# Troubleshooting AWS Clean Rooms
<a name="troubleshooting"></a>

This section describes some common issues that might arise when using AWS Clean Rooms and how to fix them.

**Topics**
+ [One or more tables referenced by the query is not accessible by its associated service role. The table/role owner must grant the service role access to the table.](#table-not-accessible)
+ [One of the underlying datasets has an unsupported file format.](#dataset-has-unsupported-file-format)
+ [Query results are not as expected when using Cryptographic Computing for Clean Rooms.](#troubleshoot-query-results-c3r)
+ [AWS Clean Rooms Spark SQL: Missing partition data](#troubleshoot-missing-partition-data)

## One or more tables referenced by the query is not accessible by its associated service role. The table/role owner must grant the service role access to the table.
<a name="table-not-accessible"></a>
+ Verify that the permissions for the service role are set up as required. For more information, see [Setting up AWS Clean Rooms](setting-up.md). ​ ​ 

## One of the underlying datasets has an unsupported file format.
<a name="dataset-has-unsupported-file-format"></a>
+ Ensure that your dataset is in one of the supported file formats:
  + Parquet 
  + RCFile
  + TextFile
  + SequenceFile
  + RegexSerde
  + OpenCSV
  + AVRO
  + JSON

  For more information, see [Data formats for AWS Clean Rooms](data-formats.md).

## Query results are not as expected when using Cryptographic Computing for Clean Rooms.
<a name="troubleshoot-query-results-c3r"></a>

If you are using Cryptographic Computing for Clean Rooms (C3R), verify that your query uses encrypted columns correctly: 
+ The sealed columns are only used in SELECT clauses.
+ The fingerprint columns are only used in JOIN clauses (and GROUP BY clauses under certain conditions).
+ That you are only JOINing fingerprint columns with the same name if the collaboration settings require it.

For more information, see [Cryptographic Computing for Clean Rooms](crypto-computing.md) and [Column types in Cryptographic Computing for Clean Rooms](crypto-computing-column-types.md).

## AWS Clean Rooms Spark SQL: Missing partition data
<a name="troubleshoot-missing-partition-data"></a>

All partitions in the AWS Glue Data Catalog must also have data in S3. The engine uses the Spark setting `spark.sql.files.ignoreMissingFiles=False`

For more information, see [https://spark.apache.org/docs/latest/sql-data-sources-generic-options.html\$1ignore-missing-files ](https://spark.apache.org/docs/latest/sql-data-sources-generic-options.html#ignore-missing-files).

If you encounter this error you will get the following error message: `"Missing partition data: One of the configured tables is partitioned and one or more of the partitions does not have data".`

Compare your data present in Amazon S3 with the partitions listed in the AWS Glue Data Catalog for the table. Delete partitions that don't have corresponding data in S3.