In Large Environments hosted on Nutanix Hypervisor, After MCS Machine Update, VDAs do not come up automatically.

In Large Environments hosted on Nutanix Hypervisor, After MCS Machine Update, VDAs do not come up automatically.

book

Article ID: CTX233530

calendar_today

Updated On:

Description

In Large Environments (1000+)  hosted on Nutanix Hypervisor, after MCS Machine Update, Desktop VDA's did not come up automatically.

It took more than 2 hours for the VDAs to come up after reboot.

The Rollout strategy was set to "Immediately" with "distribution time" set to 30 mins.

Rollout

Hosting connection was set to 50 new power actions per minute.

Hypervisor


 

Resolution

WORKAROUND

As a workaround we suggested to increase the "Distribution time" in Machine Update Rollout strategy to give ample time to shutdown the VMs while spreading it out evenly over a period of time.

We also suggested to reduce the new power actions per minute under Hosting Connections's advanced setting so that the resources would not have been stretched beyond its capacity. 

SOLUTION

We suggested engaging Nutanix and they provided new Nutanix Hypervisor AOS 5.1.2 that has scalability enhancements to address this issue.

 

Problem Cause

From the CDF Traces collected on Delivery Controller,  we found that Nutanix plugin was unable to offer bulk power state updates, and also there was a huge delay in carrying out power ops (shutdown in this case case taking more than 10 minutes) . Updating or even shutting down 1000+ machines in 30 with 50 new power actions to be processed per minute by the hypervisor seemed to be an overzealous target. 
 

Issue/Introduction

In Large Environments hosted on Nutanix Hypervisor, After MCS Machine Update, VDAs do not come up automatically.