This KB article provides insights into the occurrence of BFD/BGP flapping when an RG (Redundancy Group) group with a value greater than or equal to 1 fails over on SRX-4200. If you're experiencing issues with BFD/BGP stability during failover events, this article offers explanations and possible solutions to mitigate flapping occurrences. Understanding these factors can help ensure the reliability and stability of network connections during failover events.

• Chassis cluster implementation is in use.
• BFD/BGP configurations are utilized.
• The issue persists during failover events but disappears upon rollback.
• Single-hop BFD is employed in distributed mode.

Ensure compliance with the following statements:

"To prevent BFD flapping during the general Routing Engine switchover event, specify a minimum interval of 5000 milliseconds for Routing Engine-based sessions. This minimum value is required because, during the general Routing Engine switchover event, processes such as RPD, MIBD, and SNMPD utilize CPU resources for more than the specified threshold value. Hence, BFD processing and scheduling is affected because of this lack of CPU resources."

"SRX Series Firewalls support a BFD failure detection time of 3 x 100 ms. We support this feature for a standalone SRX Series Firewall. It is not supported for chassis clusters."

Note that single-hop BFD in distributed mode is not supported on chassis clusters. SRX only supports BFD in Centralized mode. Follow these additional steps:

• Change BFD from Distributed Mode to Centralized Mode.
  • Enable routing-options ppm no-delegate-processing and then run the clear bfd session command.
  • Validate the changes with the command: >show ppm adjacencies detail
• Adjust the BFD timer settings to comply with the following statement:
  • "BFD failure detection time > 500 ms."
• Consider modifying the BFD timers according to the following recommendations:
  • To prevent BFD flapping during a general Routing Engine switchover event, specify a minimum interval of 5000 milliseconds for Routing Engine-based sessions. This minimum value is required because processes such as RPD, MIBD, and SNMPD utilize CPU resources for more than the specified threshold value during the general Routing Engine switchover event, impacting BFD processing and scheduling.
  • To maintain BFD sessions during a dual chassis cluster control link scenario, specify a minimum interval of 6000 milliseconds to prevent LACP flapping on the secondary node for Routing Engine-based sessions when the first control link fails.

This scenario is entirely expected, as the configurations mentioned were enabled on a chassis cluster set up with Active/Passive redundancy. It appears that BFD interprets this as a standalone scenario. However, when you fail over one of the RG (Redundancy Group) groups and the chassis cluster transitions to Active/Active mode, BFD encounters issues.

BFD for BGP Sessions