Configuring a Delay in Releasing Obtained Labels in a BGP LSP FRR Switchover Scenario

Usage Scenario

In a BGP LSP FRR switchover scenario shown in Figure 1, Device A and Device D belong to AS 100, whereas Device B, Device C, Device E, and Device F belong to AS 200. The optimal route selected by Device B is a route learned from its EBGP peer Device A, and the suboptimal route selected by Device B is a route learned from its IBGP peer Device E. In normal cases, Device B preferentially selects the labeled route learned from its EBGP peer Device A and sends the route to its IBGP peer Device C. Device B delivers an incoming label map (ILM) entry and next hop label forwarding entry (NHLFE), and Device C also delivers an NHLFE. If Device A restarts due to a fault, Device B withdraws the route learned from Device A; in this case, the original suboptimal route learned from its IBGP peer Device E becomes the new optimal route. Then Device B reflects the new optimal route to Device C instead of sending the route to Device C through the IBGP peer connection. Therefore, Device B releases the obtained label and deletes the ILM entry. If the NHLFE on Device C is not updated in time, traffic is still sent to Device B. Due to the lack of the ILM entry, Device B discards the traffic, causing packet loss once again.

Figure 1 BGP LSP FRR switchover networking

To prevent this problem, configure a delay in releasing obtained labels (deleting ILM entries) on Device B.

Pre-configuration Tasks

Before configuring a delay in releasing obtained labels in a BGP LSP FRR switchover scenario, complete the following tasks:

• Configure basic BGP functions.

• Configure BGP Auto FRR.