LSF Resource Connecter should allow start/stop of *existing* AWS VMs.

I see. You are effectively wanting to power on/off an existing host (instance) - this isn't what the resource connector is designed to do. It would be a little cumbersome as the RC could only ever create 1 instance from a given suspended instance.

However, it should be possible to leverage the existing power management functions to "power" on/off those VM's.

In lsb.resources, create a Power Policy

Begin PowerPolicy
NAME = policy_special_vm's
HOSTS = hostGroupofvms
MIN_IDLE_TIME= 600
CYCLE_TIME= 60
End PowerPolicy

In lsb.params

POWER_SUSPEND_CMD = cmd_to_suspend_the_vm
POWER_RESUME_CMD = cmd_to_resume_the_vm
POWER_RESET_CMD = cmd_to_reset_the_vm

These commands/scripts would be called with the hostname/vmname to suspend/resume

This policy would suspend the VM after being idle for 600 seconds.
It would be resumed if there was a demand for it - e.g. a job explicitly targeted to it.
The cycletime is the minimum time between it being suspended and resumed (and vice versa) to allow time for the image to write/resume state etc

I think this should achieve what you want.

That is correct. We already created a number of VMs in AWS that are "secured" (additional steps beyond just spinning up an AWS instance). We want to keep them in a stopped state when they are not in use. This way, we don't get charged by AWS for running instances that are sitting idle. What we would like to have is the ability for LSF Resource Connector to start those instances, run the desired jobs and then when they're done, put them back into stopped state. We still pay for storage whether the AWS instances are running or not, that's fine.

Do you mean that you have a number of existing VM's that have been pre-created and are currently suspended that you want the RC to resume/suspend rather than create/destroy?