Sometimes you can get the following error when enabling HA on a ESX 3.5 server.
cmd addnode failed for secondary node: Internal AAM Error – agent could not start.: Unknown HA error
The best way to resolve this problem is to disable both HA and DRS on the cluster.
Then first enable HA on the cluster. Everything should go fine now.
After HA is completly enabled on the cluster enable DRS on the cluster.