This document lists supported and unsupported Serverless for Apache Spark runtime versions and describes their characteristics.
Runtime node location
This section describes how Google Cloud Serverless for Apache Spark runtime versions allocate nodes within a region to execute batch workloads and interactive sessions.
Runtime versions
3.0+: To help with obtainability and avoid zonal stockouts, Serverless for Apache Spark might allocate nodes to execute a batch workload or interactive session in multiple zones within the workload or session region. There is no additional cost for cross-zonal network traffic between nodes in a region.Runtimes versions prior to
3.0, Serverless for Apache Spark allocates nodes to execute a batch workload or interactive session in a single zone within the workload or session region.
Runtime availability and support
Serverless for Apache Spark LTS (Long-Time-Support) runtime versions are supported for 30 months. Serverless for Apache Spark non-LTS runtime versions are supported for 12 months.
Serverless for Apache Spark runtime versions continue to be available for two years after their end-of-support date.
Runtime subminor version pinning is not supported.
Supported Serverless for Apache Spark runtime versions
| Version | Last updated | Released on | Supported until | Available until | Notes |
|---|---|---|---|---|---|
| Spark runtime 3.0 | 2025/10/16 | 2025/09/08 | Preview release. | ||
| Spark runtime 2.3 LTS | 2025/10/06 | 2025/05/28 | 2027/11/26 | 2029/11/26 | General Availability release. |
| Spark runtime 2.2 LTS | 2025/09/11 | 2024/03/27 | 2026/09/30 | 2028/09/30 | General Availability release. This is the default version. |
| Spark runtime 1.2 LTS | 2025/09/11 | 2024/03/27 | 2026/09/30 | 2028/09/30 | General Availability release. |
How to choose a Serverless for Apache Spark runtime version
The current default Serverless for Apache Spark runtime version is listed in Supported Serverless for Apache Spark runtime versions. You can choose a different major.minor runtime version when you submit a batch workload or create a interactive session or session template.
Unsupported Serverless for Apache Spark runtime versions
The following Serverless for Apache Spark versions are unsupported.
| Version | Includes | Released On | Last Updated | Available Until | Notes |
|---|---|---|---|---|---|
| Spark runtime 2.1 | Apache Spark 3.4.0 Cloud Storage connector 2.2.20 BigQuery connector 0.28.1 Java 17 Conda 23.3 Python 3.11 R 4.2 Scala 2.13 | 2022/01/19 | 2024/5/30 | 2026/08/28 | Unsupported as of 2024/05/30. 2.1.50 was the final released version. |
| Spark runtime 2.0 | Apache Spark 3.3.4 Cloud Storage connector 2.2.20 Java 17 Conda 22.9 Python 3.10 R 4.1 Scala 2.13 | 2022/10/21 | 2024/08/22 | 2026/08/28 | Unsupported as of 2024/08/22. 2.0.84 was the final released version. |
| Spark runtime 1.1 | Apache Spark 3.3.4 Cloud Storage connector 2.2.28 BigQuery connector 0.28.1 Java 11 Conda 22.11 Python 3.10 R 4.2 Scala 2.12 | 2023/01/27 | 2025/07/31 | 2027/07/31 | Unsupported as of 2025/07/31. 1.1.111 was the final released version. |
| Spark runtime 1.0 | Apache Spark 3.2.3 Cloud Storage connector 2.2.11 Conda 4.11 Java 11 Python 3.9 R 4.1 Scala 2.12.17 | 2022/01/19 | 2023/2/03 | 2025/05/30 | Unsupported as of 2023/02/03. 1.0.29 was the final released version. |