As you may have noticed, the servers again went up and down over the past few days. We want to apologize for the recurring inconvenience.
We were able to identify the source of the problem, and implemented a fix earlier today. In doing so, we minimally altered all the PennController.js files hosted on the PCIbex Farm to deactivate the resource-fetching method that would cause excessive memory consumption.
You do not need to change anything in your experiments, in particular you can keep your resource files under chunk_includes. We made sure to also update the PennController.js files on our Github account, so you can safely use the sync method to update your projects.
One last note: please refrain from sending requests to server.py in your script. The more requests you send to server.py, the slower the PCIbex Farm servers will become for everyone.
Please let us know if you still experience runtime issues.