MSCS Clusters can fail or crash when adding additional hardware to the active cluster member

Products

• VMware vSphere ESXi

Issue/Introduction

Symptoms:

• Cluster crashes or becomes inconsistent after adding additional hardware to an active MSCS cluster member (primary node)

Environment

VMware vSphere ESXi 6.5
VMware vSphere ESXi 6.7
VMware vSphere ESXi 6.0

Cause

Hot-adding new hardware to an Active virtual machine, causes a "stun" operation (shown in the logs as such):

2019-04-18T15:56:50.970Z| vmx| I125: Migrate: VM starting stun, waiting 100 seconds for go/no-go message.
2019-04-18T15:56:50.977Z| vcpu-0| I125: Migrate: VM successfully stunned.

This message is normally tolerated by the majority of virtual machine configurations, but Microsoft Clusters are highly latency sensitive, and this sort of operation violates the latency constraints of that configuration.

Resolution

Hot-adding hardware to MSCS is not supported. This is expected Behavior.

Workaround:

1. Schedule appropriate maintenance for MSCS cluster members in situations where you require adding additional hardware.
2. Fail the cluster over to the Secondary member, power down the first member and add hardware while the primary virtual machine is powered off.

Additional Information

Impact/Risks:
Executing "hot-add" on active MSCS cluster members can result in downtime for whatever application they are serving.