Misogynetes v0.2.1

Patch release: enforce mode no longer lets changes slip through during a webhook rollout, and a crash-looping pod counts as one grievance.

Fixes

• Webhook drains on shutdown (#10): after SIGTERM the webhook keeps serving for --drain (default 5s) before shutting down gracefully. Until now, requests the API server still sent to a terminating pod during a rollout were refused and, with failurePolicy: Ignore, passed unchecked, so in enforce mode a change could slip past the refusal on every webhook rollout.
• A crash loop is one grievance (#11): each restart of a crash-looping pod made a heard problem unheard again, so enforce mode kept refusing fixes the person had already listened to. A grievance now starts at the beginning of the loop, and being heard covers the whole loop.
• Clean krew manifest header (#8): the released misogyn.yaml now names its own version and install URL instead of placeholder text and a v0.1.0 link.

Behaviour change

• A heard problem that keeps going is no longer announced again after 15 minutes. It becomes new only after the pod has been healthy for at least 5 minutes. A restart within 5 minutes of the previous one means the pod is still restarting (was 1 minute), in both the webhook and the CLI.

Docs

• The README installs the latest release instead of v0.1.0 (#9) and describes the crash-loop rule.

Install

• krew: kubectl krew install --manifest-url=https://github.com/tym83/misogynetes/releases/download/v0.2.1/misogyn.yaml
• Helm: helm install misogynetes oci://ghcr.io/tym83/charts/misogynetes --version 0.2.1 (image ghcr.io/tym83/misogynetes-webhook:v0.2.1, linux amd64/arm64)

Full Changelog: v0.2.0...v0.2.1