Skip to main content

Changelog

Lakemeter follows Semantic Versioning:

  • Major (X.0.0) — Schema migrations and breaking database changes
  • Minor (0.X.0) — Data-only database updates
  • Patch (0.0.X) — Application-only fixes with no database changes

v0.2.0​

2026-08-31

Data and application release expanding workload coverage, DSU accounting, estimate-level add-ons, and cross-surface calculation parity. The release includes reference-data updates but no Lakebase schema migration.

New workload coverage​

  • Added AI Extract and AI Classify quantity-based estimates (PR #23)
  • Added Unity AI Gateway inference-table and usage-tracking estimates (PR #28)
  • Added Agent Evaluation labels, token, and synthetic-question estimates (PR #29)
  • Added AI Runtime serverless GPU training for AWS and Azure (PR #32)
  • Added General Storage (Databricks Default Storage) with stored-data and Tier 1 and Tier 2 operation DSUs (PR #45)
  • Added standard and OpenTelemetry Zerobus ingestion using Jobs Serverless DBUs (PR #57)
  • Refreshed Databricks and proprietary Foundation Model API catalogs (PR #43)

Platform and export enhancements​

  • Added Enhanced Security and Compliance and Mission Critical Platform Add-ons, calculated from DBU and DSU Product Spend at List (PR #57)
  • Added first-class DSU totals and regional DATABRICKS_STORAGE pricing (PR #45)
  • Expanded Excel exports to 34 columns with DSU costs, a pre-add-on workload summary, a Platform Add-on section, and a final estimate summary (PR #45, PR #57)
  • Added AI Search reranker pricing and aligned AI Search and Lakebase storage with DSU-based calculations (PR #31)
  • Grouped region selectors into geographic areas while preserving each cloud provider's region codes (PR #61)
  • Fixed quantity-based Excel formulas so recalculated workbooks retain the correct totals (PR #25)

Calculation fixes​

  • Fixed single-node cluster pricing so only the driver VM is charged (PR #38)
  • Fixed Model Serving GPU estimates so replica counts multiply both DBUs and VM costs (PR #37)
  • Fixed All-Purpose Serverless rate parity (PR #42)
  • Fixed saved always-on workloads so the UI, calculation APIs, and Excel consistently resolve missing usage to 730 hours (PR #58)
  • Fixed Databricks Apps counts so multiple apps multiply DBUs and cost across the API, UI, and Excel (PR #59)
  • Preserved explicit zero hours and run-based usage precedence (PR #58)

Security and release reliability​

  • Updated frontend and documentation dependencies to patched versions where compatible fixes are available (PR #62)
  • Hardened data upgrades for open PostgreSQL transactions, legacy Lakebase installations, stopped apps, and overlapping Databricks Apps deployments (PR #62)
  • Extended the release gate to verify a fresh installation, upgrade, manual rollback, and a second upgrade of the rolled-back installation (PR #62)

Upgrade notes​

  • Supports direct upgrades from v0.1.0, v0.1.1, and v0.1.2 (PR #62)
  • Applies data updates 020 through 027 for the new workload reference entries (PR #62)
  • Does not add or alter Lakebase columns (PR #62)
  • Use the Upgrade Guide to validate and apply the release from a clean checkout (PR #63)

v0.1.2​

2026-08-08

Code-only patch release focused on calculation accuracy, estimate persistence, and safer release validation. No Lakebase schema change, data migration, or pricing refresh is required.

Bug fixes​

  • Issue #10: Fixed AI-generated estimate application so authenticated requests persist supported workload configuration and roll back cleanly on failure
  • Issue #11: Fixed regional VM pricing parity across the UI and Excel exports for Jobs, All-Purpose, DLT/Lakeflow, and DBSQL workloads
  • Issue #9: Fixed daily, monthly, and run-based usage normalization across calculation APIs
  • Issue #7: Fixed Databricks Apps estimates to use the correct serverless compute SKU and reject unsupported workload types instead of silently falling back
  • Fixed AI Parse and Shutterstock ImageAI quantity fields so saved estimates, cloned estimates, and Excel exports retain the entered values
  • Fixed Lakebase responses to report the effective billable hours used by the calculation

Security and release reliability​

  • Updated frontend and documentation dependencies to address known security alerts
  • Added a tested release-candidate gate that validates installation, upgrade, end-to-end behavior, and rollback before release assets can be published

Upgrade notes​

  • Supports code-only upgrades from v0.1.0 and v0.1.1
  • Leaves Lakebase online and does not modify existing estimates, database schema, or pricing data
  • Use the Upgrade Guide to run plan, doctor, apply, and rollback validation from a clean v0.1.2 checkout

v0.1.1​

2026-08-01

Patch release introducing safer upgrades and correcting AI Parse estimate persistence. No database schema or data migration is required.

New capabilities​

  • Added a version-aware upgrade utility with status, plan, doctor, apply, and rollback commands
  • Added immutable runtime staging, authenticated health checks, concurrency locks, resumable execution, and automatic recovery
  • Added Lakebase backup branches for future minor data updates and major schema migrations; patch upgrades never modify Lakebase
  • Updated new installations to provision Lakebase Autoscaling projects, branches, and endpoints directly

Bug fixes​

  • Fixed AI Parse fields so calculation method, complexity, DBU quantity, page count, mode, and page volume persist and clone correctly

Documentation updates​

  • Added the Upgrade Guide and updated installer and deployment documentation

v0.1.0​

2026-07-24

Initial public open-source release.

  • Workload coverage for Jobs, All-Purpose, DBSQL, DLT/Lakeflow, Model Serving, FMAPI (Databricks + Proprietary), AI Search, Lakebase, Databricks Apps, AI Parse, and Shutterstock ImageAI
  • AI assistant with streaming chat, workload suggestions, and one-click accept
  • Excel export with full cost breakdowns, SKU details, and discount calculations
  • One-command installer (scripts/install.sh) using Databricks Asset Bundles
  • Lakebase-backed estimate storage
  • Multi-cloud support: AWS, Azure, GCP
  • SSO authentication via Databricks Apps
  • Interactive API docs at /api/docs and /api/redoc