Re: Working network but problem with OSPF

danabersoch · ‎12-13-2009

Hey guys,

I have 3 routers all connected to one another: R1-----has two links (serial), one to R2 and one to R3

R2-----has two links (serial), one to R1 and one to R3

R3-----has two links (serial), one to R1 and one to R2

Each router has a f/ethernet to a L3 switch: R1------Site1S1

R2------Site2S1

R3------Site3S1

All hosts on the network from each site CAN ping one another (full convergence if you like), all routing tables are populated fine......

I am am massively confused because I wanted R3 to be the DR and R2 to be BDR .... so I configured these router interfaces (all of the ints after doing just one there was no elected DR or BDR) on R3 with priority 200 and R2 with priority 100 and left all other R's and S's on 1.

When I: show ip ospf neighbor------ on any of the routers they all think the attached L3 is the BDR and nothing is the DR. (same command) on the L3's they all think the attatched router is the DR

the area is 0 throghout I honestly cant figure this one out???????????????????? I have it on packet tracer if anyone fancies a look as nothing stands out!

I will appreciate any advice, or if you just want a look at the network give me a shout!

Thank you in advance

danabersoch · ‎12-13-2009

With the command Show ip ospf interface

Ive come to the conclusion that it is somthing to do with the fact that the links to the L3's are BROADCAST and the links between the routers are point-to-point and that the default priority values are different for each.

I still want it to be all one "Instsance" (R3 as the DR and R2 as the BDR throughout the network) and I thought it would because they are all in area 0, I am very confused please can someone explain whats going on???

thank you again

Dan

Jon Marshall · ‎12-13-2009

Could you post output of "sh ip ospf neigh" from each router.

Also, are you aware that DR/BDR are per interface particpating in OSPF and not for the router as a whole ?

Jon

danabersoch · ‎12-13-2009

I wasn't aware of the fsct that the DR and BDR is per int and not for the whole router!! how do you get loads of routers in a situation where there is 1 DR and 1 BDR and loads of DROthers if it is per interface?

I have at the moment: R1----R2------R3-----then R3 has another link back to R1 ( for redundancy and so R1 and R3 can communicate without having to go through R2)

How would the DR and BDR election differ if I didn't have the link from R3 back?

here are the outputs:

(R1=Site1R1 R2=Site2R1 R3=Site3R1)

The layer3 switches are linked to there relative routers.

Site1R1#sh ip ospf n

Neighbor ID     Pri   State           Dead Time   Address         Interface
192.168.100.1     1   FULL/BDR        00:00:37    192.168.10.1    FastEthernet0/1
192.168.100.22 100   FULL/ -        00:00:35    192.168.10.6    Serial0/0/0
192.168.100.33 200   FULL/ -        00:00:37    192.168.10.21   Serial0/2/0

Site1S1#sh ip ospf ne

Neighbor ID Pri State Dead Time Address Interface
192.168.100.11 1 FULL/DR 00:00:31 192.168.10.2 FastEthernet0/24

Site2R1#sh ip ospf n

Neighbor ID     Pri   State           Dead Time   Address         Interface
192.168.100.2     1   FULL/BDR        00:00:32    192.168.10.9    FastEthernet0/1
192.168.100.33 200   FULL/ -        00:00:32    192.168.10.14   Serial0/0/0
192.168.100.11    0   FULL/ -        00:00:32    192.168.10.5    Serial0/2/0

Site2S1>en
Password:
Site2S1#sh ip ospf n

Neighbor ID Pri State Dead Time Address Interface
192.168.100.22 100 FULL/DR 00:00:31 192.168.10.10 FastEthernet0/24

Site3R1#sh ip ospf n

Neighbor ID     Pri   State           Dead Time   Address         Interface
192.168.100.3     1   FULL/BDR        00:00:33    192.168.10.17   FastEthernet0/1
192.168.100.11    0   FULL/ -        00:00:32    192.168.10.22   Serial0/0/0
192.168.100.22 100   FULL/ -        00:00:33    192.168.10.13   Serial0/2/0

Site3S1#sh ip ospf n

Neighbor ID Pri State Dead Time Address Interface
192.168.100.33 200 FULL/DR 00:00:38 192.168.10.18 FastEthernet0/24

Jon Marshall · ‎12-13-2009

danabersoch wrote:

I wasn't aware of the fsct that the DR and BDR is per int and not for the whole router!! how do you get loads of routers in a situation where there is 1 DR and 1 BDR and loads of DROthers if it is per interface?

