As you know, HA is triggered on an ESX host after 15 seconds when it cannot contact it's cluster partners and the default gateway. By Default, all virtual machines running on the ESX host are shut down and restarted on the remaining ESX hosts in the cluster (using a Legato AAM agent to do it's job).
The only thing you could do up until now was changing the default gateway (via the das.isolationaddress option/value set). However, for most companies this is not enough. When a general network maintenance is scheduled, you should disable HA to avoid an unnecessary HA trigger.
New in VirtualCenter 2.0.2 are the following option/value pairs:
- das.failuredetectiontime = <value> (default 15 seconds)
- das.isolationaddress2 = <value>
These values can be set under Your Cluster => Edit Settings => VMware HA => Advanced Options.
See also VMware's HA_Tech_Best_Practices.pdf document for more info!