hub (unreleased): a deleted box or household leaves no rows in the tables that had no delete; the delete journal stays (R-1376, D73)
gates / gates (push) Successful in 2m4s

Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
This commit is contained in:
admin committed 2026-11-14 18:32:34 +01:00
1 parent 0ce752a370
commit 73d99d06c5
12 files changed
+440 -5

No files matched your search

@@ -437,6 +437,32 @@ repository's fake provider (`hetznerapi.Fake`); 9 mutations seen red, one of the
added (`audits/night-build-2026-11-14/B/redproof-hub-offsite-record.txt`).
### 14.2 What a deleted box or household leaves in the hub [FACT — built, not released: hub `main` after v0.181.0, 2026-11-14; D73, `09` §3 decision 293, R-1376]
The endings list found thirteen hub tables with no delete statement anywhere. Seven of them hold rows that name a box
or a household. The ruling: a delete removes them, with one exception — the delete journal.
| Table | Named by | Goes when | How |
|---|---|---|---|
| `os_host_settings`, `os_reports`, `os_kernel_notices` | the box | the box is deleted | `store.DeleteHost`, in its one transaction |
| the hub's own alarm marks that name the box (`hub_settings`, `os_alarm:<kind>:<box>…`) | the box | the box is deleted | the same transaction; matched on the whole id between colons, so a box whose id only starts with this id keeps its marks |
| `offsite_keys`, `offsite_windows`, `offsite_abandon_requests` | the household | the household is deleted (the purge leg, after the outside legs) | `store.PurgeCustomerResidue`; counted in the delete dialog's residue |
| `operator_actions` (the operator's own action log) | the household | **one year after the deletion**, with `events` and `notification_log` | `store.PruneDeletedCustomerAudit`. *Decided by CC unattended — the operator may reverse:* it is an audit record of the same kind as the other two, and the one-year rule (decision 181) is the design's rule for those; the other option, at once, loses „who did what” on the day of the deletion |
| **`host_deletions`, `customer_resets`** — the delete journal | both | **never** | the ruling's one exception: an id and timestamps, no personal data. The privacy text must say so (proposed, not published) |
**Rows left by boxes and households deleted before this release** (Peti's box, the test boxes of 2026-11): the daily
prune runs `store.PurgeOrphanRows` — the six tables of the first and third rows, for a box with no `hosts` row and for
a household with neither a config row nor a box. A household in the middle of its delete (boxes gone, config still
there) and a „ghost” that still has a box lose nothing. It finds something once.
**Not in this:** the six tables that name no box or household (`api_tokens`, `os_candidates`, `os_releases`,
`r172_probe` and the two journals). The update reports of a LIVE box still grow with no end: an age limit there needs
its own design, because the kernel lane reads a box's old reports to know which kernel it must never offer again.
**Not proven:** on a deployed hub; no hub database was read. Tests on the real store with look-alike ids
(`TestD73_*`), 16 mutations seen red (`audits/night-build-2026-11-14/B/redproof-hub-d73.txt`).
## 15. Customer e-mails — what the hub writes to a household [hub v0.118.0, R-558]
**The section this document did not have.** The hub composes every sentence a household reads before
@@ -0,0 +1,82 @@
== MUTATION: a deleted box keeps its update reports (the old behaviour)
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_DeleteHostRemovesTheBoxesRows (0.05s)
d73_gone_rows_test.go:92: os_reports: 2 row(s) left for the deleted box (want 0)
== MUTATION: a deleted box keeps its update settings
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_DeleteHostRemovesTheBoxesRows (0.05s)
d73_gone_rows_test.go:92: os_host_settings: 1 row(s) left for the deleted box (want 0)
== MUTATION: a deleted box keeps its kernel notices
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_DeleteHostRemovesTheBoxesRows (0.05s)
d73_gone_rows_test.go:92: os_kernel_notices: 1 row(s) left for the deleted box (want 0)
== MUTATION: the alarm marks are matched by LIKE on the box's id (a look-alike box loses its marks)
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_DeleteHostRemovesTheBoxesRows (0.05s)
d73_gone_rows_test.go:92: alarm marks: 0 row(s) left for the deleted box (want 10)
d73_gone_rows_test.go:107: another box (box-a1b) has 0 alarm mark(s) left, want its 5
d73_gone_rows_test.go:107: another box (x-box-a1) has 0 alarm mark(s) left, want its 5
== MUTATION: the alarm marks of a deleted box stay
