When NSX Edge TEPs are not connected, they send BFD packets more frequently than the Edge Cluster Profile specifies

Products

• VMware NSX

Issue/Introduction

• NSX Edge TEPs use a custom Cluster Profile
• The BFD Probe Interval (ms) is higher than default. For this example, this value is 4000ms (4 seconds)
• The BFD Declare Dead Multiple is 3 (default) or more. For this example, this value is 3 (default)
  
• The expectation is that the TEPs will send BFD packets at an interval of 4 seconds, and will not declare the tunnel as down until 12 seconds have passed with 3 missed BFD packets
• Upon inspection, after the TEP tunnels have been declared down, BFD packets are seen on the network at an interval of 1 second, appearing not to honor the 4 second Probe Interval

Environment

VMware NSX 

Cause

This is expected behavior.

Resolution

The BFD Probe Interval multiplied by the BFD Declare Dead Multiple allows for the TEPs to wait 12 seconds before declaring a TEP tunnel is down.

However, after this initial 12 second period has expired, the TEPs begin sending out BFD Down packets at an interval of every 1000 ms (1 second) in an attempt to bring the TEP tunnels up as fast as possible. Once the TEP tunnels are re-established, the configured BFD Probe Interval goes into effect again.