Reverting from a Cloud Director Appliance Deployment with Database HA to a Cloud Director Appliance Deployment Without Database HA
search cancel

Reverting from a Cloud Director Appliance Deployment with Database HA to a Cloud Director Appliance Deployment Without Database HA

book

Article ID: 373619

calendar_today

Updated On:

Products

VMware Cloud Director

Issue/Introduction

  • Cloud Director has been deployed as a Cloud Director Appliance Deployment with Database HA including a Primary Cell and Standby Cells.
  • Desire to revert to a Cloud Director Appliance Deployment Without Database HA, a single Primary Cell and no Standby Cells.
  • Cloud Director Appliance Cluster Health is in a Read_Only_Primary state.

Environment

VMware Cloud Director

Cause

After Standby Cells are added to a Cloud Director Appliance Cluster the Primary Cell will only be in a healthy state if there is at least one operational Standby Cell following it.

Broadcom does not provide support for VMware Cloud Director appliance deployments without database HA as outlined in the documentation, VMware Cloud Director Appliance Deployments and Database High Availability Configuration.

 

Resolution

The supported solution to a Cloud Director Appliance Cluster Health status of Read_Only_Primary is to deploy two new Standby Cells as per the documentation, View Your VMware Cloud Director Appliance Cluster Health and Failover Mode and Recover from a VMware Cloud Director Appliance Standby Cell Failure in a High Availability Cluster.
 
After two hours of running without any operational Standby Cells, the Primary Cell should remove any Standby Cell folders from the /opt/vmware/vcloud-director/data/transfer/appliance-nodes directory and the Primary Cell should become operational.
The Standby Cells should then be redeployed to return to a supported configuration as per the documentation, Recover from a VMware Cloud Director Appliance Standby Cell Failure in a High Availability Cluster.
 
To confirm if the Primary Cell requires an active Standby Cell, check the synchronous_standby_names setting:
  1. SSH to Primary Cell as root
  2. Retrieve the current synchronous_standby_names setting:

    grep synchronous_standby_names /var/vmware/vpostgres/current/pgdata/postgresql.conf

    If the Primary Cell requires a Standby Cell they will be listed by name in the output, for example the following represents the output for a cluster with one or two Standby Cells respectively:

    synchronous_standby_names = 'ANY 1 ("<standby_cell_name>")'
    synchronous_standby_names = 'ANY 1 ("<standby_cell_name>","<standby_cell_name>")'

    If the Primary Cell is not expecting any Standby Cells and is running as a Cloud Director appliance deployment without database HA an empty list will be returned:

    synchronous_standby_names = ''