# Windows consumer leg on 49a08a07db7f5da0ca3eb77129d5504ae107f8a9 — disposition (doyle, 2026-09-14)

Shape: RELEASE-RUNBOOK.md:152-166 consumer receipt; run.ps1 mirrors consumer-windows/run.ps1 (attempt3 on b8482445): HEAVY verbatim from the candidate's golden.yml, NEXTEST_PROFILE=ci-windows, `--workspace --no-fail-fast`, SPT_TEST_EPHEMERAL_ADVISORY_PORTS=1, CARGO_BUILD_JOBS=2, asm/304-v3 worktree's own (warm) target, pool claimed lane `consumer-windows-49a08a07` and released 09:27Z. Env scrub by NAME: attempt1 before=[SPT_ENDPOINT_ID,SPT_INJECT_VERIFY_ECHO,SPT_RELEASE_SEED,SPT_SESSION_NAME] after=[]; attempt2 see attempt2-env.txt. Detached pwsh drivers (pids in *-driver-pid.txt).

## attempt1 (full A+B) — receipt attempt1-receipt.txt
- Phase A exit 0: 3459 run / 3459 passed (4 slow, 8 leaky), 1 skipped, 499.985 s. +2 cells vs attempt3's 3457 on b8482445 = the two new bootstrap_firewall unit cells (`a_rule_heavy_prewrite_query_still_reaches_reconcile_write`, `a_hung_firewall_child_is_killed_at_its_operation_budget`). Skipped cell = the tree's only `#[ignore]` (spt-net mdns_discovery_connects_by_id), identical under golden.
- Phase B exit 100: 239 run / 237 passed (1 slow), 2 failed, 944.741 s.
  - FAIL `spt::rc_attach_truth harness_only_refuses_truthfully_pre_stream` (4.859 s) — rc_attach_truth.rs:582 remote arm: expected "harness-hosted", got `No live session for 'hh2' — not running on this node or any visible subnet.` (log lines 68-89)
  - FAIL `spt-daemon::restart_replay_lifetime a_raw_dead_connection_bound_opener_is_not_re_served_by_the_next_generation` (0.901 s — fast red; passes take 10-13 s) — restart_replay_lifetime.rs:594 RESURRECTION assertion, controlled=true driven_by=None; sibling probe passed; full LEG-A transcript in log lines 216-280.
- Post-run: 4 orphaned test daemon/brain pairs from asm-304-v3\target\debug (created 08:29Z during phase A, parents dead) reaped 09:00:37Z, survivors 0 (attempt1-orphans-census.txt).

## Rate measurements (NOT acceptance evidence — intermittent green carries no weight on its own)
- Isolation repeat (repeat-redcells.ps1, same target, scrubbed): both cells 5/5 PASS each, 10/10 (attempt1-redcells-repeat.log).
- attempt2 (phase B only, run-phase-b.ps1) — receipt attempt2-receipt.txt: exit 0, 239 run / 239 passed (1 slow), 1110.164 s, 09:05:07-09:23:46Z. Both attempt1 red cells PASS at the same positions (64: 3.359 s; 191: 13.479 s). Post-run census found 0, survivors 0.
- In-suite rate to date on this box: each red cell 1 red / 2 full phase-B runs on 49a08a07; 0 / 1 on parent b8482445 (consumer-windows/attempt3, 2026-09-13).

## Delta facts
The 49a08a07 diff (one commit over b8482445) touches crates/spt-daemon/src/bootstrap_firewall{.rs,/windows.rs,/linux.rs}, crates/spt/src/serveverb.rs (elevated-retry branch), CHANGELOG.md, docs-site changelog, one ADR. Neither red cell's seam is in it; both binaries run inside the serialized `heavy-broker-pty` test-group. Linux consumer on 49a08a07 (todlando, kitsubito): GREEN A 3417 / B 224.

## Disposition: PENDING hertz classification (dispatched 09:04Z, lane B2FGTAFA, evidence under hertz-phase-b-B2FGTAFA/)
Outcomes: (a) test/rig defect -> repin/harden in hertz's lane, receipt stands as GREEN-with-classified-flake and the classification rides the hand-off; (b) product race independent of 49a08a07 (reproduced on b8482445) -> #307-class filing, receipt stands with the filing linked; (c) caused by 49a08a07 -> BLOCKER, back to todlando. Until one of these lands, the Windows consumer receipt for 49a08a07 is NOT granted.