BFD Single Hop session keeps flapping after the link-down event; sometimes the EVO BFD app does not show online, and the PPM receive shows as state none.

When creating bfd endpoint with extended SiP tunnel, if there is an SDK API failure in the tunnel creation, the allocated lif index is not freed, leaving a stale entry that could cause an issue later. BFD endpoint creation is not effectively using the dedicated oamp bfd source ipv4 table that does not require creating an extended sip tunnel thereby minimizing the usage of eedb resources.

show system applications app bfd   
<NO OUTPUT>

Working output:

>show system applications app bfdd   
Applications Information:
Application       : bfdd
Node          : re0
App State        : online ready 
App Weight       : 0
Object Producer details
Producer ID       : 0
 Epoch ID        : 0

pfe> show ppm control protocol bfd session 
Ppm Control BFD Sessions:

Type             Discriminator  UDP Port    Distrib Key  Interval(ms)  State  Remote Address
receive           37             single-hop  0x0100003d   900           none   X.X.X.X 

Workaround suggested if running a non-fixed release:

Apply inline-disable and ppm no-delegate-processing, after that BFD session will observe stability.

set protocols isis interface et-x/x/x family inet bfd-liveness-detection inline-disable

set routing-options ppm no-delegate-processing

This issue will be fixed in the following EVO releases:

evo:23.4R2-S4-EVO

evo:25.2R1-EVO