A VMware Cloud Foundation (VCF) administrator cannot deploy Virtual Machines (VMs) to a compute cluster. The administrator discovers that the vCLS…

vSphere Cluster Services (vCLS) is a mandatory feature for vSphere Distributed Resource Scheduler (DRS) to function. When the vCLS agent virtual machines (VMs) are powered off and cannot be started, DRS becomes non-operational. A non-functional DRS can prevent the initial placement of new VMs, thus blocking their deployment.

To resolve this and enable VM deployment, the administrator can place the cluster into "Retreat Mode." This is an advanced cluster setting that disables vCLS on that specific cluster. By disabling vCLS, the dependency for DRS is removed, allowing an administrator to proceed with operations like deploying VMs, albeit without the benefits of DRS. This is intended as a temporary measure while the underlying issue preventing the vCLS VMs from running is investigated and resolved.

A. Delete all resource pools in the affected cluster.

Deleting resource pools is a destructive action that does not address the core issue of non-functional vCLS VMs and is not the recommended procedure.

B. Disable HA on the affected cluster.

The primary dependency for vCLS is DRS, not vSphere High Availability (HA). Disabling HA will not resolve the issue preventing DRS from functioning.

D. Set DRS Automation level to fully automated.

Changing the DRS automation level is irrelevant because the DRS service itself is not running due to the powered-off vCLS VMs.

1. VMware vSphere Documentation. vSphere Cluster Services (vCLS). "vSphere DRS is a critical vSphere feature that is dependent on the vCLS VMs. DRS will not function if vCLS VMs are not running." This establishes the dependency that causes the problem.

Source: VMware vSphere 8.0 Documentation, "vCenter Server and Host Management," Chapter: "Managing Clusters," Section: "vSphere Cluster Services (vCLS)".

2. VMware Knowledge Base Article 80472. vSphere Cluster Services (vCLS) in vSphere 7.0 Update 1 and newer versions. This article explicitly describes Retreat Mode as the solution for disabling vCLS when necessary.

Section: "How to disable vCLS" states: "In some corner case scenarios where vCLS VMs cannot be deployed... you can disable the vCLS service for a specific cluster by navigating to the cluster's Advanced Settings... This is called Retreat Mode."

3. VMware vSphere Blog. vSphere 7 - vSphere Cluster Services (vCLS), by Niels Hagoort, October 6, 2020. This official blog post explains the procedure for disabling vCLS.

Section: "Disabling vCLS" details the steps to enter retreat mode: "To disable vCLS on a cluster, you need to go into the advanced cluster settings and add the setting config.vcls.clusters.domain-c.enabled and set the value to False."