Please check out our migration guide on this topic.
Option one would be your best bet. You can set "spark.sql.sources.default" to parquet and "spark.sql.legacy.createHiveTableByDefault" to true on your cluster.
Option two would be to change your code.