relay-shutdown-ci.json

1{
2  "format": 1,
3  "recordedAt": "2026-10-25T00:18:28.327440+00:00",
4  "source": "3d87e07737f722a4cbeb3095f9f22ff38a050a07",
5  "url": "https://github.com/jasona/darkwind-web/actions/runs/40145501400",
6  "status": "completed",
7  "conclusion": "failure",
8  "jobs": {
9    "core": "success",
10    "container": "success",
11    "browserPlayer": "success",
12    "browserMultiplayer": "success",
13    "browserProbe": "failure"
14  },
15  "container": "Production Docker build, model imports, persisted shared room and clean shutdown/restart acceptance passed.",
16  "probe": {
17    "passed": 8,
18    "failed": 2,
19    "tests": [
20      "worker exceptions produce a bounded failure and the next run can recover",
21      "replay worker failure and cancellation release ownership and allow recovery"
22    ],
23    "finding": "Both recovery runs report Render worker timed out during model-world. The worker request deadline is 20 seconds. The inner operation at the timeout is not recorded, so the root cause remains unproven. The main Observatory run and six other probe scenarios passed.",
24    "artifact": {
25      "name": "playtest-evidence-probe",
26      "id": 12581185965
27    },
28    "contexts": [
29      "relay-shutdown-ci-worker-error.md",
30      "relay-shutdown-ci-replay-error.md"
31    ],
32    "action": "Keep the failure visible; instrument authored-model probe stages before attributing or changing its deadline."
33  },
34  "previousAlphaCi": {
35    "url": "https://github.com/jasona/darkwind-web/actions/runs/40145062459",
36    "source": "107e9339dc36921a8a0fdbc6bddf0bd63cd8e254",
37    "jobs": {
38      "core": "success",
39      "container": "failure",
40      "browserPlayer": "success",
41      "browserMultiplayer": "failure",
42      "browserProbe": "success"
43    },
44    "multiplayer": {
45      "passed": 3,
46      "failed": 1,
47      "test": "graphics recovery keeps the shared room and receives door changes during restoration",
48      "finding": "Shared-door/recovery assertions completed, but the final screenshot took 69292.045ms and the test exceeded its 120000ms deadline. The capture stall cause remains unproven.",
49      "traceCall": "call@328",
50      "screenshotStartMs": 237432.322,
51      "screenshotEndMs": 306724.367
52    },
53    "container": "SIGTERM exit 1; the subsequent shutdown fix passes container CI above."
54  }
55}
56