Control actions can be executed from the HQ portal in both the OSS and Enterprise version of the software. The main difference for Enterprise is that it can schedule control actions (run backup, run vacuum, run garbage collection etc) at a specific time in the future and can also run control actions in response to an alert condition (eg, helping you keep tomcat up by restarting the webapp while focusing on fixing your latest memory leak).
You can't reboot the operating system out of the box. It'd be a trivial plugin to write with JNI or if you're even lazier runtime.exec(), but I personally wouldn't do it. I wouldn't want system calls like that on my servers exposed through what is essentially a web service on the HQ Agent. All platforms have ways of doing that that require proper authentication and put the right information into the system's security logs. If you let an HQ agent do it all your OS will know is your platform's version of 'local daemon thread with system context told me to shut down.'