LDP Transport Address Troubleshooting
Load ldp.transport.address.cfg
There are two LDP adjacencies pre-configured: R1-R2 and R6-XR1. These are not coming up.
Find two separate ways to solve this issue. R1 and R2’s adjacency should still allow for session protection to work. XR1 and R6’s adjacency does not have this requirement.
Answer
Explanation
The LDP TCP session uses the LDP transport address for the endpoints, not the interface address. By default, the LDP transport address is the LDP RID. Because the LDP RID uses the same default methods as IGP protocols (configured RID, highest loopback IP, then highest interface IP), the LDP RID is by default Lo0 for all routers. However, Lo0 is not being advertised into the IGP in this lab.
On R1 we can see that we have no LDP neighbors. If we use the command show mpls ldp discovery we can see the reason why. R2 is using 2.2.2.2 as its discovery address, however we have no route to it.
If we use the interface as the transport address, this does not allow us to use LDP session protection. LDP session protection works by creating a targeted session that can re-route around link failure. If the TCP session uses the link addresses, it can’t re-route if that link goes down.
To demonstrate this, I’ve set the discovery address as the interface address and configured session protection. We can see that now the LDP neighbor is up, but session protection is not active.
This means our only option is to advertise the loopbacks into the IGP on R1 and R2. (Alternatively, configuring static routes for the loopback addresses would solve this lab as well).
We don’t have this same restriction on R6 and XR1. Because we have to use a different method to bring the adjacency up, we can use the interface address as the transport address.
On each router we can verify that the transport address is the interface address instead of the LDP RID.
We can also see that the session uses the interface addresses as the TCP endpoints:
Last updated
