v0.1.0-alpha — First alpha release

First alpha release

ShutdownCheck terminates your service the way your orchestrator will — under real load — and tells you exactly which stage of shutdown you got wrong, and how to fix it.

What's in this release

• 7-stage shutdown model — every check maps to exactly one stage (signal received → readiness flipped → lame-duck window → listener closed → connection close signalled → in-flight drained → clean exit), and every diagnosis names the stage that broke
• Named, explainable findings — e.g. SC003 IN_FLIGHT_DROPPED, SC006 NO_DEREGISTRATION_WINDOW, each with an explanation and a fix (shutdowncheck explain SC006)
• Three targets — processes, managed commands, and Docker containers
• Deterministic 0–100 scoring with multi-trial consistency; INCONCLUSIVE instead of a false PASS when in-flight load can't be proven
• NDJSON record/replay — record a run, re-judge it later under a different profile (--profile kubernetes) without re-running
• GitHub Action — fails the step on defects, writes the report to the job summary, uploads JSON
• shutdowncheck demo — a real check against a real process receiving a real SIGTERM, no setup needed
• Supply-chain hardened — cosign keyless-signed binaries, SBOMs, SLSA build provenance; install script verifies checksums automatically
• Conformance suite proving identical defect detection across Go, Node.js, and Python fixtures

Status

Alpha — the tool runs end to end and produces a verdict. Not yet recommended as a merge gate you depend on. Kubernetes targeting is planned follow-up work.

Binaries: linux, macOS, Windows on amd64 and arm64.