On 2026-10-19 09:24 the public path to Gitea failed once and the sidecar exited (max-failures defaults to 0).
Anonymous in-cluster fetch proven first (main and installer-v1.34.0 resolve to the public hashes).
The first draft (5121536f, unpushed) re-issued the claim code and the off-site login on a
reuse whose old box was stale or switched off. A security review flagged it: the caller
holds only the household's retrieval passphrase, and both re-issues are operator
session-only actions (05 section 17.5, 09 decision 252, R-1055/R-1059). Now the reuse
path classifies the old box and re-issues nothing: gone -> host_reinstall_over_old_host
(warning, mailed to the operator: delete the old host and re-run, or re-issue by hand);
live -> host_reenroll_while_live (info). The mint path after the operator's delete is
unchanged and still re-issues. The once-per-silence record and the claim engine comment
edit are reverted. 05 section 8, README and CHANGELOG updated; tests red-proved.
R-1150: the agent's localapi:backup-cadence-cap is a separate per-host state in
HostCapabilityChecker: alone it raises whole_guest_backup_refused (warning, plain text,
agent reason in details) and whole_guest_backup_refusal_cleared (info); other critical
caps keep agent_capability_degraded/_recovered unchanged. New types operator-only and
hub-origin only (not in allowedEventTypes: the suspected controller must not post them).
agent_capability_degraded/_recovered join operatorOnlyEvents (the checker's "never the
customer" claim was configuration only).
R-1154: handleHostEnroll's reuse path re-issues (claim code + off-site login) when no
controller wait is open and either deliberate_off_at is held or both the host report and
the controller report are older than alerting.stale_threshold (wired from main); once per
silence (host_reenroll_reissued record, also written at mint); a live box gets nothing
re-issued and the skip is recorded (host_reenroll_not_reissued). 05 section 8 [DESIGN].
Tests red-proved; go build/vet/test green.
Measured 2026-10-18 on controller jobs 1944 and 1959: same job id,
run_attempt 2, conclusion and log are the rerun's; actions/tasks keeps one
task per attempt (1962 failure + 1964 success; 1978 failure + 1980 success).
The alarm step built its link from GITHUB_SERVER_URL, which on this runner
is the in-cluster Gitea address, and from the run NUMBER, which Gitea 1.26
reads as a run id first (felhom.eu /actions/runs/1116 opened run id 1116 =
run number 613). Now https://gitea.dooplex.hu + GITHUB_RUN_ID, and the link
is printed into the job log as ALARM-LINK. Same change in all four repos.
Rebuilt sizing-data.json; data cache-bust v5; SIZING.md table and the pinned test cases moved
with it. Before/after: audits/day-2026-10-18/P6-before-after.md.
Live on demo-hp (--rotate-recovery-only): vault time moved, new value 200, old 401. Harness 52/0;
the R1145 cases fail against HEAD. Publishes only with the next installer tag.
Golden 0.329.0 baked at the floor's version, NOT vouched. Hub 0.158.0 built, NOT deployed.
Sunday rotation: no duplicate bake of an unvouched golden at the floor's version.
Hub: the R-120 golden guard judges only a CHANGED golden selection; offsite_snapshots_vanished /
_unexplained (operator-only, hub-origin) on the 07:10 listing; internal/accesslog writes path-only
lines to /data/access-log, 90 days. Installer: --rotate-recovery-only runs step 4b alone (rides
the next installer tag). 03: the orphan check runs after a failed backup (agent 6545ccc). 02: the
allow-list is empty, appdata_folders (controller v0.329.0). All red-proofed.
Closed R-971, R-954, R-1124, R-1012 (174 -> 175 rows); opened R-1139..R-1143; R-895 READY with the design
and hub slice 1; R-972 waits on the vouch. [DESIGN] "The catalog allow-list" in 02; the catalog trust line in
01; FELHOM_THINCHECK and unit-forget in 03. Evidence audits/agent-allowlist-rotation-2026-10-17/.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
- hub: `app_definition_refused` in allowedEventTypes + operatorOnlyEvents (controller v0.328.0's catalog
allow-list refusal); SaveHostRecoveryCredential moves set_at on a re-vault (found live: both rotated hosts
kept their first vault's time); the daily key audit also lists the box's restic snapshots/ read-only over the
login the hub already holds and logs count/newest/oldest (no alarm yet).
- installer (main only, rides the next tag): --uninstall removes /usr/local/sbin/felhom-thin-check (agent
v0.159.0's bundle file); test_hostinstall.py knows it.
- runbook whole-guest-backup-orphan-thin.md: the human repair the agent's new orphan finding points to.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
admin