FAILURE MAP
← Case archive

FA-19597 / Protocols / Member archive

Peer restart-drain coordination controller: draining / work finished · case 02

While draining, event work_finished produces stopped/terminate instead of ready/send_ready.

Member previewVariant 2 · 3 implementations · 14 checks per implementation

Case contract

For the stipulated Peer restart-drain coordination controller, initialize the explicitly shown resource registers from seed (1..5), then consume classified events sequentially. After each event return a deep snapshot of phase, resource registers, and cumulative emitted control frames. The explicit ten phase/event rules and their resource effects define admission and response semantics. Unlisted events preserve phase and emit rejection. Payload is the consecutive integers seed through 2*seed+1. Empty event input returns no observations.

Why this case matters

Incorrect event admission now changes concrete queued data, resource ownership, or emitted protocol frames and subsequent trace behavior.

One recorded failure

Sample boundary fixture

This sample comes from the broken implementation of a controlled reproducer.

Boundary fixtureActualExpectedOutcome
draining/work_finished[{"frames": [], "phase": "stopped", "registers": {"admission": false, "delivered": [], "generation": 2, "owned_session": null, "ready": false, "work": []}}][{"frames": [["DRAINED", 2]], "phase": "ready", "registers": {"admission": false, "delivered": [], "generation": 2, "owned_session": 2, "ready": true, "work": [2, 3, 4, 5]}}]Failed

MEMBER ARCHIVE

The complete case is available to members.

This record includes three runnable implementations, regression fixtures, execution results, and source hashes.

Member access is invitation-based. Sign in with your invited account to inspect the sources.

Sign in to the archive ↗