You can enable vSphere HA and Virtual SAN on the same cluster. As with traditional datastores, vSphere HA provides the same level of protection for virtual machines on Virtual SAN datastores. This level of protection imposes specific restrictions when vSphere HA and Virtual SAN interact.
ESXi Host Requirements
- The cluster's ESXi hosts all must be version 5.5 Update 1 or later.
- The cluster must have a minimum of three ESXi hosts. For best results, configure the Virtual SAN cluster with four or more hosts.
Networking Differences
Virtual SAN uses its own logical network. When Virtual SAN and vSphere HA are enabled for the same cluster, the HA interagent traffic flows over this storage network rather than the management network. vSphere HA uses the management network only when Virtual SAN is disabled.vCenter Server chooses the appropriate network when vSphere HA is configured on a host.
When a virtual machine is only partially accessible in all network partitions, you cannot power on the virtual machine or fully access it in any partition. For example, if you partition a cluster into P1 and P2, the VM namespace object is accessible to the partition named P1 and not to P2. The VMDK is accessible to the partition named P2 and not to P1. In such cases, the virtual machine cannot be powered on and it is not fully accessible in any partition.
The following table shows the differences in vSphere HA networking whether or not Virtual SAN is used.
Virtual SAN Enabled | Virtual SAN Disabled | |
---|---|---|
Network used by vSphere HA | Virtual SAN storage network | Management network |
Heartbeat datastores | Any datastore mounted to more than one host, but not Virtual SAN datastores | Any datastore mounted to more than one host |
Host declared isolated | Isolation addresses not pingable and Virtual SAN storage network inaccessible | Isolation addresses not pingable and management network inaccessible |
If you change the Virtual SAN network configuration, the vSphere HA agents do not automatically acquire the new network settings. To make changes to the Virtual SAN network, you must reenable host monitoring for the vSphere HA cluster by using vSphere Web Client:
- Disable Host Monitoring for the vSphere HA cluster.
- Make the Virtual SAN network changes.
- Right-click all hosts in the cluster and select Reconfigure HA.
- Reenable Host Monitoring for the vSphere HA cluster.
Capacity Reservation Settings
When you reserve capacity for your vSphere HA cluster with an admission control policy, this setting must be coordinated with the corresponding Primary level of failures to tolerate policy setting in the Virtual SAN rule set and must not be lower than the capacity reserved by the vSphere HA admission control setting. For example, if the Virtual SAN rule set allows for only two failures, the vSphere HA admission control policy must reserve capacity that is equivalent to only one or two host failures. If you are using the Percentage of Cluster Resources Reserved policy for a cluster that has eight hosts, you must not reserve more than 25 percent of the cluster resources. In the same cluster, with the Primary level of failures to tolerate policy, the setting must not be higher than two hosts. If vSphere HA reserves less capacity, failover activity might be unpredictable. Reserving too much capacity overly constrains the powering on of virtual machines and intercluster vSphere vMotion migrations. For information about the Percentage of Cluster Resources Reserved policy, see the vSphere Availability documentation.
Virtual SAN and vSphere HA Behavior in a Multiple Host Failure
After a Virtual SAN cluster fails with a loss of failover quorum for a virtual machine object, vSphere HA might not be able to restart the virtual machine even when the cluster quorum has been restored. vSphere HA guarantees the restart only when it has a cluster quorum and can access the most recent copy of the virtual machine object. The most recent copy is the last copy to be written.
Consider an example where a Virtual SAN virtual machine is provisioned to tolerate one host failure. The virtual machine runs on a Virtual SAN cluster that includes three hosts, H1, H2, and H3. All three hosts fail in a sequence, with H3 being the last host to fail.
After H1 and H2 recover, the cluster has a quorum (one host failure tolerated). Despite this quorum, vSphere HA is unable to restart the virtual machine because the last host that failed (H3) contains the most recent copy of the virtual machine object and is still inaccessible.
In this example, either all three hosts must recover at the same time, or the two-host quorum must include H3. If neither condition is met, HA attempts to restart the virtual machine when host H3 is online again.