The key was the display string cut after ten names, so two different app sets
with the same first ten names shared one wait and one operator mail.
largeSetKey is a SHA-256 of the full sorted set; the display string is only
shown. Test seen red; 2 mutations seen red (redproof-review-s2.txt).
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
The wait of a large removed-app request was timed from a search of the
windows' refusal text, which also held names a box had claimed. A box could
plant the marker in a claimed name a week ahead and skip the wait and the mail.
- offsite_large_requests (household, key of the app set, first_seen): written
only by the hub; at most 50 rows per household; it goes with the household.
No text is searched any more.
- A claimed name that has not the shape of an app name is written nowhere: the
ledger, the answer and the operator's message carry a fixed placeholder.
Tests seen red; 6 mutations seen red (redproof-review-s1.txt).
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
offsite_end_scheduled bypasses the household's switches and went through the
ordinary customer leg: subject and body were the box's own sentence, and 50
invented subjects gave 50 mails.
- processReceipt: the details must name a checked app (app-name shape, not
reserved) and two parsed dates; the mail is rendered only from the hub's
bundle with that app and the parsed end date — never the box's message.
- The cooldown key is the checked app + the parsed time of the press.
- At most 5 receipts per household in 24 hours, counted in the hub's sent log;
drops are said once a day; old receipt keys leave the cooldown map.
- Pinned: a blocked household and a household with no address get none.
Tests seen red before the fix; 12 mutations seen red, 3 after a redo
(documentation/audits/build-release-4-2026-11-15/A/redproof-review-4.txt).
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
The hub's 30-day check rests on the box's own reports: a box could name an app
that never existed, or stop listing a living one, wait 30 days, and get every
snapshot agreed beside the cap, with no operator signal.
- One window agrees to at most the size of its own cap, from the hub's OWN
count of the store; the oldest files first; the rest waits for the next
window. A narrowing of D84, decided by CC unattended — the operator may
reverse it (07 §8.2f holds the question, both options and their cost).
- offsite_end_claim_unseen (warning, mailed, operator-only): a box first names
an app the hub never saw installed — once per household and app, at most 8
per household, 30 days before anything can be agreed.
- offsite_ended_large_agreement (warning, mailed, operator-only): a request
for more than half of the store waits one week; that window agrees to none.
- 07 §8.2f: the limit rewritten (no honest ended app is needed, only 30 days),
what bounds it, what is still open. 08 §6.5: the two events.
Also carries the red-proof outputs of the controller's review fixes 2 and 3
(felhom-controller 3928c3d and the next commit).
Tests seen red on bb226f23; 15 mutations seen red (redproof-review-1.txt).
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
- 07 §8.2f: [FACT — built, not released]; D84 ruled (decision 309); the five
parts as built, the allowance, the audit, the mail; every edge case with the
test that pins it; what differs from the design; the limit (the hub cannot
read a snapshot's tags); every household text as built, Hungarian and
English, with the mail; what a box must show before a release. A household
with no notification address gets the page only.
- 08 §6.5 and §8.1: the receipt offsite_end_scheduled and the two operator
records.
- 05: the app record's table in the delete rules; the window API's new fields.
- The controller's red-proof outputs (56 mutations).
- hub tests: the English fixture sentence uses the box's quote marks.
Nothing is released and no box has run it.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
A box could send 300 new invented app names in every report and the
household's presence rows grew with no end.
- At most 400 names per household. A new name above that takes the place of
the oldest ABSENT name the same report does not claim; when there is none it
is not stored. Said once in the hub's log, not per report.
- A name is stored only when it is shaped like an app name (64 characters,
letters, digits, . _ -) and is not reserved (_shares, the marker, the box's
infrastructure stacks) — from the app list and from an end record alike.
The same check decides what a window may agree to.
- No age limit on a row, on purpose: an aged-out row would start a real ended
app's 30 days again. Making room never makes an app eligible earlier.
Tests TestD72_AppRecord_* (a box that sends new names in 40 reports);
10 mutations seen red, 2 after a redo.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
Built, not deployed. `07-backup-architecture.md` §8.2f parts 3-4 and decision 309.
- store: the hub's own record of which apps a box runs, by the hub's clock
(offsite_app_presence, fed by every controller report); it goes with the household.
