Read Snowflake Iceberg tables from Databricks UC

[jeanptello]

Hi folks!
I'm trying to read Iceberg tables that I created in Snowflake from Databricks using catalog federation. I set up a connection to Snowflake, configured an external location pointing to the S3 folder that contains the Iceberg files, and used that to create a catalog.

The first time, I'm able to query the Iceberg table from Databricks successfully. However, if I update the table in Snowflake and then try to query it again from Databricks, I get the following error:

Failed to convert the table version 2 to the universal format iceberg. Clone validation failed - Size and number of data files in target table should match with source table. srcTableSize: 168818176, targetTableSize: 314383872 srcTableNumFiles: 25, targetTableNumFiles: 25 SQLSTATE: KD00E

The table contains around 5 million rows.

[Isi]

Hey @jeanptello 

It’s possible that there’s a mismatch between what Snowflake has written and what Databricks is trying to read. This often happens if Snowflake has performed an operation that rewrites the table files (like compaction or a bulk update), and then Databricks tries to validate an older version that no longer matches the current state.

I’d recommend:

• Refreshing the table from Databricks to ensure it’s accessing the most recent version.

• Checking the table history in both Snowflake and Databricks to verify whether both systems are seeing the same snapshot, or if there’s a discrepancy in recent operations (like file size or number of files).

The error likely stems from a difference between the current state of the table in Snowflake and what Databricks expects. Reviewing the history should help confirm if there was a shift or an operation that triggered this conflict.

Hope this helps, 🙂

Isi