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_DeleteHostRemovesTheBoxesRows (0.05s)
d73_gone_rows_test.go:92: alarm marks: 15 row(s) left for the deleted box (want 10)
== MUTATION: the box delete also removes the HOUSEHOLD's off-site key record
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_DeleteHostRemovesTheBoxesRows (0.05s)
d73_gone_rows_test.go:97: offsite_keys: 1 → 0 — a box delete must leave the HOUSEHOLD's rows (the household still exists)
6 mutations, 0 not a red-proof
== MUTATION: a deleted household keeps its off-site key record
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_PurgeRemovesTheHouseholdsRows_JournalStays (0.04s)
d73_gone_rows_test.go:137: the preview's residue does not count the three tables: &{Reports:0 AppTelemetry:0 AppLogTails:0 LogTailRequests:0 NotificationPrefs:0 SelfBindTokens:0 ApplianceRegistrations:0 AppLogIssueRefs:0 OffsiteKe
== MUTATION: a deleted household keeps its abandon requests
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_PurgeRemovesTheHouseholdsRows_JournalStays (0.04s)
d73_gone_rows_test.go:137: the preview's residue does not count the three tables: &{Reports:0 AppTelemetry:0 AppLogTails:0 LogTailRequests:0 NotificationPrefs:0 SelfBindTokens:0 ApplianceRegistrations:0 AppLogIssueRefs:0 OffsiteKe
== MUTATION: the three tables are purged and not counted in the preview
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_PurgeRemovesTheHouseholdsRows_JournalStays (0.04s)
d73_gone_rows_test.go:137: the preview's residue does not count the three tables: &{Reports:0 AppTelemetry:0 AppLogTails:0 LogTailRequests:0 NotificationPrefs:0 SelfBindTokens:0 ApplianceRegistrations:0 AppLogIssueRefs:0 OffsiteKe
3 mutations, 0 not a red-proof
== MUTATION: the operator's action log of a deleted household is never deleted
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_PurgeRemovesTheHouseholdsRows_JournalStays (0.04s)
d73_gone_rows_test.go:171: operator_actions a year after the deletion: 1, want 0
== MUTATION: the one-year prune also empties the delete journal
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_PurgeRemovesTheHouseholdsRows_JournalStays (0.05s)
d73_gone_rows_test.go:178: journal host_deletions after the one-year prune: 0 row(s), want 1
2 mutations, 0 not a red-proof
== MUTATION: the sweep removes a LIVE box's reports (no owner check)
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_PurgeOrphanRows_OnlyTheGone (0.05s)
d73_gone_rows_test.go:214: live / os_reports: 0 row(s), want 2
d73_gone_rows_test.go:214: ghost / os_reports: 0 row(s), want 2
d73_gone_rows_test.go:219: the sweep reported map[offsite_abandon_requests:1 offsite_keys:1 offsite_windows:1 os_host_settings:2 os_kernel_notices:2 os_reports:8] (total 15), want 11 rows: 8 box rows of mid and gone, 3 household r
== MUTATION: the sweep removes the off-site rows of a household in the middle of its delete
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_PurgeOrphanRows_OnlyTheGone (0.04s)
d73_gone_rows_test.go:214: mid / offsite_windows: 0 row(s), want 1
d73_gone_rows_test.go:219: the sweep reported map[offsite_abandon_requests:1 offsite_keys:1 offsite_windows:2 os_host_settings:2 os_kernel_notices:2 os_reports:4] (total 12), want 11 rows: 8 box rows of mid and gone, 3 household r
== MUTATION: the sweep removes the off-site rows of a household that still has a box
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_PurgeOrphanRows_OnlyTheGone (0.05s)
d73_gone_rows_test.go:214: ghost / offsite_keys: 0 row(s), want 1
d73_gone_rows_test.go:219: the sweep reported map[offsite_abandon_requests:1 offsite_keys:2 offsite_windows:1 os_host_settings:2 os_kernel_notices:2 os_reports:4] (total 12), want 11 rows: 8 box rows of mid and gone, 3 household r
== MUTATION: the sweep leaves the kernel notices of a gone box
RED on the mutant, GREEN on the real file
--- FAIL: TestD73_PurgeOrphanRows_OnlyTheGone (0.05s)
d73_gone_rows_test.go:214: mid / os_kernel_notices: 1 row(s), want 0
d73_gone_rows_test.go:214: gone / os_kernel_notices: 1 row(s), want 0
d73_gone_rows_test.go:219: the sweep reported map[offsite_abandon_requests:1 offsite_keys:1 offsite_windows:1 os_host_settings:2 os_reports:4] (total 9), want 11 rows: 8 box rows of mid and gone, 3 household rows of gone
4 mutations, 0 not a red-proof
== MUTATION: the daily prune does not call the sweep
RED on the mutant, GREEN on the real file
--- FAIL: TestR1383_MainWiresTheSilentFailureChecks (0.00s)
r1383_wiring_test.go:74: main.go: orphan sweep in the daily prune — found 0 time(s), want 1
1 mutations, 0 not a red-proof
+1 -1
View File
@@ -355,7 +355,7 @@ stopping line that lies.
