Skip to content

CHANGELOG.md

Latest commit

 

History

History
181 lines (133 loc) · 7.95 KB

CHANGELOG.md

File metadata and controls

181 lines (133 loc) · 7.95 KB
Raw

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

[0.6.0] - 2026-03-27

Added

  • 15 new detectors across 3 families (total: 45 detectors)
    • mongopulse (MongoDB): ConnectionExhaustion (CRITICAL), OplogWindow (CRITICAL), ReplicationLag, LockPercentage, CursorTimeout (WARNING)
    • mysqlpulse (MySQL/MariaDB): ConnectionExhaustion (CRITICAL), ReplicationLag, Deadlocks, SlowQueries, InnoDBBufferPoolPressure (WARNING)
    • airflowpulse (Apache Airflow): SchedulerHeartbeat, PoolExhaustion (CRITICAL), DAGFailureRate, TaskQueueBacklog, ZombieTasks (WARNING)
  • Runbook docs for all 15 new detectors

[0.5.0] - 2026-03-25

Added

  • 12 new detectors for pgpulse (PostgreSQL) and clickpulse (ClickHouse) metrics (total: 30 detectors)
    • pgpulse: ConnectionExhaustion (CRITICAL), ReplicationLag, DeadTupleRatio, LockChainDepth, SlowQueries (WARNING)
    • clickpulse: StuckMutations, PartCountExplosion, KeeperOutstandingRequests (CRITICAL), MergePressure, ReplicaLag, DDLQueueStuck, KeeperHighLatency (WARNING)
  • Runbook docs for all 12 new detectors

0.4.1 - 2026-03-18

Fixed

  • Silence klog output in TUI mode so client-go port-forward errors no longer corrupt the display

0.4.0 - 2026-03-11

Added

  • 3 tote image salvage detectors (total: 18 detectors)
    • ToteSalvageFailure: detects failing node-to-node image transfers (CRITICAL)
    • TotePushFailure: detects failing backup registry pushes (WARNING)
    • ToteHighFailureRate: detects when most image pull failures use tags instead of digests (WARNING)
  • Runbook docs for all 3 tote detectors
  • CLAUDE.md removed from git tracking (was committed since v0.1.0)

0.3.1 - 2026-03-11

Fixed

  • Port-forward errors now surface immediately instead of waiting 10s and returning generic "timeout" message

0.3.0 - 2026-03-03

Added

  • SARIF 2.1.0 output for GitHub Code Scanning (--output sarif)
  • Cross-detector correlation with 6 built-in rules (deployment failure, memory pressure, mesh cascades)
  • Runbook links for all 15 detectors with ? key in TUI and HelpURI in SARIF
  • Problem history with local SQLite backend (--history flag, history list/prune subcommands)
  • Recurring problem badges in TUI, plain text, and SARIF output
  • Multi-cluster sweep command (infranow sweep) — scans all kubeconfig contexts via port-forward
  • Sweep parallel mode (--parallel), context glob filtering (--contexts), namespace filtering
  • Sweep output in text, JSON, and SARIF formats with cluster annotations
  • Table-based TUI with selectable rows, detail panel, copy/yank, and digit jump keys

Changed

  • TUI rebuilt with bubbles/table replacing viewport-based renderer
  • Detail panel expanded to 7 lines (added history badge and runbook link)
  • Named constants replace magic numbers throughout codebase

[0.2.0] - 2026-03-02

Added

  • Plain text output mode (--output text) for piped and CI usage
  • --once flag for single detection cycle then exit
  • Auto-detection of piped stdout — falls back to text mode when not a TTY
  • --json flag on version subcommand for machine-readable version output
  • Tiered exit codes: 0=clean, 1=warnings, 2=critical/fatal
  • GoReleaser config for automated builds, checksums, and Homebrew tap updates
  • SKILL.md badge in README linking to docs/SKILL.md

Changed

  • Version output unified to single-line format: infranow 0.2.0 (commit: X, built: Y, go: Z)
  • Release workflow replaced with GoReleaser action (from custom shell script)
  • SKILL.md moved from repo root to docs/SKILL.md per ANCC convention
  • Makefile LDFLAGS now include -s -w for smaller binaries
  • Exit codes renumbered: invalid input=3, runtime error=4

