I am scratching my head for the last 2 days because of this issue. This error is intermittent on the production server as sometimes the task scheduler works and sometimes not.
The same settings work in the development server.
I also checked the execution policy on both servers and it looks the same.








In your second screenshot, you can choose "Stop existing instance" in the latest dropdown list (if the task is already running). Then the retry option might trigger your task again correctly.