Athena to_iceberg overwrite_partitions without partition as primary key

[Max0u]

Hello, I am using to_iceberg to drop dataframe to our iceberg table. This table is composed of 2 partitions:

• _batch_ts
• _event_ts

I would like to override the previous batch, naturally I've done

        wr.athena.to_iceberg(
            df=df,
            database=config.load_database,
            table=config.load_table,
            temp_path='s3://bucket/tmp_upload/',
            boto3_session=lakehouse_aws_credentials.get_session(),
            mode="overwrite_partitions",
            workgroup="primary",
            partition_cols=["_batch_ts"],
            dtype={
                "_batch_ts": 'timestamp',
                "_event_ts": 'timestamp',
                "data": 'string',
            }
        )

The problem is that we have more than one line per partition and the following query is failling

MERGE INTO "raw"."table" target
USING "raw"."temp_table_ea8727d33972439b8f72b0c11ab5ddc4" source
ON target."_batch_ts" = source."_batch_ts" AND target."_event_ts" = source."_event_ts"
WHEN MATCHED THEN
DELETE

with the following error MERGE_TARGET_ROW_MULTIPLE_MATCHES: One MERGE target table row matched more than one source row

How can I avoid this?