
	  within a grouping.

	  we wish to minimize the number of LSAs per
	  group.

	  We wish to only inject routes from other
	  routing protcols into one group.

	  We wish to support a different set of routers
	  per ISP and maybe have a limited number of
	  merge points.

	  We wish to support a subset of routers for
	  only transit traffic..

	  We wish to support a very fast hello interval
	  for one group and that functionally or someother
	  functionally is ONLY supported by a subset of
	  the routers..

	  etc...

	And unintentionally because one or more adj early
	  adj formation parameters, (hello fields) are diff,
	  thus dropping the hello pkt..

	  Because a full adj is unstable and past history
	  is delaying our newest adj formation.

	  Or,,, because we are in the latency time period
	  before we identify that we have two DRs..

	OR, because the new router / DR on that interface
	has a higher capability and we would rather have
	a re-election of DR, versus taking down all the
	routers for dead router interval..

	OR  ... 

	But it is too simple to declare "because the
	adj is broken".. It is because their are MANY
	reasons.


	Mitchell Erblich
	----------------------
	

	

Kishore Rao wrote:
> 
> > What is b/n them is broken?
> 
> Comunication between them is broken.
> 
> >
> > Combine areas and former entry/exit interfaces between
> > the areas will result in two or more DRs announcements
> > per pseudonode.
> >
> > This is taken care of in the Spec? Can anyone find the
> > section? :-)
> 
> Section F: Multiple interfaces to the same network/subnet.
> 
> I could not find any other section other than this.
> 
> F:  says both the interface should be taken care of together so that there
> are no
> multiple DRs elected by the same router. In your case, DRs were elected by
> two different routers,
> which I think could happen only if they have no communication beween them ?
> 
> Kishore
> 
> >
> > Mitchell Erblich
> > -------------------------
> >
> > Kishore Rao wrote:
> > >
> > > > Second, "in the case where more than 1 router"
> > > > is declaring itself as the DR, shouldn't enough
> > > > time pass (1.5 to 2x) hello interval pass to identify
> > > > this situation before exiting wait state and
> > > > poss determine whether 2-ways are forming.  Yes,
> > > > it could/should exit early, but on that first
> > > > hello???
> > >
> > > A possible case of A & B declaring themselves as DR would be when comm.
> b/n
> > > them is broken; in which case shouldn't router C prefer to elect itself
> as
> > > BDR ASAP on receiving the hello from either A or B instead of waiting
> for a
> > > longer period of time without the possiblity of neither A or B being
> elected
> > > BDR during that time ?
> > >
> > > >
> > > > Thus, this section of the spec covers the rare simple
> > > > case where no BDR has yet been elected or is eligible
> > > > to be elected.
> > > >
> > > > Mitchell Erblich
> > > > -----------------
> > > >
> > > >
> > > >
> > > > Kishore Rao wrote:
> > > > >
> > > > > Not NeighborChange but Backupseen
> > > > >
> > > > > "If the neighbor is both declaring itself to be Designated
> > > > >             Router (Hello Packet's Designated Router field =
> Neighbor IP
> > > > >             address) and the Backup Designated Router field in the
> > > > >             packet is equal to 0.0.0.0 and the receiving interface
> is in
> > > > >             state Waiting, the receiving interface's state machine
> is
> > > > >             scheduled with the event BackupSeen."
> > > > >
> > > > > > I think the NeighborChange events are ignored while an interface
> is in
> > > > > waiting
> > > > > > state.
> > > > > >
> > > > > > Thanks,
> > > > > > Mukul
> > > > > >
> > > > > >
> > > > > > Quoting Kishore Rao <kishore@IND.ALCATEL.COM>:
> > > > > >
> > > > > > > The question was not about how DR or BDRs are elected. John's
> > > question
> > > > > was
> > > > > > > if the router should exit Wait Timer (and run DR election) on
> > > receving
> > > > > Hello
> > > > > > > from a router declaring itself as DR. Well, from section 10.5 it
> > > should.
> > > > > > >
> > > > > > > Kishore
> > > > > > >
> > > > > > >
> > > > > > > > My guess is that if an interface comes out of the waiting
> state on
> > > > > > > receiving a
> > > > > > > > Hello from DR (without having received a Hello from BDR), it
> may
> > > elect
> > > > > > > itself
> > > > > > > > as BDR. This way many routers may elect themselves as BDR. Now
> all
> > > > > these
> > > > > > > BDR
> > > > > > > > claimants (except one) will ultimately take their claims to
> > > BDRship
> > > > > back
> > > > > > > but in
> > > > > > > > the process each router on the LAN may have to do several DR
> > > > > elections.
> > > > > > > >
> > > > > > > > Here is a paper we wrote recently that may shed further light
> on
> > > this:
> > > > > > > > http://cs.uwm.edu/~mukul/ospflan.pdf
> > > > > > > >
> > > > > > > > Thanks,
> > > > > > > > Mukul
> > > > > > > >
> > > > > > > > Quoting John Smith <jsmith4112003@YAHOO.CO.UK>:
> > > > > > > >
> > > > > > > > > Hi,
> > > > > > > > >
> > > > > > > > > When a router comes up it starts the Wait Timer before it
> elects
> > > the
> > > > > > > DR/BDR.
> > > > > > > > > It either
> > > > > > > > > waits for the Wait Timer to expire or it waits for a router
> > > > > declaring
> > > > > > > itself
> > > > > > > > > as the BDR
> > > > > > > > > before it decides that it needs to get out of the 'Waiting'
> > > state
> > > > > (it
> > > > > > > does
> > > > > > > > > this by
> > > > > > > > > generating the Backupseen event).
> > > > > > > > >
> > > > > > > > > My question is why does it wait only for the BDR? Why not
> the
> > > DR? It
> > > > > can
> > > > > > > when
> > > > > > > > > it recieves
> > > > > > > > > a HELLO from the DR know that their exists a DR and a BDR.
> Why
> > > not
> > > > > then
> > > > > > > get
> > > > > > > > > out of the
> > > > > > > > > 'Waiting' state?
> > > > > > > > >
> > > > > > > > > Thanks,
> > > > > > > > > John
> > > > > > > > >
> > > > > > > > > Send instant messages to your online friends
> > > > > > > http://uk.messenger.yahoo.com
> > > > > > > > >
> > > > > > >