I use PyCharm version 2018.3.2 build #PY-183.4886.43 (just upgraded to 2018.3.3 #PY-183.5153.39, same issue)
I have a local git repository loaded as a PyCharm project. This repository is deployed to a remote machine. The remote (ssh) interpreter used by the PyCharm project uses this code. This setup worked fine for a long time, but recently I've been having problems every single day.
When I try running e.g. unittests from PyCharm (rightclick -> Run Unittests in file), I get the following errors:
Can't run remote python interpreter: Error connecting to remote host xxx
Couldn't upload helpers for remote interpreter: Auth cancel
However, when I change code, changes do get automatically pushed toward this same remote host with no problem. I can also browse to the virtual environment on the remote host (through PyCharm). Usually (almost all of the behavior I describe seems completely random and arbitrary) I can even open the python console and it will run the remote interpreter. Yet when I try running a file it refuses to work.
When I go to settings -> project interpreter, the selected interpreter and mappings are correct. However, under the package list it states "Nothing to show."
I have been unable to find out the steps to fixing this problem. I usually get it to work eventually through a process of repeatedly restarting, invalidating cache, creating a new ssh interpreter configuration that is -exactly- the same as the previous one (this fixes the "Nothing to show." but it still doesn't run files), etc. If I try this for long enough it tends to end up working but I can't find a pattern to it. It seems completely arbitrary and I fail to see why it can so clearly connect to the interpreter and the remote machine in some ways, but can't in others.
PyCharm is such a great tool to work with but sometimes it's infuriating due to its completely unexplainable and intransparent behavior.
Thanks for any help! Let me know if more information is needed.