Glue ETL script, Java keeps running out of memory

Hi all, I'm relatively new to Glue, but I've got a Python ETL script that I've built that works pretty well. It reads two CSV files into dataframes and then unions them together into one normalized dataframe. I then use that to create other dataframes I write to a MySQL database using JDBC. This works well when the DB is empty (in my Sandbox), but if I run it a second time with a different set of data files, it runs for an hour and then reports in cloudwatch that Java is out of memory. The input files aren't particularly large, ~ 55K lines of CSV data. When I run the script with the second set of input data and watch the metrics of the job in the AWS Console I see only one executor running, which eventually uses all the resources and is removed for lack of memory. I also see no workers running at all. At this point the system just seems to hang for a hour till it ends, I guess for a timeout reason? Can anyone help me understand what's going on and how to address this problem? Thanks! Doug