In VMware vCenter, when a new cluster created, the default HA configuration is disabled:
When opening the HA configuration for this cluster (Select cluster in inventory -> "Configure" tab -> vSphere Availability section -> "Edit" button for vSphere HA) we can see that HA is disabled and the response for "Datastore with PDL" and "Datastore with APD" is pre-defined as "Power off and restart VMs".
While this new cluster is still empty, if vShere HA is turned on from this configuration window, it unlocksconfiguration sections for PDL/APD response:
If the response under "Datastore with APD" was not changed to "Disabled" before adding host(s) to the cluster then, later when freshly imaged ESXi host(s) (with Misc.APDhandlingEnable = 0 set by foundation process) are added to cluster, the APD response configuration section is locked and cannot be changed to "Disabled" according our recommendations for standalone clusters.
As a result, attempting to enable HA on this cluster fails with the following message: “Cannot enable vSphere HA VM Component Protection for specified cluster, because it contains a host <name/IP> with APD Timeout is disabled.”
This is due to the selected "Power off and restart VMs" response for PDL or PAD conditions, which will require APD handling to be enabled on all ESXi hosts in the cluster (Misc.APDhandlingEnable =1).
There are two workarounds for this issue:
PREFERRED - When a new cluster still does not contain any ESXi hosts, it is possible to enable HA and change PDL/APD response to “Disabled”. Later when you add ESXi hosts (with APD handling disabled) to this cluster, APD response configuration will be locked as "Disabled" and will not prevent HA from being enabled on the cluster.
If ESXi hosts are already added to the cluster and HA cannot now be enabled, you can temporarily enable APD handling on hosts (by setting Misc.APDHandlingEnable to "1") to unlock the response configuration for APD failures. After this, you can change PDL/APD responses in HA cluster settings to "Disabled" and then enable HA on cluster. Later, you would need to disable APD handling on hosts (by setting Misc.APDHandlingEnable to "0") so as to remain within a supported configuration. See KB-3328 for instructions on how to change advanced parameters on ESXi and to find more details regarding the related ncc check esx_misc_settings_check.
More details - https://portal.nutanix.com/page/documents/kbs/details?targetId=kA00e000000brHPCAY