In case of partitions this behavior is only applied on specific partition we are affecting during operation, which is useful in case of ETL workloads.
Recently I tried using airflow and inserted data every 15mins into external table from postgres to hive using INSERT OVERWRITE behavior and by setting hive.insert-existing-partitions-behavior=OVERWRITE as session property. It worked fine for one partition but as soon as new partition was added when the date changed, I saw duplicate data was added in the newly added partition as well. I will reconfirm this again later next week but meanwhile Could you please explain what do you mean by “only applied for specific partition”? Was this duplicate data because of some misconfiguration I did on running airflow job or trino is not aware about previously added partition and its data?