Well it is important to know when the services stopped, and if they stop at the same time everyday. The Windows Event View could help in this process.
We saw an issue recently where a customers CATALOG.Z was corrupted and every time their Pull Task ran the services stopped.
So just wondering if you have some type of Automation task running when the service stops.
I also recently saw a customer who's Eventparser Reporting Plugin for VSE was causing his services to stop randomly. Updating the VSE reporting extension resolved this issue.
Also you should look at the following logs for clues as to what part of ePO is causing the services to stop..
I would agree with the previous reply to your question. I have also seen this type behavior on an ePO server this week. Everytime a repository pull would start the application service would stop. It did not make a difference if it was a pull now or a scheduled pull task, HTTP or FTP. Like Ryan mentioned you can review some logs on the server to get a better idea of whats going on. I would add that if you put the orion log in debug mode while reproducing the issue you may see more information about the problem. Use KB 52369 to learn how to enable debug logging for orion. We saw the following which pointed us to rebuilding the master repository.
ERROR [http-8443-Processor23] action.CheckInPackageWizardAction - Unable to extract package file
We rebuilt the master repository and resolved this problem on the server. We used KB53736 for step by step instructions for rebuilding the master repository. You can access the articles I mentioned above at