I have at the moment: R1----R2------R3-----then R3 has another link back to R1 ( for redundancy and so R1 and R3 can communicate without having to go through R2)

How would the DR and BDR election differ if I didn't have the link from R3 back?

here are the outputs:

(R1=Site1R1 R2=Site2R1 R3=Site3R1)

The layer3 switches are linked to there relative routers.

Site1R1#sh ip ospf n

Neighbor ID     Pri   State           Dead Time   Address         Interface
192.168.100.1     1   FULL/BDR        00:00:37    192.168.10.1    FastEthernet0/1
192.168.100.22 100   FULL/ -        00:00:35    192.168.10.6    Serial0/0/0
192.168.100.33 200   FULL/ -        00:00:37    192.168.10.21   Serial0/2/0

Site1S1#sh ip ospf ne

Neighbor ID Pri State Dead Time Address Interface
192.168.100.11 1 FULL/DR 00:00:31 192.168.10.2 FastEthernet0/24

Site2R1#sh ip ospf n

Neighbor ID     Pri   State           Dead Time   Address         Interface
192.168.100.2     1   FULL/BDR        00:00:32    192.168.10.9    FastEthernet0/1
192.168.100.33 200   FULL/ -        00:00:32    192.168.10.14   Serial0/0/0
192.168.100.11    0   FULL/ -        00:00:32    192.168.10.5    Serial0/2/0

Site2S1>en
Password:
Site2S1#sh ip ospf n

Neighbor ID Pri State Dead Time Address Interface
192.168.100.22 100 FULL/DR 00:00:31 192.168.10.10 FastEthernet0/24

Site3R1#sh ip ospf n

Neighbor ID     Pri   State           Dead Time   Address         Interface
192.168.100.3     1   FULL/BDR        00:00:33    192.168.10.17   FastEthernet0/1
192.168.100.11    0   FULL/ -        00:00:32    192.168.10.22   Serial0/0/0
192.168.100.22 100   FULL/ -        00:00:33    192.168.10.13   Serial0/2/0

Site3S1#sh ip ospf n

Neighbor ID Pri State Dead Time Address Interface
192.168.100.33 200 FULL/DR 00:00:38 192.168.10.18 FastEthernet0/24

Maybe i should explain a bit better.

It is per interface but that interface connects to a network segment. So lets say you have an ethernet segement and 4 different routers are attached to this segment. Each router has an interface in that network so you will have a DR, a BDR and 2 DROthers.

Couple of other points -

1) It is important to understand also that the DR and BDR form a full adjacency with the DROthers

2) On a point to point network there is no DR/BDR and the 2 ends of the point to point link form a full adjacency.

So as an example, from your output above -

Site1R1#sh ip ospf n

Neighbor ID     Pri   State           Dead Time   Address         Interface
192.168.100.1     1   FULL/BDR        00:00:37    192.168.10.1    FastEthernet0/1
192.168.100.22 100   FULL/ -        00:00:35    192.168.10.6    Serial0/0/0
192.168.100.33 200   FULL/ -        00:00:37    192.168.10.21   Serial0/2/0

R1 has one neighbor on the fa0/1 interface and that would be the L3 switch. The layer 3 switch is the BDR for the ethernet segment.

R1 also has 2 point to point links (remember no DR/BDR) and has establised full adjacency on bit links.

Site1S1#sh ip ospf ne

Neighbor ID Pri State Dead Time Address Interface
192.168.100.11 1 FULL/DR 00:00:31 192.168.10.2 FastEthernet0/24

Your switch S1 that is connected to R1 shares the same ethernet segment. R1 is seen as the DR which makes sense because

1) there are only 2 interfaces on the ethernet segment

2) from the output of R1 we have already established that S1 is the BDR so that leaves R1 to be the DR

Jon

danabersoch · ‎12-13-2009

Thank you for your help jon that has cleared a few confusing things up for me.

Is the setup that I have common, un-common, good, bad? and I dont suppose you can see from what you know about the network if there is a way of making this any better if it is not very good or not common practice?

thanks again

dan

Giuseppe Larosa · ‎12-14-2009

Hello Dan,

Jon has explained you that all what you see is normal and he has clarified the DR/BDR concept (that is per network segment and not per area or per router)

About your setup: it is quite common and correct as far as we can see.

Area 0 must be on the center or each OSPF domain. One non zero area per site could be used to provide a point where some control (inter-area summarization and route filtering are possible on ABR that are the OSPF nodes that connect to two or more areas, with one of them being area 0)

The choice of deploying multiple areas can be good if the number of sites grow over time.

Hope to help

Giuseppe