Databricks Asset Bundles began automatically migrating Terraform state to the direct engine on clean deploys.

August marked a major architectural transition as Databricks Asset Bundles shifted default deployments away from Terraform [16]. The CLI v1.14.0 release introduced automatic migration from Terraform state to the direct deployment engine during clean deploys, allowing teams to opt out by specifying their engine configuration [7]. This update also improved direct engine support for cluster policies and removed fields, while adding automatic SSH keepalives [7]. Later in the month, CLI v1.14.1 corrected app updates for existing deployments and fixed reference resolution for resource keys with leading underscores [4]. Earlier, CLI v1.13.0 added detailed action summaries and file sync counts for deploy and destroy runs [12].

As the direct engine rollout progressed, community guides outlined migration steps and trade-offs when leaving the Terraform backend [14]. Practitioners examined design patterns across project repositories, comparing monorepo and multi-repo architectures for bundle maintenance [17], and detailed deployment workflows targeting Azure DevOps pipelines [11].

Resource management details drew significant attention. Developers analyzed approaches to bind existing Lakeflow jobs into bundle definitions [8] and managed resource lifecycles [9]. Others highlighted operational hazards when defining Unity Catalog schemas within bundles [10]. Toward the end of the month, discussions and demos focused on syncing UI workspace changes back into bundle definitions [1, 2], alongside features like file watching [19] and immutable folders [18].