Network Connectivity Is Lost in the Virtual SAN Cluster

When the connectivity between the hosts in the cluster is lost, Virtual SAN determines the active partition and rebuilds the components from the isolated partition on the active partition if the connectivity is not restored.

Component Failure State and Accessibility

Virtual SAN determines the partition where more than 50 percent of the votes of an object are available. The components on the isolated hosts are marked as absent.

Behavior of Virtual SAN

Virtual SAN responds to a network failure in the following way:

Parameter	Behavior
Number of failures to tolerate	If the Number of failures to tolerate in the VM storage policy is equal to or greater than 1, the virtual machine objects are still accessible from another ESXi host in the cluster. If resources are available, Virtual SAN starts an automatic reprotection. If the Number of failures to tolerate is set to 0, a virtual machine object is inaccessible if the object's components are on the isolated hosts.
I/O operations on the isolated hosts	Virtual SAN stops all running I/O operations for 5-7 seconds until it re-evaluates whether an object is still available without the failed component. If Virtual SAN determines that the object is available, all running I/O operations are resumed.
Rebuilding data	If the host rejoins the cluster within 60 minutes, Virtual SAN synchronizes the components on the host. If the host does not rejoin the cluster within 60 minutes, Virtual SAN examines whether some of the other hosts in the cluster can satisfy the requirements for cache, space and placement rules for the objects on the inaccessible host. If such a host is available, Virtual SAN starts the recovery process. If the host rejoins the cluster after 60 minutes and recovery has started, Virtual SAN evaluates whether to continue the recovery or stop it and resynchronize the original components.

Parameter

Behavior

Number of failures to tolerate

If the Number of failures to tolerate in the VM storage policy is equal to or greater than 1, the virtual machine objects are still accessible from another ESXi host in the cluster. If resources are available, Virtual SAN starts an automatic reprotection.

If the Number of failures to tolerate is set to 0, a virtual machine object is inaccessible if the object's components are on the isolated hosts.

I/O operations on the isolated hosts

Virtual SAN stops all running I/O operations for 5-7 seconds until it re-evaluates whether an object is still available without the failed component.

If Virtual SAN determines that the object is available, all running I/O operations are resumed.

Rebuilding data

If the host rejoins the cluster within 60 minutes, Virtual SAN synchronizes the components on the host.

If the host does not rejoin the cluster within 60 minutes, Virtual SAN examines whether some of the other hosts in the cluster can satisfy the requirements for cache, space and placement rules for the objects on the inaccessible host. If such a host is available, Virtual SAN starts the recovery process.

If the host rejoins the cluster after 60 minutes and recovery has started, Virtual SAN evaluates whether to continue the recovery or stop it and resynchronize the original components.