HIVE_CURSOR_ERROR: Failed to read Parquet file:

Question

Getting HIVE_CURSOR_ERROR: Failed to read Parquet file: s3:////output.parquet I have created an external table on Parquet files generated by Pandas & Pyarrow library in Python. If I execute SELECT count(*) FROM it's giving me correct output But If I execute SELECT * FROM

//output.parquet'

It's throwing the above exception What's going wrong ? The Athena table structure is below . CREATE EXTERNAL TABLE `mmx_india_parquet_edr_part`( `uid` string COMMENT 'from deserializer', `customerid` string COMMENT 'from deserializer', `productid` string COMMENT 'from deserializer', `edrmode` string COMMENT 'from deserializer', `destinationid` string COMMENT 'from deserializer', `protocolid` string COMMENT 'from deserializer', `host` string COMMENT 'from deserializer', `servicetype` string COMMENT 'from deserializer', `t1` bigint COMMENT 'from deserializer', `t2` bigint COMMENT 'from deserializer', `status` string COMMENT 'from deserializer', `ruleid` string COMMENT 'from deserializer', `supplierid` string COMMENT 'from deserializer', `commandstatus` string COMMENT 'from deserializer', `httpstatus` string COMMENT 'from deserializer', `messageid` string COMMENT 'from deserializer', `fragmented` string COMMENT 'from deserializer', `fragnumber` string COMMENT 'from deserializer', `fragtotal` string COMMENT 'from deserializer', `messagestate` string COMMENT 'from deserializer') PARTITIONED BY ( `traffic_date` date) STORED AS PARQUET LOCATION 's3://

Answer

Hello,

This error normally indicates there’s a mismatch between the table DDL and the actual underlying data. You can use this [parquet tool](https://pypi.org/project/parquet-tools/) to inspect parquet file schema. If the file is quite large, you may also want to test on a small sample of the data first, in case there’s mixed data types for one column.

Apart from that, you can also test on different Athena engine versions to see if this is caused by some engine specific behaviours. (Athena engine V2 uses Presto and V3 uses Trino). [This doc](https://docs.aws.amazon.com/athena/latest/ug/engine-versions-changing.html#engine-versions-changing-changing-the-engine-version) shows how to change the engine version under a workgroup.

HIVE_CURSOR_ERROR: Failed to read Parquet file:

Relevant content