Summary
The SDK adds AI Gateway workspace support, new methods for bundle deployment updates and serving endpoint telemetry configuration, plus numerous new fields across compute, jobs, and catalog services. Breaking changes include a revised create_deployment() signature and removal of deployment_id and lifetime fields.
Summary generated by brickster.ai. For the full changelog and any code/binary attachments, follow the GitHub link above.
More from databricks/databricks-sdk-py
The SDK adds new methods for PostgreSQL CDF configuration and clean room task runs, plus numerous fields across compute, ML, pipelines, and networking services. Breaking changes include revised IAM v2 method paths, internal_id type changes to string, and removal of code_source_path from AiRuntimeTask that may require code updates.
This release includes breaking changes that remove deprecated fields from secrets management and aggregation functions while making the PostgreSQL role field required. New capabilities include list() and list_effective() methods for grants management, expanded model serving workload types, and enhanced authentication documentation.
The SDK adds SQL conditions for job triggers, telemetry configuration for serving endpoints, Postgres endpoint management and credential enhancements, and Genie visualization support across multiple services. The release removes the name field from User objects as a breaking change.
The SDK adds cluster policy enforcement methods and fields, including a new cancel_pending_cluster_enforcement() method, and introduces AI runtime task support in jobs APIs. The replicate_workspace_assets field in WorkspaceSet is now optional instead of required.
The SDK adds Data API management methods for Postgres workspaces and introduces serverless compute support for Databricks Workflows pipelines. Vector search indexes now support endpoint IDs, Azure compute attributes can specify capacity reservation groups, and Dynamics365 is available as a new connection type.
Spark Connect clusters avoid crashes through lazy dbutils initialization and runtime fallback, while OIDC authentication performance improves via token caching that reduces latency and rate-limits. Type_overrides fields are added for synced tables and bundle operation resource_id is no longer required.