- offsitekeys: window-open takes the box's due end records and answers
ended_allowed / ended_refused. An app is agreed only after 30 days of absence
in the hub's record; an id only when the hub's own listing holds it, uploaded
before that absence. The cap (max_remove) is unchanged; the close keeps only
agreed ids from removed_for_ended_apps and allows them beside the cap.
- The 07:10 audit explains the ended apps' vanished files by name and still
raises for every other vanished file.
- notify: offsite_end_scheduled is a household receipt — one mail per press to
the household's own address, whatever its switches; the operator is not mailed.
- Two operator-only hub records: offsite_ended_app_removed (info),
offsite_ended_app_refused (warning).
Tests TestD72_*, TestD84_*; 44 mutations seen red, 5 after a redo
(documentation/audits/build-release-4-2026-11-15/A/redproof-hub-*.txt).
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
A hub daily check skipped while the hub was down; an operator mail that was never sent; a signed job a box did not
take; a signed job a box refused, and the guest-power watchdog's give-up (both from the agent). Operator-only.
No new table. Not deployed.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
The host report's new sent_at (the box clock at the moment of sending) is compared with the hub's clock at arrival.
Both directions; told once; clears within 30 s. A report that waited on the box raises nothing (reported_at is not
read). A second box off in the same direction points at the hub's own clock. Operator-only. Not deployed.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
- R-528: NOT closed. demo-felhom's night oom_check failed once (2026-11-11 04:40 UTC, Docker 29.9.3,
containerd 2.4.2-3); 56 others passed; all four app_oom events since 10-12 explained. Due date 2026-11-28.
This clears the due-checks gate that kept nightly CI red (runs 2628, 2630, 2632).
- R-1391: Go 1.27.3 (hub, controller, agent) has GO-2026-6859; fixed in 1.27.4.
- R-1392: the hub's "root files behind" compares the whole bundle sha, which moves with every agent release
even when configs/ did not change; the os_config_bundle_behind alarm is due ~11-17 on all three boxes.
- STATUS: D81 + the 11-14 do line; old Updated paragraph moved to STATUS-HISTORY. REPORT overwritten.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
- Golden 0.352.0: bake, publish, one fresh install, the operator's vouch (R-468).
- No clone holds a login in its address; the credential helper script is in the repository (R-925, R-1390).
- N100 clock: a stale cached AppArmor profile stops chrony (R-1301, new R-1388). Read only.
- Certificate pin spike on demo-hp (R-1380): the break, the proven hand repair (runbook), the CA-file test,
03-host-agent.md 12a, decision D80.
- Rulings of 13:17 written down (09 decisions 297-302). Register 245 -> 248 (R-1388, R-1389, R-1390).
- Docs, evidence and one helper script. No release, no floor change, no hub deploy.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
Read-only review of every job on a timer and of everything that ends on a date,
with measurements from the three test boxes. Nothing ends in the next 14 days.
Register 236 -> 245 (R-1379..R-1387), pointer lines on twelve rows, D76-D79 in
STATUS.md. Part C: felhom-agent's clone address holds no token now; the other
clones wait for the operator's yes. R-1035's 4-week check moved to 2026-11-18
(operator read needed).
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
The close-out of the attended release run of 2026-11-10 (afternoon).
- Evidence: documentation/audits/release-space-fixes-2026-11-10/ (agent, controller, hub) and the hub row after the
fallback-kernel test.
- Rows: dated lines on R-1368, R-1371, R-1301, R-1353. 236 open before and after; opened 0, closed 0.
- Capability map: dated notes under C13 and G5; no status changed.
- disk-full runbook and 16 §14: the self-update folder is PROVEN as a product path on three test boxes.
- STATUS: new "Updated" paragraph; the old one moved to STATUS-HISTORY.
- REPORT-release-space-fixes-2026-11-10.md: wrong claims first, one deviation (the hub deploy ran before CI for the
manifest-bump commit was read), what stays not proven.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
The one-shot flag booted 7.0.14-26-pve: every health point good after 313 s. An orderly restart
brought 7.0.14-28-pve back: 324 s. The pinned default, grub.cfg and the step's state file did not
change. No kernel report and no alarm at the hub. Evidence: evidence/fallback-boot/.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
D71: the hub keeps the record of an off-site account after the tier is unticked (R-1375).
D72: off-site copies of a removed app end 30 days after 'delete backup data' (R-1374).
