VM replications are frequently changing to a paused state

Products

• VMware Cloud Director
• VMware Live Recovery

Issue/Introduction

• VM replications are entering 'paused' state randomly after some successful synchronizations.
• vSphere DRS is migrating the source VM often to different ESXi hosts.
• Within the hbrsrv.log on the replicator you may see the following log:
  
  error hbrsrv[###] [Originator@6876 sub=AgentConnection opID=###########-####-####-####-###########] Connection failed to agent ###########-####-####-####-###########:host-###/hostd (<source-esxi-vmkernel-ip>): Can't login to the host

Environment

VMware Cloud Director Availability 4.7.x

Cause

This issue can occur when the source VM is located on an ESXi host which does not have a dedicated vmkernel port for vSphere Replication traffic. The replication traffic will have to share with the management vmkernel port and performance bottlenecks can exist.

Resolution

Dedicating a VMkernel port for VMware Cloud Director Availability (VCDA) isolates replication traffic from your management and VM networks. To implement this best practice, create an isolated VMkernel port in a dedicated IP subnet and enable the vSphere Replication and vSphere NFC Replication as outlined in the Deployment Requirements.