Resiliency improvements for vRA database clustering in 3VA / "SYNC" mode (automatic-DB-failover enabled)

Products

• VMware Aria Suite

Issue/Introduction

Symptoms:
Upon failover, one or more nodes of the system may become in "N/A" state resulting in systems outage.

Environment

VMware vRealize Automation 7.3.x
VMware vRealize Automation 7.4.x

Resolution

This issue is resolved in vRealize Automation versions 7.4 and later.

Workaround:
To work around this issue in vRealize Automation 7.3.x, update the attached RPM file for psql manager to the nodes:

1. Log in to VAMI on either VA node.
2. Navigate to vRA Settings > Database tab.
3. Click Async mode button. Wait for it to complete.
   
   Note: If the button reads Sync mode - do not do anything!! We need DB cluster to be in Async mode (e.g. manual-failover mode).

4. Open SSH connection to all VA nodes.
5. Stop the psql-manager service on each node by running the command:
   
   $ service psql-manager stop  

6. Download the attached updated RPM file for psql manager to the nodes and execute the command:

vRealize Automation 7.3.x:

$ rpm -U psql-manager-1.0.0.0-8686561.noarch.rpm

Note: if update fails, remove the current version and re-install the new one with rpm:

$ rpm -ev <old_package_name>

$ rpm -ivv <new_rpm_package_file>

 

$ service psql-manager start

7. After the package upgrade, the psql-manager service might need to be started by running the command (all nodes):

service psql-manager restart

8. Enable automatic failover again:
   1. Log in to VAMI page on either VA node.
   2. Navigate to vRA Settings > Database tab.
   3. Choose Sync mode.

Alternatively, to workaround the issue, disable automatic failover on the vRA cluster.

To disable automatic failover on the vRA cluster:

1. Log in to VAMI page on either VA node.
2. Navigate to vRA Settings > Database tab.
3. Switch database to Async mode.