From: Godswill Oletu (oletu@inbox.lv)
Date: Thu Jun 01 2006 - 15:23:54 ART
Petr,
What you have just stated happens, if you have more than two routers in a
multiaccess segment and a DR has already been elected. My scenario only have
two routers. When there is only one OSPF router on a segment, there is no
concept of DR/BDR, because the election of BD/BDR happens at OSPF state equal
or greater than 2-way. With only one router on the segment, it will not get to
the 2-way state by itself.
When the second OSPF router comes online, both routers will go through the
various OSPF states one after the other and when they get to 2-way (ie having
received and seen each other's RID in the exchanged hello packets), they will
decide among other things to elect DR/BDR, to form adjacency/or not to form,
etc...
If you have the time read through the RFC2328
http://www.faqs.org/rfcs/rfc2328.html, but, if you do not have the time, read
section 9.4(2). It states that, during the initial OSPF process, the DR/BDR
are deterministic; the highest priority wins and if there is a tie, then the
highest RID will win. And since it is not preemptive, a DR that transistion
from up to down, will not regain it DR status back, in a segment containg
OTHER DR/BDR ELIGIBLE ROUTERS.
However, in a two router segment, when the DR transistion from up to down, the
OSPF adjacency will be teared down, when it comes back up, both routerS will
negotiate the OSPF states all over again, as if it never occurred before.
There is no case of preempt/no preempt here; because, when the DR goes down,
so does the BDR and the OSPF adjacency and neighbor relationship. oOne OSPF
enabled router cannot be a DR or BDR by itself, so the concept of preemption
does not apply here. The concept of preemption only come to play, when there
are more than two routers in the segment and another DR was elected when the
old DR went down.
According to RFC2328 and then Cisco, the highest priority wins and in case of
a tie, the highest RID will wins, but that is not what my two router segment,
broadcast multiaccess topology is telling me. The results are not consistent.
The concept of the first router that boots up, will not even fly, because it
takes at least two OSPF enabled routers to kickoff the OSPF process.
Suggestions are welcome.
Thanks.
Godswill Oletu
----- Original Message -----
From: Petr Lapukhov
To: Godswill Oletu
Cc: Cisco certification
Sent: Thursday, June 01, 2006 1:24 PM
Subject: Re: DR Election: Incase of a Priority Tie, Highest RID Wins - Truth
or Fallacy?
Godswill,
The main idea for DR election with OSPF is that it is
"non-preemptive", i.e not deterministic, as it is
with IS-IS.
Actually, when router come alive on a shared network, it first
spend some time "waiting" for DR/BDR announces (WAIT time).
That is, if it hears DR/BDR announce in neighbors' hello packets,
it will not even bother participating in new elections, even if it has
higher priority or RID.
So if you really want some router to ALWAY be a DR, you should
set all other routers' priorities to zero.
That thing is explained in details in RFC2328 :)
HTH
Petr
2006/6/1, Godswill Oletu <oletu@inbox.lv>:
Hi,
This topic was beaten to death the past few weeks on the group and the
general
concession is that, when there is a tie on the priority vlaues, the
highest
Router-ID wins, Cisco online documentation have various pages confirming
this
as well. But, I do not know if anyone labbed this up and fool-proof this
concept.
I am have labbed that exact scenario, that required that a particular
router
be elected the DR in segment of two routers, the neighbor or priority
commands
are not to be used.
The results I am getting is not consistent across the board:
Little preview of my configures:
Rack1R2:
interface Serial0
no ip address
encapsulation frame-relay
no frame-relay inverse-arp
!
interface Serial0.204 point-to-point
ip address 144.1.24.2 255.255.255.0
ip ospf network broadcast
frame-relay interface-dlci 204
!
router ospf 1
router-id 222.2.2.2
log-adjacency-changes
network 144.1.24.2 0.0.0.0 area 0.0.0.0
network 150.1.2.2 0.0.0.0 area 0.0.0.0
!
-------------------------------------------------------------------------
----
--
Rack1R4:
interface Serial0/0
no ip address
encapsulation frame-relay
no frame-relay inverse-arp
!
interface Serial0/0.402 point-to-point
ip address 144.1.24.4 255.255.255.0
ip ospf network broadcast
frame-relay interface-dlci 402
!
router ospf 1
router-id 150.1.4.4
log-adjacency-changes
network 144.1.24.4 0.0.0.0 area 0.0.0.0
network 150.1.4.4 0.0.0.0 area 0.0.0.0
!
-------------------------------------------------------------------------
----
-------
Results:
Rack1R2#clear ip ospf process
Rack1R2#sho ip ospf nei
Neighbor ID Pri State Dead Time Address
Interface
150.1.4.4 1 FULL/DR 00:00:37 144.1.24.4
Serial0.204
Rack1R2#
Rack1R4#show ip ospf nei
Neighbor ID Pri State Dead Time Address
Interface
222.2.2.2 1 FULL/BDR 00:00:32 144.1.24.2
Serial0/0.402
Rack1R4#
Despite the fact that Rack1R4 have the lowest Router-ID, it was elected
the DR
for that segment and Rack1R2 who have the highest Router-ID settled for
the
less fancy job of a BDR.
Now............
Rack1R4#clear ip osp nei
Rack1R4#sho ip ospf nei
Neighbor ID Pri State Dead Time Address
Interface
222.2.2.2 1 FULL/DR 00:00:39 144.1.24.2
Serial0/0.402
Rack1R4#
Rack1R2#sh ip ospf nei
Neighbor ID Pri State Dead Time Address
Interface
150.1.4.4 1 FULL/BDR 00:00:34 144.1.24.4
Serial0.204
Rack1R2#
Now, the roles have been revised, completely negativing the 'supposed'
influence that a higher Router-ID should have in the DR/BDR election
process.
Or, are mine missing something here? Maybe my coffee have not sink in
yet...but your contribution is highly welcome.
Thanks.
Godswill Oletu
_______________________________________________________________________
Subscription information may be found at:
http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Sat Jul 01 2006 - 07:57:31 ART