Datasets with more than 1 question don't work

I try to use my own dataset while evaluating a QA system, however, after I configure and start the experiment nothing happens -- "The experiment is still running.", but the system doesn't receive the questions.

Occasionally, I tried to upload the dataset with only one question in it -- it worked, but I still want to evaluate on all the data I have.

I use this endpoint: http://gerbil-qa.aksw.org/gerbil/config

Start by reproducing the problem at the reported configuration endpoint with datasets containing one and multiple questions, and compare the experiment status and system behavior. Trace the dataset upload and experiment execution path to find where multi-question input stops progressing; done means all questions are delivered and the experiment completes.

Written by the indexing model from the issue text.