BFD: When BGP peer is shutdown by admin, associated BFD session should also be brought down or not ?

[sougata-github-nvidia]

Hi,

When a BGP peer is administratively shut down, should the associated BFD session also be brought down automatically?

I couldn’t find any explicit guidance on this behavior in the relevant RFCs. Could you please clarify whether it is recommended for BGP to terminate the BFD session in such cases, or should this decision be left to the user? That is, the user can manually unbind or disable the BFD session if needed.

While automatically tearing down the BFD session may help with resource optimization, implementing this behavior consistently would require the same logic to be applied across all BFD client protocols for consistency.

Looking forward to your input.

[ton31337]

When a BGP peer is administratively shut down, should the associated BFD session also be brought down automatically?

Yes, but it will still allow establishing a BGP session even if BFD is down unless a strict mode is used.

[sougata-github-nvidia]

Hi @ton31337 ,
Thanks for your quick response. My question is the opposite of what happens in BFD strict-mode. Suppose both the BGP session and its associated BFD session are up, and then the user shuts down the BGP peer using the neighbor shutdown command.
In this scenario, will the corresponding BFD session also be brought down automatically?
If not, is it acceptable to keep the BFD session up even when BGP is administratively shut down? From a resource optimization perspective, wouldn’t it be better to bring down the BFD session as well, to free up resources?

[ton31337]

I can't answer this because I'm not familiar with what should be followed for the BFD session once the BGP session goes down. Free up resources...maybe unless there are other protocols relying on BFD session too?

[sougata-github-nvidia]

Yes , if there is no other protocol dependent on that session, it can be freeed.