vCenter Server might not choose a datastore that you specify as a preference for vSphere HA storage heartbeating.
You can specify the datastores preferred for storage heartbeating, and based on this preference, vCenter Server determines the final set of datastores to use. However, vCenter Server might not choose the datastores that you specify.
This problem can occur in the following cases:
The specified number of datastores is more than is required. vCenter Server chooses the optimal number of required datastores out of the stated user preference and ignores the rest.
A specified datastore is not optimal for host accessibility and storage backing redundancy. More specifically, the datastore might not be chosen if it is accessible to only a small set of hosts in the cluster. A datastore also might not be chosen if it is on the same LUN or the same NFS server as datastores that vCenter Server has already chosen.
A specified datastore is inaccessible because of storage failures, for example, storage array all paths down (APD) or permanent device loss (PDL).
If the cluster contains a network partition, or if a host is unreachable or isolated, the host continues to use the existing heartbeat datastores even if the user preferences change.
Verify that all the hosts in the cluster are reachable and have the vSphere HA agent running. Also, ensure that the specified datastores are accessible to most, if not all, hosts in the cluster and that the datastores are on different LUNs or NFS servers.