The current core network of the MAN consists of two 12416 routers made by Vender C and respectively located at site A and site B. The total bandwidth of the MAN for outgoing packets is 3x2.5 Gbit/s. The bandwidth from C12416 at site A to the C3 node on the backbone network is 1x2.5 Gbit/s. The bandwidth from C12416 at site B to the C3 node on the backbone network is 2x2.5 Gbit/s. The bandwidth of the two C3 nodes on the backbone network is 2x10 Gbit/s. To adapt to the business development, the carrier attempts to optimize the MAN and the backbone network. Two C3 devices (the NE5000E and the C12416 router) on the backbone network are reduced to the core routers of the MAN for outgoing packets and become the EBGP peer of C2. The original core router of the MAN is kept for other use (MAN SR) and the private AS number of the MAN is also kept. After the cutover, IBGP neighborship is established between the two core networks of the MAN The SR and BRAS originally connected to C12416 are connected to the NE5000E and C12416.
In the service test after the cutover, some MPLS VPN services under the BRAS run abnormally and some run normally (users in the same VPN).
1. Checking the data configured on the NE5000E, the engineer found no error.
2. Checking the MTU and BRAS configurations at the NE5000E interface, the engineer found no error.
3. Run the ping lsp command on the NE5000E with the source address (the loopback address of the NE5000E) to ping the loopback address of the BRAS. If the loopback address can be successfully pinged, packets can be forwarded between the NE5000E and the BRAS normally.
4. On the BRAS, the engineer adopted the same method to ping the PE device that could not be connected in the VPN. The PE could not be successfully pinged if LSP is contained in the IP address of the PE.
5. Checking the LDP PEER on the BRAS, the engineer found that the PEER address of the C12416 is LOOPBACK1 rather than LOOPBACK0. Set the same LOOPBACK1 address on the C12416 as that of the NE5000E to meet the special business demand of the customer.
6. Checking the LDP PEER on the NE5000E, the engineer found that the PEER address of the C12416 is also LOOPBACK1 rather than LOOPBACK0.
7. Checking the configuration on the C12416, the engineer found that the PEER address was set incorrectly. The address of the C12416 should have been LOOPBACK0, but it was set to LOOPBACK1. This was why some services forwarded by the C12416 could not be accessed normally. This was because the traffic, accessed through the original core router of the MAN before the cutover for the NE5000E, was switched to the C12416 during the cutover and was not switched back to the NE5000E after the successful cutover.
1. The incorrect data configuration of the NE5000E.
2. The incorrect setting of MTU.
3. The MPLS forwarding problem.
The main reason for this problem is the incorrect MPLS LSP forwarding.
When handling this type of failure, you are recommended to ping the IP addresses of PEs with LSP to solve the problem soon.