AWS Glue Schema mismatch

We are trying to use Glue to query and aggregate some Parquet files in S3.

We get this error related to schema mismatch:

An error occurred while calling o106.pyWriteDynamicFrame. org.apache.spark.sql.execution.datasources.SchemaColumnConvertNotSupportedException

We did notice that one of the columns in some of the parquet files was sometimes read as int or binary instead of string.

I tried changing the schema using both the Glue visual tool and a Spark job in Glue ETL. I even tried dropping that column. But no matter what I do I get that error and it feels like there is no way to change this in the Glue ETL

Any help with fixing this or getting the script to ignore the mismatch (since we don't even need the field) would really help! Thanks in advance!