BFD Tshoot #1
Last updated
Last updated
Load bfd.tshoot1.cfg
R4 and R6 are OSPF neighbors. BFD is pre-configured, but the administrator found that when the link goes down on one end, it still takes the full OSPF dead interval to detect this. Explain the issue and find a way to fix it.
The main issue that is causing BFD failure detection to be so slow is that R6 is currently using the max BFD interval (9999 msec):
We can also see that BFD is “not using echo function.” So control packets are used, and they are being paced at ~10 second intervals.
If we bring down Gi2.46 on R6, we’ll see that OSPF detects the failure on R4 faster than BFD. BFD right now is detecting the failure in 10*5 seconds = 50 seconds!
Since OSPF detects the failure first, the BFD session then is destroyed because the neighbor is gone. So BFD never really detects the failure anyways.
The first way to fix this would be to fix the interval on R6:
BFD is still not using echo packets, but at least the interval has been reduced to 750msec.
BFD failure detection now happens in .750 * 5 = less than 4 seconds. Now on R4, BFD does detect the failure and notifies OSPF.
To improve this even more, we should enable echos on R6. Echos are enabled by default on R4 since BFD is enabled directly under the interface. Since R6 is using a template, echos are disabled by default, so we must use the echo keyword.
