Known Issues Iceberg
Learn about the known issues in Iceberg, the impact or changes to the functionality, and the workaround.
- CDPD-55243: Spark can read stale data under certain conditions
-
Using inconsistent cases for database and table names of Iceberg tables in queries can lead to Spark reading stale data. In particular, after any write to the table (append, update, delete) in the same session, an old snapshot may still be read, unless all queries use consistent casing for database and table names.
- CDPD-84220: Cannot query Iceberg tables
- You cannot query existing Iceberg tables after you enable HDFS HA. This is because Iceberg stores the table path in the manifest files differently depending on whether the HDFS HA is enabled or not. After you enable HDFS HA, you might not be able to query the tables created prior to you enabling HDFS HA.
