Fault Symptom
OSPF multi-instance is run between PEs and CEs. The CEs are Layer 3 switches of other manufacturers. The PEs deliver OSPF default routes to interwork the networks of the two cities. PE1 and PE2 are connected to the same UMG. The same IP address 10.1.1.33 is set for the interface connecting PE1 to the UMG and the interface connecting PE2 to the UMG, and the two interfaces are bound to the VPN instance of the UMG. Normally, the link between the UMG and PE2 is Down. Therefore, the two interfaces with the IP address being 10.1.1.33 on the two PEs cannot be in the Up state at the same time.
CE1 can successfully ping PE1, and CE2 can successfully ping PE2. When a CE pings its remote peer or a device on the remote network, packet loss, however, occurs irregularly.
Fault Analysis
1. 10.1.1.33 is the largest IP address in the VPN instance to which the two PEs are bound, and the following command is run to configure OSPF multi-instance:
<PE1> ospf 4 vpn-instance www
Therefore, PE1 and PE2 select 10.1.1.33 as their router ID.
2. On CE1, you can find that the router ID of PE1 is 10.1.1.33; on CE2, you can find that the router ID of PE2 is also 10.1.1.33.
3. The debugging information on the CEs shows that a device with the router ID being 10.1.1.33 sends LSAs every five seconds and the sequence numbers of LSAs are incremental and unstable.
4. The CEs receive LSAs sent by two devices with the same router ID. Therefore, the OSPF default routes in the routing tables of the CEs constantly change. When the default route of CE1 is learned by CE2 and the default route of CE2 is learned by CE1, a routing loop occurs. As a result, routes are unreachable and packet loss occurs.
Procedure
1. Run the ospf 4 router-id 10.2.2.9 vpn-instance www command on PE1 to specify the router ID of the OSPF multi-instance as the unique address of PE1, and run the ospf 4 router-id 10.2.2.10 vpn-instance www command on PE2 to specify the router ID of the OSPF multi-instance as the unique address of PE2.
[PE1] ospf 4 router-id 10.2.2.9 vpn-instance www
[PE2] ospf 4 router-id 10.2.2.10 vpn-instance www
2. Restart the OSPF process associated with the VPN instance on PE1, and then perform the same operation on PE2. After that, services are restored.
Summary
Specify the router ID of the OSPF multi-instance as the unique address of on the PEs.
