Disconnection of esxi hosts with vCenter dual stacked ipv6/ipv4 configuration due to an ipv6 routing disruption

Products

• VMware NSX

Issue/Introduction

• Dual stack configuration for ipv4/ipv6
• vCenter runtime settings for vCenter Server managed address blank or hard set to ipv6 or ipv4
• ipv6 or ipv4 routing is disrupted, hosts disconnected

Environment

vSphere 7

vSphere 8

Cause

There is no failover mechanism between ipv4 or ipv6 in the vpxa agent configured managed address for vCenter server.

Resolution

Use the same broadcast domain for vCenter and esxi so that neither dual stack of ipv4 or ipv6 is affected by routing disruptions.

Workaround the issue by manually configuring the vpxa agent managed vcenter server_ip for the preferred dual stack protocol that is routable and make sure to match the vCenter Runtime Settings for managed Address. Host management services may need to be restarted or host rebooted. Manually reconnect the host in the vSphere client.

Note: The host_ip and server_ip which contain the current ESXi management IP address and the vCenter Server IP address, are not part of the vpxa service configuration in the config store but are instead stored in the vpxa solution user configuration.

Additional Information

vCenters in a VCHA configuration need static IP addresses for the management network. The management and cluster network addresses must be IPv4 or IPv6. They cannot be mixed mode IP addresses.