SDK Types¶

The decorator keyword arguments accept SDK-native dataclasses from databricks.bundles.jobs. Below is a complete reference of which classes are used by each config TypedDict, and where to import them.

Importing SDK Classes¶

All classes listed below are imported from a single module:

from databricks.bundles.jobs import CronSchedule, AutoScale, ClusterLogConf  # etc.

Classes Used by `JobConfig` (for `@job(...)`)¶

Field	Type	Description
`continuous`	`Continuous`	Continuous trigger configuration
`email_notifications`	`JobEmailNotifications`	Email notification settings for the job
`git_source`	`GitSource`	Git repository source configuration
`health`	`JobsHealthRules`	Health rules for the job
`lifecycle`	`Lifecycle`	Lifecycle management settings
`notification_settings`	`JobNotificationSettings`	General notification settings
`performance_target`	`PerformanceTarget`	Performance target level
`queue`	`QueueSettings`	Queue settings for concurrent runs
`run_as`	`JobRunAs`	Service principal or user to run the job as
`schedule`	`CronSchedule`	Cron-based schedule
`trigger`	`TriggerSettings`	File-arrival or table-update triggers
`webhook_notifications`	`WebhookNotifications`	Webhook notification destinations

Primitive fields (budget_policy_id, description, max_concurrent_runs, tags, timeout_seconds, usage_policy_id) accept plain Python types (str, int, dict[str, str]) — no SDK import needed.

Classes Used by `TaskConfig` (for `@task(...)`)¶

Field	Type	Description
`compute`	`Compute`	Compute configuration override
`email_notifications`	`TaskEmailNotifications`	Email notification settings for the task
`health`	`JobsHealthRules`	Health rules for the task
`new_cluster`	`ClusterSpec`	Inline new cluster spec (prefer `@job_cluster` instead)
`notification_settings`	`TaskNotificationSettings`	General notification settings
`run_if`	`RunIf`	Conditional execution enum
`webhook_notifications`	`WebhookNotifications`	Webhook notification destinations

Primitive fields (description, disable_auto_optimization, disabled, environment_key, existing_cluster_id, max_retries, min_retry_interval_millis, retry_on_timeout, timeout_seconds) accept plain Python types — no SDK import needed.

Classes Used by `ClusterConfig` (for `job_cluster(...)`)¶

Field	Type	Description
`autoscale`	`AutoScale`	Autoscaling worker range
`aws_attributes`	`AwsAttributes`	AWS-specific cluster attributes
`azure_attributes`	`AzureAttributes`	Azure-specific cluster attributes
`cluster_log_conf`	`ClusterLogConf`	Cluster log delivery configuration
`data_security_mode`	`DataSecurityMode`	Data security mode enum
`docker_image`	`DockerImage`	Custom Docker container image
`driver_node_type_flexibility`	`NodeTypeFlexibility`	Driver node type flexibility
`gcp_attributes`	`GcpAttributes`	GCP-specific cluster attributes
`init_scripts`	`list[InitScriptInfo]`	Init script configurations
`kind`	`Kind`	Cluster kind enum
`runtime_engine`	`RuntimeEngine`	Runtime engine enum
`worker_node_type_flexibility`	`NodeTypeFlexibility`	Worker node type flexibility
`workload_type`	`WorkloadType`	Workload type configuration

Primitive fields (apply_policy_default_values, autotermination_minutes, cluster_name, custom_tags, driver_instance_pool_id, driver_node_type_id, enable_elastic_disk, enable_local_disk_encryption, init_scripts, instance_pool_id, is_single_node, node_type_id, num_workers, policy_id, remote_disk_throughput, single_user_name, spark_conf, spark_env_vars, spark_version, ssh_public_keys, total_initial_remote_disk_size, use_ml_runtime) accept plain Python types — no SDK import needed.

Nested Classes (used inside the above)¶

These classes are commonly needed when constructing the top-level objects:

Class	Used inside	Purpose
`DbfsStorageInfo`	`ClusterLogConf`	DBFS log destination
`S3StorageInfo`	`ClusterLogConf`	S3 log destination
`GcsStorageInfo`	`ClusterLogConf`	GCS log destination
`VolumesStorageInfo`	`ClusterLogConf`	Unity Catalog Volumes log destination
`DockerBasicAuth`	`DockerImage`	Docker registry credentials
`JobsHealthRule`	`JobsHealthRules`	Individual health rule
`Webhook`	`WebhookNotifications`	Individual webhook destination
`SparseCheckout`	`GitSource`	Sparse checkout configuration
`FileArrivalTriggerConfiguration`	`TriggerSettings`	File arrival trigger
`TableUpdateTriggerConfiguration`	`TriggerSettings`	Table update trigger
`PeriodicTriggerConfiguration`	`TriggerSettings`	Periodic trigger
`Library`	`@job(libraries=...)`	Library dependency (whl, jar, pypi)
`PythonPyPiLibrary`	`Library`	PyPI package specification
`MavenLibrary`	`Library`	Maven artifact specification
`JobParameterDefinition`	(internal)	Job parameter — prefer `@job(params={"key": "default"})`

Complete Import Example¶

from databricks.bundles.jobs import (
    # Job-level
    CronSchedule,
    JobEmailNotifications,
    JobRunAs,
    QueueSettings,
    TriggerSettings,
    FileArrivalTriggerConfiguration,
    WebhookNotifications,
    Webhook,
    # Cluster-level
    AutoScale,
    ClusterLogConf,
    DbfsStorageInfo,
    DockerImage,
    DockerBasicAuth,
    InitScriptInfo,
    Library,
)

For the underlying TypedDict source definitions, see Internals > SDK Types.