I was teaching my first course last week on the new source code (V3.5). I was suprised to see the message about insufficent resources to power on a VM and HA Failover – when our cluster had 3 ESX hosts and were running very small VMs. This message was kicking far sooner than you would normally expect. I raised this as support issue with VMware Education – and learnt something about HA in the new version.
There’s a new setting called das.vmMemoryMinMB) that is set to 256 MB by default for all VMs in an HA cluster. Taking this amount and the overhead for each VM, we’d run out of memory resources. You can learn more about the HA settings (which are located in the Advanced Settings section for each ESX Host) here