Option	Value	Description
Host monitoring	Turned on	Uses network heartbeat to determine the status of hosts participating in the cluster, and whether a corrective action is required, such as restarting a VM on a different host from the cluster
Response for Host Isolation	Power off restart VMs	Determines what happens to the VMs on an isolated host, that is a host that cannot communicate to other hosts in the cluster nor reach the isolation response IP address. You configure this setting to Power off and restart, because a clean shutdown is not possible on an isolated host, and the access to the vSAN Datastore and the ability to write to disks is lost.

The compute capacity in a vSAN stretched cluster is typically equally divided between the two sites in the cluster. You need to ensure that all the VMs on site 1 can be restarted on site 2 when an entire site goes down. To achieve this, you reserve 50% of the cluster's capacity for failover to allow restart of all the VMs during site failure.

With this setting, vSphere HA ensures that 50% of cluster's aggregate CPU and memory resources are reserved for failover.

In addition to network heartbeating, vSphere HA provides datastore hearbeating for determining the state of the hosts in the cluster. However, a vSAN datastore cannot be used for heartbeating and you must deactivate this function for the vSAN datastore.

To not use datastore heartbeating for the vSAN datastore, select Use datastores only from the specified list and make sure that the vSAN datastore is not selected. You will still see an alert that there are less than the required heartbeat datastores, you can disable that alarm by configuring das.ignoreInsufficientHbDatastore = true.

When vSphere HA is enabled on a vSAN cluster, HA uses network heartbeat to validate the state of an ESXi host.

In a vSAN environment, vSphere HA uses the vSAN traffic network for communication. This is different from traditional vSphere environments where vSphere HA uses the management network for communication. However, even in a vSAN environment, vSphere HA continues using the default gateway of the management network for isolation detection responses. For this reason, you must configure isolation response IP addresses that reside on the vSAN network to allow HA to react to a vSAN network failure and trigger host isolation response.

It is recommended that you specify two additional isolation response addresses, and each of these addresses should be site-specific - one isolation address residing in site 1 and the other residing in site 2. This enables vSphere HA to validate host isolation even in the case of network failure between sites. Use the following settings for additional isolation response addresses:

• das.isolationaddress0. Set the value to an IP address on the vSAN network which resides on site 1.
• das.isolationaddress1. Set the value to an IP address on the vSAN network that resides on the site 2.
• das.usedefaultisolationaddress. Set to false.

In case of host or site failure, a possibility exists that the resources are insufficient to restart all the VMs in the vSAN cluster. Therefore, you have to ensure that certain VMs that are of high importance, such as the Supervisor control plane VMs, are restarted first. To do so, configure vSphere HA restart priority to individual VMs, as follows:

• Highest - Supervisor control plan VMs, NSX Edge VMs, and NSX Advanced Load Balancer Service engine VMs.
• High - all TKG cluster control plane VMs.
• Medium - all TKG cluster worker VMs.