Spark calling `count()` on DataFrame is causing stage to fail and application to crash

Question

I am processing data with spark and trying to get a distinct count on one of the dataframes. The data is read in from parquet files hosted in AWS S3.

I read in all th