Skip to content

Releases: rh-aiservices-bu/litemaas

0.1.2

17 Oct 18:39
@guimou guimou
0.1.2
c91c814
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
0.1.2 Latest
Latest

What's Changed

  • fix(migrations): skip inactive keys in litellm_key_alias backfill by @guimou in #66
  • chore: release v0.1.2 - fix critical migration issue by @guimou in #67

Full Changelog: 0.1.1...0.1.2

Contributors

guimou
Assets 2
Loading

0.1.1

17 Oct 16:30
@guimou guimou
0.1.1
696e6ad
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
0.1.1

Release Summary

This PR merges two releases from dev to main:

  • v0.1.0 - Enterprise features milestone
  • v0.1.1 - Patch release with reliability improvements

Brings main from v0.0.19 → v0.1.1

🎯 v0.1.1 (Latest) - Patch Release

Fixed

  • Orphaned API Key Handling: Database migration now gracefully handles API keys deleted in LiteLLM

    • Marks orphaned keys with orphaned_ prefix instead of failing
    • Eliminates verbose error logging during backfill operations
    • Improved 404 error handling during key synchronization

  • Server Startup Performance: Backend initialization is faster and more resilient

    • LiteLLM-dependent operations moved to background tasks using setImmediate
    • Server starts immediately without waiting for external services
    • Non-blocking initialization prevents startup delays

  • Frontend Test Reliability: Fixed missing /api/v1/banners mock endpoint

    • BannerProvider now works correctly in Layout component tests
    • Eliminated test failures related to banner initialization

🚀 v0.1.0 - Enterprise Features Milestone

This major release introduces 3 enterprise-grade features for production-ready governance and observability.

New Features

1️⃣ Restricted Model Subscription Approval Workflow

Enterprise governance for sensitive/costly AI models with comprehensive approval system.

Key Capabilities:

  • Three-state workflow: pending → active / denied
  • Bulk approval/denial operations with detailed result tracking
  • Full audit trail in subscription_status_history table
  • Automatic access revocation when models become restricted
  • LiteLLM API key synchronization for immediate enforcement
  • Granular RBAC (admin vs adminReadOnly permissions)
  • Complete i18n support (9 languages)

Technical Details:

  • New database table: subscription_status_history
  • New API endpoints: /api/v1/admin/subscriptions
  • New frontend page: AdminSubscriptionsPage
  • 500+ lines of new tests

2️⃣ Comprehensive Admin Usage Analytics

System-wide visibility with intelligent caching and multi-dimensional filtering.

Key Capabilities:

  • Day-by-day incremental caching (permanent historical + 5-min current day TTL)
  • Multi-dimensional filtering: users, models, providers, API keys
  • Trend analysis with automatic comparison period calculations
  • Rich visualizations: usage trends, model distribution, weekly heatmap
  • CSV/JSON data export with filter preservation
  • Configurable cache TTL via backend configuration
  • 70%+ backend test coverage

Technical Details:

  • New services: AdminUsageStatsService, DailyUsageCacheManager
  • New API endpoints: /api/v1/admin/usage
  • New frontend page: AdminUsagePage with 4 tabs
  • Refactored into specialized modules (~5000+ lines total)

3️⃣ Model Configuration Testing

Real-time validation for custom model creation with user-friendly feedback.

Key Capabilities:

  • Test API endpoint and credentials before model creation
  • Check model availability at specified endpoint
  • Display available models when requested model not found
  • Clear error messages for auth failures and network issues
  • i18n support for all test states

Code Quality & Refactoring

  • Major Refactoring: Decomposed monolithic admin analytics service

    • Split 2900-line service into 5 specialized modules
    • Improved separation of concerns and testability
    • New infrastructure utilities (advisory locks, pagination, date validation)

  • Frontend Enhancements:

    • Reusable usePagination hook
    • Shared chart utilities and accessibility improvements
    • New filter components with cascading dependencies

Documentation

72+ new documentation files including:

  • Feature guides: subscription approval, admin analytics, model testing
  • Development guides: pattern reference, chart components, code review checklist
  • Enhanced architecture docs with new tables and services
  • Complete refactoring documentation (6 phases documented)

Bug Fixes

  • OAuth TLS verification bypass option for k8s self-signed certs (#58)
  • Daily usage cache token breakdown calculation accuracy

📊 Release Statistics

Combined Changes (v0.1.0 + v0.1.1):

  • 362 files changed
  • ~127,000+ lines added
  • 16 commits merged to dev
  • New database table: subscription_status_history
  • New backend services: 7 major new services
  • New API endpoints: 12+ new admin endpoints
  • Enhanced test coverage: 2,000+ lines of new tests

🔧 Technical Details

Database Changes

  • New table: subscription_status_history for audit trail
  • Migration: Fix daily usage cache token breakdowns
  • Schema updates for subscription status and restricted models

Backend Changes

  • 7 new major services (admin usage, subscription approval, notifications)
  • New infrastructure utilities (advisory locks, pagination, rate limiting)
  • Enhanced RBAC with granular permission checks
  • Improved error handling and validation

Frontend Changes

  • 3 new major pages (AdminSubscriptionsPage, AdminUsagePage, enhanced AdminModelsPage)
  • 20+ new reusable components
  • Enhanced i18n with 400+ new translation keys across 9 languages
  • Improved chart accessibility and shared utilities

⚙️ Upgrade Notes

Required Actions

✅ Run database migrations (creates subscription_status_history table)

Optional Configuration

  • Set K8S_SKIP_TLS_VERIFY=true for self-signed k8s certs (dev only)
  • Configure admin analytics cache TTL via backend config
  • Review new rate limiting configuration options

Compatibility

✅ No breaking changes - Fully backward compatible
✅ All existing features continue to work as before
✅ New features are additive only

🧪 Testing

Coverage

  • Backend: 70%+ coverage for new admin features
  • Integration tests: Complete endpoint coverage
  • Frontend: Comprehensive component and accessibility tests
  • 2,000+ lines of new test code

Validated Scenarios

  • Orphaned key handling in various states
  • Subscription approval workflow (bulk and individual)
  • Admin usage analytics filtering and caching
  • Model configuration testing with various providers
  • Server startup resilience

👥 Contributors

  • Guillaume Moutier (@guimou)
  • Co-authored-by: Claude (AI pair programming)

📚 Documentation

See CHANGELOG.md for complete details on both releases.

Key Documentation:

What's Changed

  • Update openshift-deployment.md by @guimou in #56
  • Very minor openshift deployment enhancements and correction by @tonykay in #57
  • feat(admin/analytics): implement comprehensive admin usage analytics system by @guimou in #59
  • refactor(admin/analytics): decompose monolithic service into specialized modules + other refactoring by @guimou in #60
  • feat(auth/oauth): add optional TLS verification bypass for k8s API by @guimou in #61
  • feat(subscriptions): implement restricted model approval workflow by @guimou in #63
  • fix(test): handle orphaned keys and improve server startup by @guimou in #64
  • chore(release): merge v0.1.0 and v0.1.1 to main by @guimou in #65

New Contributors

Full Changelog: 0.0.19...0.1.1

Contributors

tonykay and guimou
Loading