Skip to content

v0.3.0

Choose a tag to compare

View all tags
@pallaprolus pallaprolus released this 14 Jun 02:09
· 2 commits to main since this release
v0.3.0
9fb82c8

Alpha release. Analysis is read-only — the CLI never changes your cluster.

Recommendation quality & safety

  • Per-resource sizing — CPU and memory are sized independently, so a CPU-wasteful workload pinned at its memory limit still gets its CPU cut.
  • Default strategy is now p99 — a safer savings/violation trade-off.
  • Sizes on raw usage — removed pre-percentile outlier filtering that could under-provision against real demand spikes.

Validation

  • New backtest harness (benchmarks/) that validates recommendations against a public production trace (Alibaba 2018) with a held-out train/test split, plus a one-command fetch script.

Distribution

  • Docker images are now published to GitHub Container Registry: ghcr.io/pallaprolus/kube-foresight.

Docs

  • Operator-first README rewrite, accurate cost framing (reclaimable capacity, not a billing forecast), and durable positioning.

Full changelog: v0.2.0...v0.3.0

Assets 2
Loading