I have run into this in my lab and its due to the time offset between the various components.  Make sure all ESXi hosts and appliances are using a valid NTP host, if using Windows as your NTP source see http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=103583...

The Windows bits at the top are rather important, though if you don't want to mess with your DCs (likely what is running NTP) setup a dedicated linux server to do so.  Also you can SSH to your appliances and run date to ensure the time is correct and in sync with your SSO server - Windows VC or VCSA for example and your NTP server.  Even a small offset of under a minute it seems can break the ability to log in.

You can attempt to force the vCAC appliance to sync its time using sntp –P no –r 192.168.1.10