Three single-scene H16 canaries sealed and passed QH readback; one Mac artifact roundtripped through Ubuntu with matching manifest bytes. These bounded operational results do not establish completed fleet generation, scientific pooling or unattended recovery.
The frozen queue contains 77 eligible TRAIN, 20 reserved exposed-validation and three excluded assignments. At the later 12 October capture it showed three successes, one active assignment and 74 remaining. That is a timestamped historical observation, not current service status. The canaries use one scene and source key; they cannot estimate scene-level performance.
The receipt-aligned packed-layout check covered 32 distinct-scene origins in each of two layouts: 64 comparisons and zero differences across all nine native post-hoc tables. This is conservation evidence for those origins and producer hashes, not a global layout optimum. Warm-cache timings use small purposive samples and uncontrolled OS caches.
A disposable packed transfer rejected an interrupted payload, resumed it and verified it before publication. Source hashes stayed unchanged. The Streamlit follow-up exercised local visual refresh and recorded repeated-call cache timings. Fixture reconnect tests do not simulate a physical two-host outage.
Remaining gates: scientific Mac/CUDA compatibility, representative cold-cache/quota evidence, reboot and sudden-death recovery, and prospective cluster access. The provider-level child-environment defect remained unresolved in these dated receipts even though a campaign workaround allowed generation to seal.
The complete commissioning record retains operational chronology and full tables. Its source README and dated supplement link compact transport, cache and post-hoc comparison receipts. This support report adds no scientific figure: these heterogeneous checks cannot support a comparative scientific performance plot. The cached-workload benchmark is a separate primary measurement.
| Finding / evidence | RQ | Thesis section | Coverage / gap |
|---|---|---|---|
| Canary, transport and conservation checks | None | sec:thesis-experimental-design | Outside scope: operational qualification; no scientific pooling |
Previous primary report retains its complete narrative, exact tables and original figures. This 15 October revision is an authored synthesis, not a new experiment or a change to the numerical snapshot.
Experiment index and evidence routes. Missing/private parents remain explicit in the native campaign owner; historical versions retain their original dates and scientific identities.
Scientific snapshot and attempt identities remain those of the cited experiment. Report synthesis: 15 October 2026; restored to the domain/date owner: 16 October 2026.