| **R-1345** | Hub & operator | P4 | **An access token's scopes cannot be changed after it is made — widening or narrowing one means a new token (and a new secret).** The operator, 2026-10-29: to give CC `hosts:read` + `hosts:write` for the live proofs he had to make `CC_HOSTS` (now in `sec`) beside `cc-releases-2`. Wanted: edit a token's scopes in the session (an audit line per change), so a scope can also be taken away after a session without retiring the secret. | OPEN | — | design + build the scope edit (session only), red-first test that an unticked scope is refused at once | CC |
| **R-1353** | Hub & operator | P4 | **Statements still over 1 s, named by the new slow-statement line (hub 0.178.0) in its first hour:** a background deadline/restore-test check reading days of full host-report JSON (`GetHostReportsSince`, `store.go`; 1.5–9.2 s cold, from `monitor/deadline.go` and `monitor/restoretest.go` — no page waits on it); the start-up prune of `events` and `app_log_issues` (1.1 s and 1.8 s; no index on `created_at` / `last_seen`); the customer page's newest-report read (`GetCustomer`, 1.3 s cold in the pod probe). With the 1 GiB limit most of this should come from the cache once warm. Evidence: the hub log 2026-10-30 09:22–10:05 UTC; `audits/release-2026-10-30/p2-r1346/`. | OPEN **2026-11-02 read (`audits/security-2026-11-02/part4/hub-slow-lines-2026-11-01-02.txt`; the log covers only the 31 h since DooPlex restarted on 2026-11-01 16:11 CET, the lines from 10-30 are gone with the old pod):** 35 lines, 15 statements. What repeats: `VACUUM INTO` 5× (avg 110 s, the 6-hourly snapshot, expected); `GetHostReportsSince` 9× (avg 7.9 s, max 26.6 s); `GetCustomer`'s newest report 3× (1.1 s). Once, at the cold start: the prune of `events` (10.4 s) and `app_log_issues` (21 s, again 04:30), the OS-release and host lists (2–22 s). No fix before the full week (11-06). **2026-11-10 read (the hub's start after a deploy, `audits/release-space-fixes-2026-11-10/hub/10-watch.txt`):** 11 slow lines in the first 2 minutes of the new pod, the longest 19.1 s (`GetHostReportsSince`), then the OS-report reads (3–4 s) and the two prunes (2.7 s, 4.2 s); none in the 28 minutes after. The 13:02 UTC snapshot's `VACUUM INTO` took 127.8 s. The same statements as the 11-02 read; no new one. | — | read a week of slow-statement lines; index or narrow what still repeats (the history read needs only a few fields, not the whole JSON) | CC |
| **R-1375** | Hub & operator | P2 | **The outside part of three endings is done by hand or not at all — the worst: an off-site account can stay at the provider with no record in the hub.** Found 2026-11-10 (the endings list, `documentation/audits/endings-list-2026-11-10/TABLE.md` endings EH06, EH01, EA21 and the ending "a tester ends"; read in source and documents; no account was called). (1) **Unticking the off-site tier** of a household takes nothing down and leaves no reminder (`hub/internal/web/configs.go:1737-1743`). A later RESET takes the account down only when the tier is ticked (`hub/internal/web/customer_reset.go:203-245`), so the sub-account and its data stay, are paid for, and the hub holds no record that they exist. (2) **Switching the second off-site tier off** takes nothing down: storage entry, token and copies stay; the source calls the teardown 'a deliberate future op' (agent `internal/pbsdr/manager.go:274`). (3) **A tester ends**: no handler and no runbook; done by hand once (2026-09-25); no list says what to collect. **Members (answers marked N):** EH06-4, EH01-4, EA21-4, and 'a tester ends' in the table's head. Related: R-810 (a household leaves), R-32. | NARROWED — part 1 (D71) is built, not released. **2026-11-14 (night build):** the hub keeps the record of an off-site account after the tier is unticked; RESET and delete take the account down when a record exists; an untick alone takes nothing down (pinned by a test); the config form says that the account still exists (`05` §14.1). The runbook „a household or a tester leaves” is written (D74, `runbooks/household-leaves.md`). **For the operator, once, after the release:** at the provider's page read each sub-account's label `felhom-customer=<id>`; for a household whose tier is unticked in the hub, tick it again and save — the account's record comes back. LEFT: part (2) has no fault (the second tier keeps its record, read in source and its test); part (3) is the runbook; a household mail at leaving stays R-810. OPEN — filed 2026-11-10 (endings list) **2026-11-10 15:19 — RULED (D71 = yes, D74 = yes; `09` §3 decisions 291, 294).** Not built. | — | Hub (D71): keep the account record after an untick; RESET and delete take the account down, ticked or not. Docs (D74): the runbook 'a household or a tester leaves', from the five code doors, in a later documents-only session. Parts (2) and (3) of this row have no ruling of their own. | CC |
| **R-1376** | Hub & operator | P3 | **Records of a thing that is gone stay in six places.** Found 2026-11-10 (the endings list, `documentation/audits/endings-list-2026-11-10/TABLE.md` endings EH00, EH02, EH21, EA00, EA11, EA27; read in source, `documentation/audits/endings-list-2026-11-10/evidence/hub-delete-statements.txt`; no hub database was read). (1) **Hub:** thirteen tables have no delete statement anywhere. Deleting a box leaves its update settings, its update reports and its kernel notices; deleting a household leaves its off-site key rows, off-site window rows and abandon requests. Most are small; the update reports of each box grow with time. (2) **Host uninstall** leaves the mount units and host mounts of household drives (said when R-179 closed; no open row carried it). (3) **A drive wiped by a signed job** keeps a record and a mount unit that name the old filesystem (agent `internal/signedjobs/wipe.go:61-110`). (4) **The token of a destroyed guest** stays valid until a new guest takes its number: the token store has no remove (agent `internal/localapi/tokenstore.go:127-210`). **Members (answers marked N):** EH00-4, EH02-4, EH21-4, EA00-4, EA11-4, EA27-4. | OPEN — filed 2026-11-10 (endings list) **2026-11-10 15:19 — RULED (D73 = yes, with one exception; `09` §3 decision 293).** Not built. | — | Hub (D73): add the tables with no delete to the two purges and to the daily clean-up; the delete journal (`host_deletions`, `customer_resets`) stays on purpose, and the privacy text says so. Agent: remove a guest's token when the guest is destroyed. Parts (2) and (3) have no ruling of their own. | CC |
