OSPF multiple interfaces to the same subnet

[berniechen]

Hi 

Did anyone test this scenario before. Assuming eth1 eth2 and eth3 are in the same area and the same subnet (Let's say 192.168.2.0/24). From the RFC, both eth1 and eth2 are considered togather to DR election cause they have same router ID. Then assuming eth2 is DR, eth3 is BDR.

My qustion is:

1. What is status of eth1?  Does eth1 have neighbors or adjacencies?

Some documents say, in this case,  all inbound traffic for the RTA will use only one interface out of eth1 or eth2. the inbound traffic to RTA is not load balanced across the interfaces.

2. The next question is, if one link fails (The interface is still active ), can RTA detect that and switch the route to ETH1.   

I tested it by other routers, the result is, ETH1 is in the BACKUP status and it does not have any neighbor or adjacency. If I deactive ETH2, ETH1 will change the status to DR or  BDR. RTA will switch the route to ETH1. But if I just make link fail, both ETH2 and ETH3 will change to DR, RTA did not switch the route to ETH1.

I am appreciate that if some one can explain that and tell me what will happen in CISCO router.  Thanks in advance!

[rais]

When Eth1 and Eth2 are in the same subnet, it means they are bridged. The backup state seems to be STP blocked. Do you see Block status on switch side as well?

Router A would detect failure and move to the other.

Thanks.

[berniechen]

I do not have the environment, all of test is running in VM with quagga. I am not sure if any Layer 2 protocol block the OSPF multicast.

Could you explain what should happen and what exact status interfaces should be.

For example, if eth1 is DR, eth3 will be BDR, and eth2 will be DRother cause, this interface has the same router ID with eth1. blabla ... Then if the link between eth2 and switch fails, what will happen.

Thanks

[ebarticel]

It may work in a simulation enviroment, but on real router I don't think you can configure 2 interfaces in the same subnet.

The DR will be the interface with the highest IP since all other values will be the same, the interface with second highest IP will become BDR and the third interface will become DROther.

eth1  192.168.2.1  DROther

eth2 192.168.2.2   BDR

eth3 192.168.2.3   DR

Hope this helps

Eugen

[berniechen]

Any reason? RFC says it is not allowed or some design consideration?

Have you tested this case in your environment. Actually, as I know, some routers can work in this case

It is appreciate if you can explain more detailed. Thanks

[ebarticel]

I have tested on 2620 and 1760 routers and I couldn't cofigured it. Router will say that it overlaps with other interface.

If you run "show ip int brief" you will see that other interface has ip address, but if you try to open the interface the router will tell you that is an incorrect ip address assignment and will stay administratively down.

I have tried on Packet Tracer, GNS3 with same results.

1. OSPF election process for DR and BDR first looks at the router-id which by default should be the same on all routers unless was manually changed changed.

2. If router-id are the same then OSPF looks at loopback interfaces and chooses the highest

3. If no loopback interfaces are configured, OSPF looks at the highest IP configured on router, even if that interface is not active.

Eugen