Skip to content

CHANGELOG.md

Latest commit

 

History

History
94 lines (88 loc) · 8.27 KB

CHANGELOG.md

File metadata and controls

94 lines (88 loc) · 8.27 KB

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

[1.0.0-alpha] - 2026-02-12

Added

  • Initialized Valiant project structure.
  • Backend (Go):
    • Implemented Orphan Event Detection: Execution events (e.g., GitOps, manual) without a corresponding intent event (e.g., CI) within a configurable correlation window are now marked as IsOrphaned.
    • Intent-Execution Linking: Implemented backend logic to link CI (Intent) and GitOps/manual (Execution) events using git_sha or image_tag within a configurable intent_execution_correlation_window. This replaces the previous IsOrphaned check based solely on services.
    • Extended PostgresStorage.GetChangeEvents to support filtering by metadata content (git_sha or image_tag).
    • Renamed OrphanCorrelationWindow to IntentExecutionCorrelationWindow in configuration for clarity.
    • Extended PostgresStorage.GetChangeEvents to support filtering by trigger type, time range, and affected services.
    • Added IsOrphaned field to domain.ImpactAnalysis for API response.
    • Added OrphanCorrelationWindow configuration to config.yaml.
    • Project initialization with Go modules.
    • Core domain models for ChangeEvent, MetricValues, and ImpactAnalysis.
    • Defined interfaces for Storage, MetricsProvider, and Collector.
    • Skeleton implementations for PostgreSQL storage, Prometheus metrics, and Kubernetes, Git, and CI/CD collectors.
    • Core impact correlation engine logic with configurable weights and thresholds.
    • Custom weights of metrics (built-in and custom) applicable in config.yaml
    • Timestamp Guardrail for ChangeEvent Ingestion: Implemented a defensive mechanism to prevent ChangeEvents with future timestamps from entering the analysis workflow.
      • Defined a max_future_skew of 2 minutes (hardcoded). Events with Timestamp or EndTime beyond this future limit are marked as invalid.
      • Invalid events are persisted with status='invalid_time', invalid_reason (e.g., timestamp_in_future), and skew_seconds for auditability.
      • Structured warning logs are emitted for invalid events.
      • Analysis workers (GetEventsPendingAnalysis) now explicitly filter and skip events with status != 'ready', preventing them from entering the analysis queue.
    • Configurable Worker Polling Interval: Made the analysis worker's polling frequency configurable via config.yaml.
      • Added a worker.polling_interval field to config.yaml (default "5m"), allowing users to adjust how often the worker checks for new events to process.
      • The backend worker now uses this configurable interval.
    • HTTP API server with health check, event submission, event listing, and impact analysis endpoints.
    • Initial database migration for the change_events table.
    • Environment-based configuration management.
    • Implemented PostgresStorage with actual SQL queries for saving and retrieving change events.
    • Added automatic execution of 001_initial_schema.sql on backend startup.
    • Implemented CORS middleware to allow cross-origin requests from the frontend.
    • Updated main.go to connect to PostgreSQL using lib/pq and configuration.
    • Implemented PrometheusClient with actual PromQL queries for error rate, latency, RPS, and saturation metrics.
    • Basic homepage component.
    • API client layer for interacting with the backend.
  • Deep Linking Utility:
    • Implemented dynamic generation of contextual links within the backend's event retrieval (specifically PostgresStorage.GetEventsPendingAnalysis) based on configurable LinkTemplates.
    • Enhanced PostgresStorage to gracefully handle invalid URL templates and missing metadata keys, skipping link generation when <no value> is produced.
    • Expanded deep linking test suite (tests/deep-linking) with comprehensive edge cases, including empty MetadataHas, static URLs, special characters in metadata, and cases with partial or missing metadata.
    • Updated example seed scripts (example/scripts/seed_data.ps1, example/scripts/seed_data.sh) to generate events with varied metadata for deep linking, including valid, partially valid, and invalid scenarios, enabling thorough testing and demonstration.
  • Documentation:
    • Added HOW_TO_USE.md with "Quick Start", "Core Concepts", and "For Dummies" guide on connecting apps.
    • Fixed missing Tailwind CSS configuration (tailwind.config.ts, postcss.config.js) and dependencies.
    • Moved styling dependencies to dependencies and updated Dockerfile to ensure npm install runs on startup, fixing volume sync issues.
    • Added --legacy-peer-deps to npm install commands to resolve React 19 peer dependency conflicts.
    • Implemented GitCollector to parse git tags and commits from a local repository.
    • Implemented immutable ImpactAnalysis snapshots. Analysis results are now stored in the database and reused to prevent historical drift.
    • Added support for config.yaml to configure analysis windows (baseline/impact durations), and now customizable Prometheus query templates with variable substitution ({{ .Services }}, {{ .Duration }}).
    • Implemented Automatic Background Analysis: A worker now periodically checks for events with expired impact windows and calculates/snapshots their impact without user intervention.
    • Fixed fetchEvents reference error in frontend by renaming handler to fetchData.
    • Improved data models in the frontend API layer.
    • BREAKING: Changed monitoring model to focus on execution boundaries (Trigger Type: CI, GitOps). Manual events are now deprecated.
    • Added trigger_type, execution_id, and end_time to ChangeEvent model and database schema.
    • Polished UI with new icons for triggers: GitBranch for GitOps, Bot for CI.
    • Updated seed scripts to exclude manual actions, reflecting the execution-only model.
    • Disabled GitCollector default behavior to align with the new model.
    • Updated seed scripts (.sh and .ps1) to generate 12 diverse events with varying timestamps and distinct affected services (fixing filtering issues).
    • Fully implemented KubernetesCollector using client-go with strict auditor logic. It now only emits events for completed rollouts (Available=True) that prove intent via valiant.io/source annotation.
    • Added support for allowed_sources configuration to trust only specific deployment systems (ArgoCD, Helm, CI/CD).
    • Refined K8s detection to capture precise rollout_start and rollout_end timestamps for impact anchoring.
    • Refactored Collector interface to be streaming/push-based (Start(ctx, chan)).
  • Frontend (Next.js):
    • Next.js 16 project skeleton with TypeScript and React 19.
    • Root layout with Inter font and global CSS.
    • Implemented Timeline and TimelineEvent components.
    • Added on-demand impact analysis triggering from the UI.
    • Added detailed metric deltas breakdown in the analysis view with improved visual design (cards, colors).
    • Added "Show More" pagination logic to the timeline view.
    • Polished UI with lucide-react icons and improved layout for events and metrics.
    • Implemented dedicated Service Dashboard pages (/services/[name]) for deep-linking and focused analysis.
    • Enhanced Service Filter UI with focus links.
    • Added "Pending Analysis" badge for events where the impact window (35m) has not yet closed.
    • Fixed filtering logic in Service Pages to handle URL encoding correctly.
    • Refined Service Filter pills on homepage to embed the focus link inside the pill.
    • Enforced impact window validation in the backend: AnalyzeImpact now returns a specific error and "PENDING" status if called too early, preventing premature analysis.
    • Added tooltips to metric cards to explain what each metric represents.
    • Added "time ago" badge (e.g., "5m ago") next to event timestamps.
    • Added safe handling of null API responses in fetchChangeEvents.
    • Modal with PromQL queries and their weights for overall score.
  • Infrastructure:
    • Dockerfiles for Backend and Frontend.
    • docker-compose.yml for full stack local development (Postgres, Backend, Frontend).
    • Added bash and PowerShell scripts for seeding mock data.