Snowflake Connector for Spark release notes for 2026¶

This article contains the release notes for the Snowflake Connector for Spark, including the following when applicable:

Snowflake uses semantic versioning for Snowflake Connector for Spark updates.

Version 3.2.1 (July 8, 2026)¶

Restricted the fallback catalog to merge only immutable Spark configuration, so runtime and session settings can no longer override trusted catalog connection options.
Registered the connector’s sensitive DataSource options with Spark’s log redaction, so they are masked in logs, plan output, and the Spark UI.
Fixed identifier quoting to escape embedded quote characters.
Fixed CREATE STAGE construction to escape SQL metacharacters in interpolated values.
Added validation of the OAuth token request URL’s scheme and host.

The Snowflake Connector for Spark now requires Snowflake JDBC driver version 4.0.2 or later at runtime. If your environment pins a specific Snowflake JDBC driver version, upgrade it to 4.0.2 or later. The AWS SDK and Azure Storage SDK are now bundled (relocated) inside the connector JAR and are no longer pulled in as transitive dependencies.

Added support for Apache Spark 4.0 and 4.1 (Scala 2.13, JDK 17).
Added a Scala 2.13 cross-build for Spark 3.5.
Added VariantType support for reading and writing Snowflake VARIANT columns under Spark 4.x, including the Parquet write path.

Prevented silent data loss that could occur when speculative tasks are killed during a stage upload.
Added missing TimestampNTZType support in the Parquet write path.

Fixed the UnsupportedOperationException: Unexpected type: NullType error that was raised when writing DataFrames with structured columns (StructType) containing all null values via the Parquet write path.