Changelog

All notable changes to k8s LogJedi are documented in this file.

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

Unreleased

0.1.0 - Initial release

Added

  • Go operator: controller-runtime; watches Pods, Deployments, Jobs; detects failure signals (CrashLoopBackOff, ImagePullBackOff, ErrImagePull, OOMKilled, failed Jobs, unavailable Deployments).
  • LLM service: FastAPI POST /analyze with Pydantic validation; Strands agent with structured output; multi-provider support (mock, OpenAI, Bedrock, Gemini).
  • Log backend: in-cluster Kubernetes and generic HTTP backend.
  • Notifications: Slack and Microsoft Teams webhooks.
  • Apply logic: APPLY_MODE=auto or manual; strategic-merge patch with dry-run option; per-resource cooldown; namespace and apply-scope filters.
  • Secret redaction in specs sent to the LLM; optional operator–LLM auth header.
  • GET /analyses and GET /reports; POST /report-outcome for operator callback (step-by-step reports).
  • Helm chart under charts/logjedi/ with configurable values.
  • README, RUNBOOK, ROADMAP, ADRs, and deploy manifests.