Breadcrumbs

5.30 Release Notes — Databricks

Release notes for Harbr version 5.30, covering features and fixes for platforms deployed using a Databricks workspace.

New Features

At-Source Tabular Asset Cataloguing for Object Stores

General Availability

At-source cataloguing for tabular assets on object stores is now supported on Databricks data planes. Operators can register and catalog tabular assets directly from cloud storage without copying data into the platform.

  • At-source assets are catalogued directly from cloud storage, such as via the S3 and Azure Blob Storages, with no data movement required.

  • Once catalogued, assets can be used in Spaces, Query, Export, and Data Sharing in the same way as on-platform assets.

Note: Currently, these assets can only be consumed via Delta Sharing if the data is in Delta format and accessed via an AWS connector. Assets loaded via the Azure Blob Storage connector will be supported this way in a future release.

For more information, see Using S3 and Azure Blob Storage.

Custom Plan Visibility

General Availability

Custom Plan Visibility gives platform operators control over which organizations and users can see a given plan. Plans can be restricted so they are visible only to intended audiences, rather than all users on the platform.

  • A new plan visibility editor is available in Manage Products / Plans.

  • Visibility can be scoped to specific organizations, specific users, or left open to all.

  • Visibility settings can be applied at plan creation or updated on existing plans.

For more information, see Subscription Plans.

Custom Plan Names

General Availability

Plans can now be given a display name, set individually for each plan in Manage Products / Plans, rather than inheriting the name from the plan template. The plan name is used as the main identifier wherever a plan is shown, across both the Harbr UI and the APIs.

Previously, every plan created from the same template shared that template's name. When a producer used one template to create several plans — for example, with different row and column filtering on each — the plans were hard to tell apart. Giving each plan its own name makes it clearly identifiable to both providers and consumers.

  • A Plan name field is available in the plan editor under Manage Products / Plans.

  • The plan name is shown wherever a plan is referenced — Exchange plan cards, subscription details, and the subscriptions tables — in place of the template name.

  • A plan name can be set when a plan is first created, or added later and released as part of a subsequent product release.

  • The plan name is subject to release control in the same way as other plan settings in Manage Products / Plans.

  • A plan name change on an already-released plan applies to new subscribers only. Existing subscribers see no change — a backfill migration is handled automatically as part of this release.

Restrict Display of Asset Usage based on Capabilities

General Availability

Asset sharing and My Collection now display only the usage types the platform can fulfill for the requesting user or organization. Usage options that cannot be fulfilled are no longer surfaced in sharing flows, collection views, or available actions.

  • Asset sharing for usage now occurs after the asset has been released, and is restricted to the usage types the asset type supports.

  • A Usage column is now shown in My Collection:

    • For products, the usage shown is what is defined on the plan. Usage types that cannot be achieved by any single asset in the product are shown greyed out — for example, if the plan includes exports but no assets in the product are of an exportable type.

    • For assets, the usage shown is the intersection of what the asset type supports and what the user has permission for.

  • Actions available on a product or asset are now aware of the usage the asset supports. For example, the Create export option is not shown for an asset if the asset type does not support export.

Note: This release updates what is shown and available in the UI — it is surface-level filtering, not backend enforcement. Enforcement of fulfillment restrictions at the capability level will be introduced in a future release.

Federation Proxy

General Availability

Federation Proxy enables Federated Connectors to securely access source systems that are not directly reachable from Databricks Serverless compute, including on-premises databases, private corporate networks, and sources protected by IP allowlisting (e.g. MySQL).

How it works

  • Routes connector traffic through a Harbr-managed proxy with fixed IP addresses.

  • Source system owners allowlist the proxy IPs instead of Databricks Serverless IP ranges.

  • Automatically applied to new federated connectors when enabled.

Benefits

  • Enables Delta Share and other Serverless-powered consumption journeys for restricted data sources.

  • Simplifies network access management with a small, stable IP allowlist.

  • No changes to producer or consumer workflows.

  • Data remains in the source system and is queried in real time.

Note: This feature requires platform-level configuration by a Harbr operator. Please contact your Harbr Account Manager to discuss.

For more information, see Configuring Federation Proxy.

‌

Fixes

  • Fixed an issue preventing collaborators from activating Spaces when joining from a different organization.

  • Extended file upload validation to cover additional file types, further protecting the platform against potentially harmful uploads.

  • Fixed logo picker replacing the original image with a degraded crop when re-opened after rotating.

  • Fixed logo picker preview going blank when rotation moved the image outside the crop boundary.

  • Improved logo upload handling when wide logos are enabled for an organization.

‌

Improvements

  • A Subset chip now appears next to product and plan names wherever row or column filtering is active on the plan. Clicking the chip opens the filtering details without leaving the current view. The chip appears in product views, subscription views, and selector screens.

  • Extended flexibility to allow users with non-eco-admin roles to create Subscription Plans with approval-based configurations.

  • Export events now identify the triggering organization in the Event stream and Activity Log. Previously, these events were logged anonymously.

  • Updated AI-powered Discovery character handling and prompt configuration.

  • Improved resiliency by fixing a queue behaviour that could cause exports to time out or not complete.

  • Improved resiliency by hardening approval request processing so persistent failures are dead-lettered rather than retried indefinitely.

  • Improved resiliency by tuning platform auto-scaling and database connection handling to better manage high-volume workloads.

  • Improved error handling to ensure clearer and more consistent responses when platform operations encounter failures.

  • Gobal search now handles UK/US spelling variants. Searching for a term such as "colour" will also return results containing "color", and vice versa. This behaviour is enabled via the enableSpellingVariants flag in the globalSearchConfig ecosystem metadata setting.