| **R-1376** | Hub & operator | P3 | **Records of a thing that is gone stay in six places.** Found 2026-11-10 (the endings list, `documentation/audits/endings-list-2026-11-10/TABLE.md` endings EH00, EH02, EH21, EA00, EA11, EA27; read in source, `documentation/audits/endings-list-2026-11-10/evidence/hub-delete-statements.txt`; no hub database was read). (1) **Hub:** thirteen tables have no delete statement anywhere. Deleting a box leaves its update settings, its update reports and its kernel notices; deleting a household leaves its off-site key rows, off-site window rows and abandon requests. Most are small; the update reports of each box grow with time. (2) **Host uninstall** leaves the mount units and host mounts of household drives (said when R-179 closed; no open row carried it). (3) **A drive wiped by a signed job** keeps a record and a mount unit that name the old filesystem (agent `internal/signedjobs/wipe.go:61-110`). (4) **The token of a destroyed guest** stays valid until a new guest takes its number: the token store has no remove (agent `internal/localapi/tokenstore.go:127-210`). **Members (answers marked N):** EH00-4, EH02-4, EH21-4, EA00-4, EA11-4, EA27-4. | NARROWED — part 1 (the hub, D73) is built, not released. **2026-11-14 (night build):** deleting a box removes its update settings, update reports, kernel notices and the hub's alarm marks that name it; deleting a household removes its off-site key, window and abandon rows; the operator's action log of a deleted household goes one year after the deletion with the other audit rows (decided by CC unattended — the operator may reverse); the delete journal stays; a daily sweep removes what earlier deletes left (`05` §14.2). The privacy sentence is proposed in the report, not published. LEFT: the agent part (a destroyed guest's token), parts (2) and (3), and an age limit for a LIVE box's update reports (needs a design: the kernel lane reads old reports). OPEN — filed 2026-11-10 (endings list) **2026-11-10 15:19 — RULED (D73 = yes, with one exception; `09` §3 decision 293).** Not built. | — | Hub (D73): add the tables with no delete to the two purges and to the daily clean-up; the delete journal (`host_deletions`, `customer_resets`) stays on purpose, and the privacy text says so. Agent: remove a guest's token when the guest is destroyed. Parts (2) and (3) have no ruling of their own. | CC |
| **R-1386** | Hub & operator | P3 | **Timed jobs with no time limit and no measured duration (25 answers).** Found 2026-11-11 (the Clock list, class W4, `documentation/audits/clock-list-2026-11-11/TABLE.md`; read in source; the hub was not measured, by ruling). The hub's 60-second sweep runs its checks one after the other with no limit for each: one slow check delays all the others, the liveness check too (CH19-3, CH19-8, CH01-3, CH08-3, CH08-8, CH10-3, CH13-8, CH18-3, CH06-8, CH02-3); the 07:10 audit has 10 minutes, measured with three boxes only (CH11-3, CH11-8). On the box: the interval jobs have no limit of their own (CC11-3, CC12-3, CC13-3, CC14-3, CC15-3, CC16-3, CC20-3, CA06-3, CA10-3, CA10-8, CA13-8, CA15-3, CA19-3). Every measured duration is from a small test store. | OPEN — filed 2026-11-11 (clock list) | — | Hub: log each sweep step's duration and alarm when one sweep takes longer than its interval; then read a week of numbers and set limits. Box: the same for the controller's interval jobs. | CC |
| **R-1393** | Hub & operator | P4 | **For about 8 minutes after a hub restart its database statements take 1 to 32 seconds; the System page took 12.7 s and a box's enrolment 11.4 s.** Measured 2026-11-14 at the 0.181.0 deploy (`audits/release-1-dates-2026-11-14/hub/05-watch.txt`): 24 `slow statement` lines from 09:58 to 10:06 UTC, none in the 20 minutes after. The database file is 678 MB; the slow ones read `host_reports`, `os_reports` and `events` (whole report JSON per row). One 32-second statement ran while CC copied the database out of the pod for a read (CC's own load). No report was lost that CC could see; not measured: whether a box's report timed out in that window. | OPEN — filed 2026-11-14 | — | measure a restart with no copy running; then decide: an index or a shorter keep on `host_reports`/`os_reports`, or accept | CC |
+2 -2
View File
@@ -71,8 +71,8 @@ The hub then runs, in this order: the journal entry → every box deleted (door
measured (`runbooks/ep0-datastore-copy.md`).
- The purge of the hub's rows: PROVEN once (482 reports, 1,036 telemetry rows, 1 log tail gone; journal #20).
**Known gap, R-1376:** rows in 13 tables stay (the box's update settings and reports, its kernel notices, the
off-site key and window rows, the abandon requests). Ruled to be removed (D73, `09` §3 decision 293); read the
row for whether that is released — until then they stay.
off-site key and window rows, the abandon requests). Built to be removed, not released (hub `main`, 2026-11-14;
D73, `05` §14.2) — until that release they stay.
### Step 5 — if a leg fails: press again.
The dialog offers Resume. A leg that is done is not done again. Nothing is half forgotten: the journal names the leg.
+6
View File
@@ -3,6 +3,12 @@
- `bundlefiles`: the line „bundle X holds the same files as the vouched Y” is logged once per pair (INFO), not at
every view of the System page (it was 6 DEBUG lines per view, seen at the 0.181.0 deploy). No behaviour change.
