Load Balancer status UNKNOWN after NSX V2T Migration

Products

• VMware NSX

Issue/Introduction

• The load balancer is migrated from NSX for vSphere (NSX-v) to NSX-T Datacenter as a native (non-AVI) load-balancer.
• One of the virtual servers is configured with a Source IP persistency (where the persistency profile has no time-out configured)
• Virtual IP addresses are reachable.
• All Virtual Servers of a given load balancer are affected.
• The NSX-T GUI shows a status UNKNOWN for the load balancer and its related component (virtual servers and server pools).
• Running the following command on the NSX-T Edge CLI returns the following:

   NSXTEdge01I> get load-balancer <Load-balancer UUID> status

   Internal Error: Query LB Engine Failed.

Cause

The issue is caused by a missing parameter in the load-balancer configuration file, which totally locks up the load-balancer and its virtual servers. Due to the missing time-out setting on the persistency profile, the “expire” parameter cannot be written in the load-balancer configuration file: locking up nginx (the load-balancer which NSX-T uses internally).

Resolution

Configure the expiry setting of the persistency profile on the NSX V LB configuration and migrate the configuration using migration coordinator.
Workaround:
Do a roll-back of the Migration through Migration-coordinator.
Update the "Expires in": field on the persistency profile on source NSX V ESG.
Attempt the configuration migration again.

Additional Information

Source side configuration on the NSX V LB ESG

 


Migrated NSX T native LB On the NSX T side

 

 

 

 


LB state on the edge node CLI
 
|


Impact/Risks:
The NSX T native Load-Balancer does not get realized after NSX V2T Migration through Migration-coordinator.