Skip to main content

Job Cluster

Controls how Databricks job tasks are assigned compute resources.

ValueBehavior
Serverless (default)Uses Databricks serverless compute; the generated job tasks carry no cluster configuration and set disable_auto_optimization: true
A databricks.yml variable nameThe job defines a job cluster (job_cluster_key with new_cluster: ${var.<name>}) from the named complex variable: Spark version, node type, scaling and other cluster attributes; the sample databricks.yml includes such a variable
An existing cluster IDUsed when Use Existing Cluster is enabled; the tasks reference the cluster with existing_cluster_id

When to Change This Setting

Keep the default Serverless for the simplest operation with no cluster definitions to manage. Point it at a databricks.yml cluster variable when workloads need a specific cluster configuration, or pair it with Use Existing Cluster to run against an always-on cluster you manage yourself.

SettingInteraction
Use Existing ClusterSwitches this value from a cluster variable to an existing cluster ID
Utils DriverJDBC is the driver option compatible with serverless compute
Git SourceThe companion databricks.yml variable reference for notebook source

Notes

  • This setting is part of the Databricks settings category.
  • The default value for this setting is Serverless.