- **Release 3 („the hub tells you”), built 2026-11-14 night, NOT deployed. Every new alarm is operator-only.**
- **A deleted box or household leaves no rows in the tables that had no delete (R-1376 part 1, D73, `09` §3 decision
293; `05` §14.2).** `DeleteHost` also removes the box's `os_host_settings`, `os_reports`, `os_kernel_notices` and the
hub's alarm marks that name it; `PurgeCustomerResidue` also removes the household's `offsite_keys`,
`offsite_windows`, `offsite_abandon_requests` (and the delete dialog counts them); `operator_actions` goes with the
audit rows one year after the deletion. **The delete journal stays** (`host_deletions`, `customer_resets`).
`PurgeOrphanRows` in the daily prune removes what earlier deletes left. Tests `TestD73_*`; 16 mutations seen red.
- **Two end dates warn 14 days before (R-1379, D77, `09` §3 decision 298; `08` §12).** `monitor.EndDateChecker`, hourly
on the sweep: `operator_token_ends_soon` for an operator access token that is not revoked, and `web_cert_ends_soon`
for a web certificate a box reports in `web_certs` (controllers after 0.352.1). Told once at 14 days and once more,
+7
View File
@@ -1214,6 +1214,13 @@ func pruneAll(s *store.Store, maxDays int, logger *log.Logger) {
} else if n > 0 {
logger.Printf("[INFO] Pruned %d stale app issues", n)
}
// D73 (`09` §3 decision 293, R-1376): the rows a box or household deleted BEFORE this release left in the six
// tables that had no delete. A delete now leaves none, so this finds something only once.
if orphans, err := s.PurgeOrphanRows(); err != nil {
logger.Printf("[ERROR] Purge rows of deleted boxes and households: %v", err)
} else if orphans.Total() > 0 {
logger.Printf("[INFO] Purged %d row(s) of boxes and households that no longer exist: %v", orphans.Total(), map[string]int64(orphans))
}
// R-901 (`09` §3 decision 181): a deleted customer's audit rows go 1 year after the deletion.
if ev, nl, err := s.PruneDeletedCustomerAudit(time.Now().UTC(), store.DeletedCustomerAuditKeep); err != nil {
logger.Printf("[ERROR] Prune deleted customers' audit rows: %v", err)
+3 -1
View File
@@ -60,6 +60,8 @@ func TestR1383_MainWiresTheSilentFailureChecks(t *testing.T) {
seen["lost-mail scan"]++
case recv == "signedJobWaitChecker" && name == "Check":
seen["signed-job sweep"]++
case name == "PurgeOrphanRows" && fn.Name.Name == "pruneAll":
seen["orphan sweep in the daily prune"]++ // D73, R-1376
case recv == "endDateChecker" && name == "Check":
seen["end-date sweep"]++ // R-1379
}
@@ -67,7 +69,7 @@ func TestR1383_MainWiresTheSilentFailureChecks(t *testing.T) {
return true
})
}
for _, k := range []string{"ledger built", `watched "deadline-check"`, `watched "offsite-key-audit"`, "lost-mail scan", "signed-job sweep", "end-date sweep"} {
for _, k := range []string{"ledger built", `watched "deadline-check"`, `watched "offsite-key-audit"`, "lost-mail scan", "signed-job sweep", "end-date sweep", "orphan sweep in the daily prune"} {
if seen[k] != 1 {
t.Errorf("main.go: %s — found %d time(s), want 1", k, seen[k])
}
+17 -1
View File
@@ -24,6 +24,13 @@ import "fmt"
// What deliberately SURVIVES (unchanged from every other tier): `events` (the audit stream),
// `notification_log` (the send-attempt audit), `host_deletions` and `customer_resets` (F-14
// provenance). The audit trail outlives every lifecycle tier — that rule is not relaxed here.
//
// D73 (`09` §3 decision 293, R-1376): „deleting a household or a box deletes its rows in the tables that have no
// delete today” — with ONE exception, the delete journal (`host_deletions`, `customer_resets`), which stays for ever
// and which the privacy text must name. So: the three off-site tables go with this purge; the box's three update
// tables go with DeleteHost; and `operator_actions` — the operator's own action log, an audit record like `events`
// — goes with the audit rows, one year after the deletion (PruneDeletedCustomerAudit; decided by CC unattended,
// the operator may reverse). Rows left by households and boxes deleted BEFORE this: PurgeOrphanRows, daily.
// CustomerResidue counts what a customer has left behind OUTSIDE the config row: the report-derived
// state that keeps a deleted customer visible and alerting, plus the credential-bearing bindings.
@@ -39,6 +46,11 @@ type CustomerResidue struct {
// AppLogIssueRefs (R-244, v0.147.0): shared app_log_issues rows naming the customer. De-referenced
// (never blindly deleted) by the purge — see app_log_residue.go.
AppLogIssueRefs int
// D73 (`09` §3 decision 293, R-1376): the three household-keyed tables that had no delete statement anywhere —
// the off-site key record, the off-site clean-up windows and the abandon requests. They outlived the household.
OffsiteKeys int
OffsiteWindows int
OffsiteAbandonRequests int
}
// Total is the single number the "is there anything left at all?" decision hangs on.
