Managed Service for Apache Spark: Confidential Compute, Executor Adjustments
Managed Service for Apache Spark now supports Confidential Compute for specific GPU machine types in cluster images 2.1, 2.2, and 2.3. Additionally, runtime 3.0 uses fewer executors by default, and all runtimes have a new listener exit timeout setting. These changes enhance security and optimize resource utilization for Spark workloads on Google Cloud.
- →Managed Service for Apache Spark
Features (1) ›
- Managed Service for Apache Spark Managed Service for Apache Spark
Managed Service for Apache Spark (formerly Dataproc on Compute Engine): The 2.1 , 2.2 and 2.3 cluster image versions now support Confidential Compute for the g4-standard-48 GPU machine type.
Enhancements (1) ›
- Managed Service for Apache Spark Managed Service for Apache Spark
Managed Service for Apache Spark (formerly Google Cloud Serverless for Apache Spark): The 3.0 runtime now uses fewer executors, as follows: 0 min executors for spark.dynamicAllocation.minExecutors property 1 min executor for spark.executor.instances and spark.dynamicAllocation.initialExecutors properties All runtimes now configure spark.scheduler.listenerbus.exitTimeout to 30 seconds.
https://docs.cloud.google.com/release-notes#July_13_2026
Related releases
- Cloud SDK 576.0.0: BigLake, GKE, Compute Engine updates Google Cloud release notes ·
- Terraform Google Provider v7.40.0: New Data Sources and Resources Terraform Google Provider Releases ·
- Managed Service for Apache Spark adds new subminor cluster image versions Google Cloud release notes ·
- Cloud SDK 575.0.0 Adds Agent Registry, AlloyDB Failover, and More Google Cloud release notes ·
- Terraform Provider for Google Cloud v7.39.0: New Resources and Compute Enhancements Terraform Google Provider Releases ·
- Managed Service for Apache Spark: New image versions, Iceberg, and Spark optimizations Google Cloud release notes ·