Summary
This release adds effective entitlements tracking to workspace assignment details and serverless compute ID support for job clusters. It also updates model serving telemetry configurations to include table names and telemetry profile IDs.
Summary generated by brickster.ai. For the full changelog and any code/binary attachments, follow the GitHub link above.
More from databricks/databricks-sdk-java
Bundle deployment APIs received breaking changes that reorder required arguments in the createDeployment method and remove the deploymentId field from create requests. The catalog service adds new special destination owner enum values for Unity Catalog schemas, tables, volumes, functions, and registered models.
The Databricks Java SDK now includes a dependencyMode field across cluster compute configurations, along with new fields for AI runtime tasks, ML schema configs, and Postgres synced tables. This release also introduces a breaking change by removing the lifetime field from the ML TimeWindow class.
Pipeline connector options now include support for Reddit Ads configurations. The window duration field is now optional across ML rolling, sawtooth, and sliding window classes.
This release introduces breaking API path updates to workspace assignment and identity resolution methods in the accountIamV2 and workspaceIamV2 services, along with replacing the longRolling field with sawtooth in
This release introduces a breaking change by removing the codeSourcePath field from AiRuntimeTask in the Jobs service. Additional updates include support for Clean Rooms JAR analysis and task run handling, JDBC Unity Catalog connection types, and new fields across bundle deployments, compute instance pools, and pipelines.