Remote interpreter: huge usage problems Follow

I am not sure if it is real problem or bug and I should report it on the bug tracker or I just misunderstand something.

I use remote python interpreter that is located on local virtual machine (vagrant). Project is located in shared folder (vagrant - host machine), so when I make any changes in local IDE I don't need to upload updated files in virtual machine.

I have 2 big really annoying problems:

1/ Stubs are cached! This means when I make change, I need go to interpreter settings and reload path - otherwise I have old code version in the stub! It takes time and I need to go to the settings each time.

2/ When I debug my code - in fact debugging is going in stub, not in real project file. So if I want add / remove breakpoint or edit something - I have to switch to original file each time (and then F7/F8 brings me back to the stub...). Naturally I ofter forget about switching back to the original file, and so my changes are lost... I need move them from stub to original file, then reload stubs (see 1 point)

Any advices?
thanks for you help

ps - pycharm 3.0.1, python 2.7.5. Host system is maс os x, guest is debian 7.