A Destroy operation on a cluster member prevents Scale Out/Scale In from working as expected in vRA
search cancel

A Destroy operation on a cluster member prevents Scale Out/Scale In from working as expected in vRA

book

Article ID: 314941

calendar_today

Updated On:

Products

VMware Aria Suite

Issue/Introduction

Symptoms:

Day 2 scale in and scale out operations shows incorrect values when you have manually destroyed any of the deployments machines.

For example:

When you manually destroy a machine that is part of a multi-machine cluster, you can no longer perform a reliable day 2 operation with Scale Out/ Scale In. You introduce a count mismatch when you manually destroy one member of a cluster using the destroy operation on the machine. With a count mismatch, a scale out operation assumes that the destroyed machine is still part of the cluster. This prevents a scale out operation from adding some or all of the needed machines.

If the count is off by 1 machine and the cluster limit is 5, there can be at most 4 actual virtual machines and 1 phantom machine. For a scale in operation, the composition service might attempt to scale in to a single machine, resulting in the destruction of all cluster members.


Environment

VMware vRealize Automation 6.2.x
VMware vRealize Automation 7.0.x
VMware vRealize Automation 7.3.x
VMware vRealize Automation Desktop 6.2.x
VMware vRealize Automation 7.5.x
VMware vRealize Automation 6.2
VMware vRealize Automation Desktop 6.2
VMware vRealize Automation 6.x
VMware vRealize Automation 7.6.x
VMware vRealize Automation 7.1.x
VMware vRealize Automation 7.x
VMware vRealize Automation 7.4.x
VMware vRealize Automation 7.2.x

Resolution

To prevent the issue, do not entitle destroy operations for deployments where Scale Out/Scale In operations are enabled. This prevents a count mismatch. If you think your deployment has a machine in a cluster that was manually destroyed, an administrator can check by counting the number of cluster members that appear on the Deployments page. If there is a cluster that has a destroyed virtual machine, redeploy the deployment and do not entitle destroy operations on the redeployed deployment.


Additional Information

To be alerted when this document is updated, click the Subscribe to Article link in the Actions box in the Actions box.