@@ -47,7 +59,8 @@ func (r *CustomerResidue) Total() int {
return 0
}
return r.Reports + r.AppTelemetry + r.AppLogTails + r.LogTailRequests +
r.NotificationPrefs + r.SelfBindTokens + r.ApplianceRegistrations + r.AppLogIssueRefs
r.NotificationPrefs + r.SelfBindTokens + r.ApplianceRegistrations + r.AppLogIssueRefs +
r.OffsiteKeys + r.OffsiteWindows + r.OffsiteAbandonRequests
}
// residueQueries is the ONE list both the counter and the purge walk, so a table can never be
@@ -63,6 +76,9 @@ var residueQueries = []struct {
{"customer_notifications", func(r *CustomerResidue) *int { return &r.NotificationPrefs }},
{"selfbind_tokens", func(r *CustomerResidue) *int { return &r.SelfBindTokens }},
{"appliance_registrations", func(r *CustomerResidue) *int { return &r.ApplianceRegistrations }},
{"offsite_keys", func(r *CustomerResidue) *int { return &r.OffsiteKeys }},
{"offsite_windows", func(r *CustomerResidue) *int { return &r.OffsiteWindows }},
{"offsite_abandon_requests", func(r *CustomerResidue) *int { return &r.OffsiteAbandonRequests }},
}
// CustomerResidue counts the customer's report-derived + binding residue. Read-only.
+225
View File
@@ -0,0 +1,225 @@
package store
import (
"testing"
"time"
)
// d73Seed gives a box rows in the three box-keyed tables that had no delete, and its household rows in the three
// household-keyed ones — plus the hub's own alarm marks that name the box.
func d73Seed(t *testing.T, s *Store, hostID, customerID string) {
t.Helper()
exec := func(q string, a ...any) {
t.Helper()
if _, err := s.db.Exec(q, a...); err != nil {
t.Fatalf("seed %q: %v", q, err)
}
}
exec(`INSERT INTO os_host_settings (host_id, ring, enabled) VALUES (?, 0, 1)`, hostID)
exec(`INSERT INTO os_reports (host_id, layer, outcome, report_json) VALUES (?, 'host', 'nothing', '{}')`, hostID)
exec(`INSERT INTO os_reports (host_id, layer, outcome, report_json) VALUES (?, 'kernel', 'applied', '{}')`, hostID)
exec(`INSERT INTO os_kernel_notices (host_id, kver, sent_at) VALUES (?, '7.0.14-22-pve', datetime('now'))`, hostID)
exec(`INSERT INTO offsite_keys (customer_id, fingerprint) VALUES (?, 'SHA256:x')`, customerID)
exec(`INSERT INTO offsite_windows (customer_id, closes_by) VALUES (?, datetime('now'))`, customerID)
exec(`INSERT INTO offsite_abandon_requests (customer_id, path, due_at) VALUES (?, '.orphaned-1', datetime('now'))`, customerID)
exec(`INSERT INTO operator_actions (customer_id, host_id, action, requested_at) VALUES (?, ?, 'rerun_backup', datetime('now'))`, customerID, hostID)
for _, k := range []string{"box_clock:" + hostID + ":ahead", "pve_trust:" + hostID + ":ca_copy_does_not_verify",
"guest_power:" + hostID + ":9201", "signed_job_wait:" + hostID + ":job1", "wg_stale:" + hostID} {
if err := s.SetOSAlarmRaised(k, time.Now()); err != nil {
t.Fatal(err)
}
}
}
func d73Count(t *testing.T, s *Store, q string, args ...any) int {
t.Helper()
var n int
if err := s.db.QueryRow(q, args...).Scan(&n); err != nil {
t.Fatalf("count %q: %v", q, err)
}
return n
}
func d73Counts(t *testing.T, s *Store, hostID, customerID string) map[string]int {
t.Helper()
return map[string]int{
"os_host_settings": d73Count(t, s, `SELECT COUNT(*) FROM os_host_settings WHERE host_id = ?`, hostID),
"os_reports": d73Count(t, s, `SELECT COUNT(*) FROM os_reports WHERE host_id = ?`, hostID),
"os_kernel_notices": d73Count(t, s, `SELECT COUNT(*) FROM os_kernel_notices WHERE host_id = ?`, hostID),
"alarm marks": d73Count(t, s, `SELECT COUNT(*) FROM hub_settings WHERE key LIKE 'os_alarm:%' AND instr(key, ?) > 0`, hostID),
"offsite_keys": d73Count(t, s, `SELECT COUNT(*) FROM offsite_keys WHERE customer_id = ?`, customerID),
"offsite_windows": d73Count(t, s, `SELECT COUNT(*) FROM offsite_windows WHERE customer_id = ?`, customerID),
"offsite_abandon_requests": d73Count(t, s, `SELECT COUNT(*) FROM offsite_abandon_requests WHERE customer_id = ?`, customerID),
"operator_actions": d73Count(t, s, `SELECT COUNT(*) FROM operator_actions WHERE customer_id = ?`, customerID),
}
}
var d73BoxTables = []string{"os_host_settings", "os_reports", "os_kernel_notices", "alarm marks"}
var d73HouseholdTables = []string{"offsite_keys", "offsite_windows", "offsite_abandon_requests"}
// D73 (`09` §3 decision 293, R-1376): deleting a BOX deletes its rows in the three box-keyed tables that had no
// delete statement, and the hub's alarm marks that name it. The household's rows stay (the household is still
// there), the other box is untouched — also one whose id only STARTS with this box's id — and the delete journal has
// its row. COMPANION RED-PROOFS (observed, audits/night-build-2026-11-14/B/redproof-hub-d73.txt).
func TestD73_DeleteHostRemovesTheBoxesRows(t *testing.T) {
s := newTestStore(t)
for _, h := range []struct{ host, cust string }{{"box-a1", "cust-a"}, {"box-a1b", "cust-b"}, {"x-box-a1", "cust-c"}} {
if err := s.UpsertHost(&Host{HostID: h.host, CustomerID: h.cust, APIKey: "k-" + h.host}); err != nil {
t.Fatal(err)
}
d73Seed(t, s, h.host, h.cust)
}
before := d73Counts(t, s, "box-a1", "cust-a")
for table, n := range before {
if n == 0 {
t.Fatalf("fixture gap: %s has no row", table)
}
}
otherB, otherC := d73Counts(t, s, "box-a1b", "cust-b"), d73Counts(t, s, "x-box-a1", "cust-c")
if err := s.DeleteHost("box-a1", true); err != nil {
t.Fatal(err)
}
after := d73Counts(t, s, "box-a1", "cust-a")
// "alarm marks" counts by a substring, so the two look-alike boxes' marks (box-a1b, x-box-a1) are in it: those
// must be all that is left.
lookAlikes := 2 * 5
for _, table := range d73BoxTables {
want := 0
if table == "alarm marks" {
want = lookAlikes
}
if after[table] != want {
t.Errorf("%s: %d row(s) left for the deleted box (want %d)", table, after[table], want)
}
}
for _, table := range append(d73HouseholdTables, "operator_actions") {
if after[table] != before[table] {
t.Errorf("%s: %d → %d — a box delete must leave the HOUSEHOLD's rows (the household still exists)", table, before[table], after[table])
}
}
for name, pair := range map[string][2]map[string]int{"box-a1b": {otherB, d73Counts(t, s, "box-a1b", "cust-b")}, "x-box-a1": {otherC, d73Counts(t, s, "x-box-a1", "cust-c")}} {
for _, table := range []string{"os_host_settings", "os_reports", "os_kernel_notices", "offsite_keys", "operator_actions"} {
if pair[0][table] != pair[1][table] {
t.Errorf("another box (%s) lost rows in %s: %d → %d", name, table, pair[0][table], pair[1][table])
}
}
if n := d73Count(t, s, `SELECT COUNT(*) FROM hub_settings WHERE key LIKE 'os_alarm:%' AND (instr(key, ':'||?||':') > 0 OR key LIKE '%:'||?)`, name, name); n != 5 {
t.Errorf("another box (%s) has %d alarm mark(s) left, want its 5", name, n)
}
}
if n := d73Count(t, s, `SELECT COUNT(*) FROM host_deletions WHERE host_id = 'box-a1'`); n != 1 {
t.Errorf("the delete journal has %d row(s) for the box, want 1 — the journal STAYS", n)
}
}
// Deleting a HOUSEHOLD (the purge leg) deletes its rows in the three household-keyed tables, counts them in the
// preview's residue, and leaves the two journals alone. The operator's action log is an audit record: it stays at the
// purge and goes with the audit rows one year after the deletion.
func TestD73_PurgeRemovesTheHouseholdsRows_JournalStays(t *testing.T) {
s := newTestStore(t)
for _, c := range []string{"cust-a", "cust-b"} {
if err := s.SaveCustomerConfig(&CustomerConfig{CustomerID: c, APIKey: "k-" + c, RetrievalPassword: "p"}); err != nil {
t.Fatal(err)
}
d73Seed(t, s, "box-"+c, c)
}
id, err := s.StartCustomerReset("cust-a", false)
if err != nil {
t.Fatal(err)
}
_ = s.UpdateResetLeg(id, "customer_delete", "ok")
_ = s.FinishCustomerReset(id)
if _, err := s.db.Exec(`INSERT INTO host_deletions (host_id, customer_id, escrow_acked) VALUES ('box-cust-a', 'cust-a', 0)`); err != nil {
t.Fatal(err)
}
res, err := s.CustomerResidue("cust-a")
if err != nil || res.OffsiteKeys != 1 || res.OffsiteWindows != 1 || res.OffsiteAbandonRequests != 1 || res.Total() < 3 {
t.Fatalf("the preview's residue does not count the three tables: %+v (%v)", res, err)
}
if err := s.PurgeCustomerResidue("cust-a"); err != nil {
t.Fatal(err)
}
after, other := d73Counts(t, s, "box-cust-a", "cust-a"), d73Counts(t, s, "box-cust-b", "cust-b")
for _, table := range d73HouseholdTables {
if after[table] != 0 {
t.Errorf("%s: %d row(s) outlived the household", table, after[table])
}
if other[table] != 1 {
t.Errorf("%s: the OTHER household has %d row(s), want 1", table, other[table])
}
}
if after["operator_actions"] != 1 {
t.Errorf("operator_actions: %d — the action log is an audit record and stays at the purge", after["operator_actions"])
}
for _, j := range []string{"customer_resets", "host_deletions"} {
if n := d73Count(t, s, `SELECT COUNT(*) FROM `+j+` WHERE customer_id = 'cust-a'`); n != 1 {
t.Errorf("journal %s: %d row(s), want 1 — the delete journal stays (the ruling's one exception)", j, n)
}
}
// the action log goes with the audit rows: not at 100 days, gone after a year
now := time.Now().UTC()
if _, _, err := s.PruneDeletedCustomerAudit(now.Add(100*24*time.Hour), DeletedCustomerAuditKeep); err != nil {
t.Fatal(err)
}
if n := d73Counts(t, s, "box-cust-a", "cust-a")["operator_actions"]; n != 1 {
t.Errorf("operator_actions after 100 days: %d, want 1", n)
}
if _, _, err := s.PruneDeletedCustomerAudit(now.Add(400*24*time.Hour), DeletedCustomerAuditKeep); err != nil {
t.Fatal(err)
}
if n := d73Counts(t, s, "box-cust-a", "cust-a")["operator_actions"]; n != 0 {
t.Errorf("operator_actions a year after the deletion: %d, want 0", n)
}
if n := d73Counts(t, s, "box-cust-b", "cust-b")["operator_actions"]; n != 1 {
t.Errorf("the OTHER household's action log: %d, want 1", n)
}
for _, j := range []string{"customer_resets", "host_deletions"} {
if n := d73Count(t, s, `SELECT COUNT(*) FROM `+j+` WHERE customer_id = 'cust-a'`); n != 1 {
t.Errorf("journal %s after the one-year prune: %d row(s), want 1", j, n)
}
}
}
// The daily sweep is for what was left BEFORE: rows of a box or household that no longer exists. A live box, a live
// household, and a household in the middle of its delete (its boxes are gone, its config is still there) lose nothing.