Fixed

  • README detector table missing trustwatch detectors (now shows all 15)
  • README architecture section showed 7 detectors instead of 15
  • ARCHITECTURE.md stale problem pruning time said 2 minutes (actual: 1 minute)

0.1.2 - 2026-02-21

Added

  • Trustwatch certificate and probe failure detectors (2 new detectors, total: 15)
    • TrustwatchCertExpiry: tiered alerts for trustwatch-monitored certificate expiry
    • TrustwatchProbeFailure: detects trustwatch endpoint probe failures
  • Safety Model section in README documenting zero-footprint guarantees
  • golangci-lint config (.golangci.yml) with gocritic, gocyclo, revive enabled
  • Trivy security scanning in CI pipeline
  • GPG signing for release checksums

Security

  • SHA-pinned all GitHub Actions to commit hashes (supply chain hardening)
  • Added go mod verify to release workflow (dependency integrity)
  • Added -trimpath to release builds (path leak prevention)
  • Scoped release workflow permissions to job level
  • Added context timeouts for Kubernetes API calls in port-forward
  • Added HTTP client timeout for SPDY port-forward transport
  • Enhanced Prometheus URL validation with link-local SSRF rejection
  • Fixed export file permissions: 0666 → 0600
  • Fixed TUI signal handling: graceful shutdown via tea.Quit instead of os.Exit
  • Capped problem map at 10,000 entries to prevent unbounded memory growth
  • Sanitized Prometheus URL in TUI header display
  • Fixed LDFLAGS to use VERSION_NUM (no v prefix)

Fixed

  • Octal literal style in baseline file permissions (0600 → 0o600)
  • Import ordering with goimports local-prefix grouping

0.1.1 - 2026-02-11

Added

  • 6 service mesh detectors for linkerd and istio
    • LinkerdControlPlane: detects linkerd deployments with zero replicas (FATAL)
    • LinkerdProxyInjection: detects linkerd pods in CrashLoopBackOff (CRITICAL)
    • IstioControlPlane: detects istiod with zero replicas (FATAL)
    • IstioSidecarInjection: detects istio-system pods in CrashLoopBackOff (CRITICAL)
    • LinkerdCertExpiry: tiered alerts for identity cert expiry (<7d WARNING, <48h CRITICAL, <24h FATAL)
    • IstioCertExpiry: tiered alerts for root cert expiry (<7d WARNING, <48h CRITICAL, <24h FATAL)

Changed

  • Total detector count: 7 → 13
  • CLAUDE.md synced with global project standards
  • CONTRIBUTING.md commit message format aligned with conventional commits
  • ARCHITECTURE.md Go version corrected to 1.25+, stale timing fixed to 1 minute

Fixed

  • Duplicate .PHONY: deps in Makefile
  • Stale problem timing documented as 2 minutes but implemented as 1 minute

0.1.0 - 2026-02-08

Added

  • 7 deterministic detectors with explicit PromQL thresholds
    • Kubernetes: OOMKill, CrashLoopBackOff, ImagePullBackOff, PodPending
    • Generic: HighErrorRate (>5% 5xx), DiskSpace (90%/95%), HighMemoryPressure (>90%)
  • Interactive TUI with real-time problem ranking by severity, recency, or count
  • JSON output mode for CI/CD pipelines (waits for first detection cycle, then exits)
  • Baseline save/compare with --fail-on-drift for regression detection
  • Severity gate via --fail-on for CI/CD exit code control
  • Native Kubernetes port-forwarding via client-go (no kubectl dependency)
  • Namespace include/exclude filtering with glob patterns
  • Configurable polling intervals, concurrency limits, and detector timeouts
  • Problem scoring based on severity weight, blast radius, and persistence
  • Stale problem pruning (removed after 1 minute without re-detection)
  • TUI keyboard navigation: scroll, sort, search/filter, pause/resume
  • Prometheus health monitoring with connection status in TUI header
  • Multi-platform builds via Makefile (Linux, macOS, Windows)