D73: a household or box delete removes its rows; the delete journal stays (R-1376).
D74: the runbook 'a household or a tester leaves' is a later documents-only session.
D75: the smallest system disk is 200 GB; the installer's act below it is open (R-1370).
Accepted: the heading correction of runbooks/ep0-datastore-copy.md (closed R-901).
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
C4: documentation/runbooks/disk-full.md (R-1369). Five disks; each hand step
is PROVEN with its date and evidence, or NOT PROVEN.
C5: 16-storage-physical-identity.md §14 (D69, decision 288). 45 places, each
with limit, cleaner and alarm; the rulings behind "no limit"; the host disk
budget; the open design points. PROMPT-TEMPLATE §N.5 gets the keep-it-true line.
Close-out: STATUS (D75 new), CONTEXT, dated lines on R-1369, R-1370, R-1371,
three small stale-document fixes, the report of the run.
Documents only. No release, no build, no deploy.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
host_disk_warning and host_disk_critical join the operator-only register, so a
household that already has the type saved gets no mail. The type leaves the
claim seed list. storage_fill_* is split by storage kind: a drive the household
can free (usb, local-dir, nfs, cifs) still reaches the household; the thin pool,
a PBS store, an unknown kind and a missing kind go to the operator only.
Tests: TestD68_HostDiskIsOperatorOnly_EvenWhenTheHouseholdSavedIt,
TestD68_StorageFillReachesAHouseholdOnlyForADriveItCanFree,
TestD68_RegisterAndGate, TestD68_SeedHoldsNoOperatorOnlyType. Red seen three
times (register lines out; gate reading the register only; type back in the seed).
No hub build, no deploy. Also: dated lines on R-1368 and R-1371 for the agent
(776025c8) and controller (844f19f1) commits, and the filesystem-identity
measurement the controller commit cites.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
One run on the demo-hp host (Tier 0), decision 286. The rule removed six of
eight kernel image packages and nothing else: 6.4 GB gained on the system
disk, every boot setting unchanged, one orderly reboot back on the same
kernel, all containers healthy 311 s after the command, the crash guard saw
a clean stop, a second run removed nothing. Five refusals shown by
simulation only.
- documentation/audits/kernel-cleanup-spike-2026-11-10/: README (one page),
evidence 00-15, the rule as a read-only plan script
- 11-os-updates.md §5.11.1: the design of the step (DESIGN, not built)
- R-1368: a dated pointer line
felhom-os-apply, the config bundle, the agent and rule R4 are unchanged.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
What every thing a customer can end leaves behind, and who is told.
- documentation/audits/endings-list-2026-11-10/: README (one page), TABLE (67 endings x 8 questions,
each answer with a mark and a pointer), MATRIX, the search list with controls, the helpers' raw
answers, and the ageing drill on scratch guest 9202 (five rounds, put back).
- Marks: A 499, D 0, K 15, N 22. 9 endings are not named in the architecture folder.
- Register: six new rows, one per class (R-1373 … R-1378); 230 -> 236 open rows. Nine existing rows
get a dated pointer line; none is closed, merged or re-ranked.
- STATUS.md: four decisions for the operator (D71 … D74).
No code change. No release. Machines: scratch guest 9202 only, put back.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
D67 yes as a separate step (R4 unchanged), D68 operator only, D69 a section
inside 16, D70 yes with three changes; DooPlex is never measured, ep0 not yet.
STATUS: D67-D70 out. Rows R-1368, R-1370, R-1371 carry the ruling.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
Read-only measurement on demo-hp, demo-felhom and Tester 1, two readings.
Audit folder documentation/audits/space-list-2026-11-10/ (README, TABLE,
MATRIX, evidence, tools). 360 answers: A 294, D 0, K 19, N 47.
Register 225 -> 230: R-1368 (host places that only grow: kernels, agent
binaries), R-1369 (no written way out of a full disk), R-1370 (no space
budget for the host disk), R-1371 (no space alarm; host-disk mail goes to
the household), R-1372 (small guest places). A dated line on nine existing
rows. Section headings set to the real counts; R-1268 and R-1312 moved to
their category's section.
STATUS: new Updated paragraph, decisions D67-D70. N100 diary line.
Two stale facts corrected in 05-hub-architecture.md and 11-os-updates.md.
Nothing was changed on any machine.
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS