You may notice that cluster(s) in vCenter 7 display a message stating the health has degraded due to the unavailability of vSphere Cluster Service (vCLS) VMs. To learn more about the purpose and architecture of vCLS, please see the following documentation and blog. In the below demo, we demonstrate how to troubleshoot and correct this state automatically with vCenter's "Retreat Mode."
At a high level, "Retreat Mode" tells vCenter to power off and remove all vCLS VMs for the specified cluster.
Important: This is just for troubleshooting purposes and is not intended for long-term operation.
Once a cluster has been configured for "Retreat Mode," virtual machine workloads will continue to operate, but cluster features such as vSphere High Availability (vSphere HA) and Distributed Resource Scheduler (DRS) will be degraded until the vSphere Cluster Service VMs have been successfully re-deployed. After observing the vCLS VMs have been removed from the given cluster, disable "Retreat Mode" so that vCenter can re-deploy the vCLS VMs and restore the health of the cluster's services.
The in-depth explanation for using "Retreat Mode" can be found in this knowledge-base article.
Simulations can be navigated by clicking the orange box hot-spots, or by using your keyboards left and right arrow keys
(Click HERE to open the simulation walkthrough in a new window)