Hi BirtyB,
The easiest way for this is to create a separate VM policy you can use for these type of scenarios.
I use VMware Intelligent Policies, and clients are selected on the fly based on a couple of criteria. The downside with these policies is that I cannot run a manual backup on just one of them, as all VM's will be backed up.
For these cases I have a separate policy VM_Test. This I can configure however I like, and I add a single client - either via a VIP query based on the VM's VMDNSname attribute, or selecting manually. You can have a full schedule in the policy with an accelerator forced rescan. Hopefully this will clear up your issue - maybe the change track log is screwed up.
I quite often use this method when I want to work with just one problem VM. I know it is not backing up under the same policy, but it does allow me to work with just the one VM.
Hope this helps,
Steve