BFD to BGP neighbors may go down after the restart of data path service

Products

• VMware NSX

Issue/Introduction

 BFD to BGP neighbors remain down. Some connected routes are missing in non-default VRFs. BFD down will bring down BGP therefore impact routing.

 

 

Environment

VMware NSX T 3.2.5

VMware NSX 4.2.0

 

Cause

The restart of the data path service may induce this error condition, since it involves the deletion and recreation of the interface added within the Linux namespace for frr's consumption.

If this "ifindex" is missing, log output similar to the below will be found in the logs. NSX T edge log contains following entry

/var/log/syslog

 

ROUTING [nsx@6876 comp="nsx-edge" subcomp="nsxa" s2comp="routing" level="ERROR" errorCode="EDG####"] FIB update 16 error: update ###.###.###.###/28 add - ifindex 104 not found

In addition to above log entries, in the output of following command . 

'get logical-router <UUID>'

Forwarding shows some routes (especially connected routes) are missing.

Resolution

Workaround:

Login to NSX T edge as user 'admin'
Restart local-controller.

>restart service local-controller

Note:  Though restarting local-controller may not cause impact to data plane forwarding, customer should follow the maintenance procedure.

 

Solution 

Upgrade NSX version to 

NSX T 3.2.5

NSX  4.2.0

Additional Information

Fixed Version
3.2.5, 4.2.0