Unable to power on a VM when DSVA, SVM, or Guest Introspection VM is down

Products

• VMware NSX

Issue/Introduction

Symptoms:

• Unable to power on a VM when DSVA, SVM, or Guest Introspection VM is down
• DRS is enabled in the vSphere environment.

Environment

VMware NSX for vSphere 6.3.x
VMware NSX for vSphere 6.2.x

Cause

This issue occurs as the powering on VMs from the vSphere Web Client has a dependency on DRS, which relies on EAM to power Guest Introspection VMs. If EAM is unresponsive, it will cause issues with manual power ons via DRS, and with DRS initiated vMotions. If the service state is preventing it it will report "Invoking callbacks"

Resolution

The reason VM has failed to power on if the appliance of Trend Mircro Deep Security was in power off status in the same cluster is due to invocation of api – updateComputeResourceAgentInfo. which sets the number of required esx agent virtual machines to specific number.

So when a user invokes power-on of a vm, we iterate over a list of esx agent vm and check if they are ready. we check if the number of esx vm ready is equal to numRequiredEsxAgents.

By default vpxd.esxAgentVmsReadyMaxWaitTime is set to 10 minutes. VM will wait until someone powers on the esxagent vm or power on will then fail at 10th minute. Tweaking the Max Wait time will not help in powering on the VM. NSX won't let vm to be powered on if ESX agent vm are not running.

Adding power on compatible hosts to testtest: strictDrsCheck = true

DRS is used to do the precheck before VM is powered on. So, that is why VM is stuck at precheck process. Check VMware VirtualCenter Operational Dashboard to get the VM’s power-on precheck status.

https://<vcenter>/vod/index.html and go to “Thread States”

TASK::task-internal-53258315::vm-812::Drm.ExecuteVmPowerOnLRO::

To work around this issue, disable DRS. NSX will not be able to perform prechecks and VM gets powered on right away.