ContextQMD
Back to Datahub

v0.3.9

docs/managed-datahub/release-notes/v_0_3_9.md

1.5.0.34.8 KB
Original Source

v0.3.9

Release Availability Date

17-Mar-2025

1.0.0

Helm Chart

1.5.5 or greater

Known Issues

  • OpenAPI Async Timeseries Aspects will not automatically create the entity's key aspect.
  • Writes for hard deleted entities will be dropped for 60 seconds after the hard delete is processed.
  • API Tracing is considered as a beta release

Release Changelog

v0.3.9.2

  • [Lineage Visualization] Fix bug where assets would not appear when hiding transformations.
  • [Remote Executor] Create upgrade job to remove legacy remote executor status
  • [Usage Ranking] Make usage feature reporting source more robust to non-standard data.
  • [Proposals] Make action request owner source more robust to non-standard data.
  • [Platform] Add error handling when deleting an aspect that does not exist.

v0.3.9

  • All changes in https://github.com/datahub-project/datahub/releases/tag/v0.15.0.1

  • Breaking Changes

  • Bug Fixes

    • [Compliance Forms] Fix a few inaccuracies in the Compliance Forms analytics tab - handle multiple forms per asset properly and fix a few bugs in the forms analytics ingestion source when generating this analytics data.
    • [Compliance Forms] Fix bug in nested select component when adding a Domain question to your Compliance Form when selecting child domains. Now you can properly select a parent and their children.
    • [Queries Tab] Fix column and user filters in the Query tab by showing the proper display name instead of urns.
    • [Search Results] Fix order of parent entity paths in the autocomplete dropdown for a given result.
    • [CSV Export] Fix missing display name for Groups when downloading search results.
    • [Users & Groups] Fix group membership inconsistencies on groups page. Correctly show the count of members and improve the experience to make it clearer who is a member of what group.
    • [Columns Tab] Fix bug in schema table where we were not showing metadata if a field changed in ingestion from V1 to V2 field path or vice versa.
    • [Search Filtering] Fix dropdowns for selecting platforms and other filters when creating Views that were not showing any results.
    • [Search Results] Fix lineage counts on search result cards that were incorrectly counting "ghost" (non-existent) entities.
    • [Compliance Forms] Fix bug when filling out documentation prompts on Datasets to update the correct aspect.
    • [Log Out] Fix logout bug in new navigation bar redesign when a user is logged in with SSO.
    • [Search Ranking] Improved memory usage of usage ranking job; fix bugs when usage was missing.

  • Product

    • Introducing a new Stats Tab redesign for Datasets and Columns! This release gives you all the functionality you had with Dataset and Columns stats previously, but now in a much cleaner, more consumable way. We also add a few new stats and charts to render additional helpful information about your assets. Note that statistics must be enabled in ingestion to view them on the stats tab.
    • Improved support for ML Models and ML Groups as part of the expanded MLflow ingestion source. Data process instances can now appear on container pages and lineage.
    • Improve the Compliance Forms experience while filling out questions by pre-filling responses for the user with existing metadata on the asset. Now, a user will have an even easier time filling out Compliance Forms for their assets.
    • The new navigation bar experience is now the default! Quickly understand & navigate to every DataHub feature.
    • Improve the navigation bar experience by centering and expanding the search bar at the top of the page.
    • Add Compliance Forms validation to prevent duplicate question IDs across and within Compliance Forms.
    • Elevate structured properties displayed on Asset Sidebar about Status section
    • Entity versioning is now supported via ingestion, for version-native sources like MLflow.

  • Platform

    • Remote Executor: Introduced new Prometheus Metrics for remote executor monitoring:
      • Ingestion: datahub_executor_worker_ingestion_requests and datahub_executor_worker_ingestion_errors to monitor ingestion jobs received from queue
      • Memory: datahub_executor_memory_* to monitor usage and limits by process group/host
      • CPU: datahub_executor_cpu_* to monitor usage and limits by process group/host
      • Disk: datahub_executor_disk_* to monitor total bytes and bytes used
      • Network: datahub_executor_net_* to monitor bytes and packets sent/received, etc.

  • Ingestion

    • MLflow: Support for experiments & runs ingestion, enhanced property ingestion for models/model versions with native versioning support
      • Dataset ↔ Run lineage is coming soon.
    • Spark: Column-level lineage now runs through a Spark DataJob rather than directly between datasets.