func TestD73_PurgeOrphanRows_OnlyTheGone(t *testing.T) {
s := newTestStore(t)
// live: a household with a box
_ = s.SaveCustomerConfig(&CustomerConfig{CustomerID: "live", APIKey: "k1", RetrievalPassword: "p"})
_ = s.UpsertHost(&Host{HostID: "live-box", CustomerID: "live", APIKey: "hk1"})
d73Seed(t, s, "live-box", "live")
// mid-delete: the config is there, the box is already gone (so its box rows ARE orphans, its household rows are not)
_ = s.SaveCustomerConfig(&CustomerConfig{CustomerID: "mid", APIKey: "k2", RetrievalPassword: "p"})
d73Seed(t, s, "mid-box", "mid")
// a ghost with a box: the config row is gone, a box row is left → the household still „exists”
_ = s.UpsertHost(&Host{HostID: "ghost-box", CustomerID: "ghost", APIKey: "hk3"})
d73Seed(t, s, "ghost-box", "ghost")
// gone: neither config nor box (Peti's box, the test boxes of 2026-11)
d73Seed(t, s, "gone-box", "gone")
got, err := s.PurgeOrphanRows()
if err != nil {
t.Fatal(err)
}
want := map[string]map[string]int{
"live": {"os_host_settings": 1, "os_reports": 2, "os_kernel_notices": 1, "offsite_keys": 1, "offsite_windows": 1, "offsite_abandon_requests": 1, "operator_actions": 1},
"mid": {"os_host_settings": 0, "os_reports": 0, "os_kernel_notices": 0, "offsite_keys": 1, "offsite_windows": 1, "offsite_abandon_requests": 1, "operator_actions": 1},
"ghost": {"os_host_settings": 1, "os_reports": 2, "os_kernel_notices": 1, "offsite_keys": 1, "offsite_windows": 1, "offsite_abandon_requests": 1, "operator_actions": 1},
"gone": {"os_host_settings": 0, "os_reports": 0, "os_kernel_notices": 0, "offsite_keys": 0, "offsite_windows": 0, "offsite_abandon_requests": 0, "operator_actions": 1},
}
for cust, tables := range want {
c := d73Counts(t, s, cust+"-box", cust)
for table, n := range tables {
if c[table] != n {
t.Errorf("%s / %s: %d row(s), want %d", cust, table, c[table], n)
}
}
}
if got.Total() != 11 || got["os_reports"] != 4 || got["offsite_keys"] != 1 {
t.Errorf("the sweep reported %v (total %d), want 11 rows: 8 box rows of mid and gone, 3 household rows of gone", got, got.Total())
}
again, _ := s.PurgeOrphanRows()
if again.Total() != 0 {
t.Errorf("a second sweep removed %v — it must find nothing", again)
}
}
@@ -65,6 +65,11 @@ func (s *Store) PruneDeletedCustomerAudit(now time.Time, keep time.Duration) (ev
}
n, _ = r.RowsAffected()
notifications += n
// D73 (R-1376): the operator's action log of the deleted household is an audit record of the same kind — it had
// no delete at all; it goes with the other two.
if _, err = s.db.Exec(`DELETE FROM operator_actions WHERE customer_id = ? AND requested_at <= ?`, d.id, cutoff); err != nil {
return events, notifications, fmt.Errorf("PruneDeletedCustomerAudit %s: operator_actions: %w", d.id, err)
}
}
return events, notifications, nil
}
+55
View File
@@ -0,0 +1,55 @@
package store
import "fmt"
// OrphanRows counts what PurgeOrphanRows removed, per table.
type OrphanRows map[string]int64
// Total is the sum.
func (o OrphanRows) Total() int64 {
var n int64
for _, v := range o {
n += v
}
return n
}
// orphanTables are the tables D73 names (`09` §3 decision 293, R-1376) — the ones that had no delete statement — each
// with the column that names its owner and the query that says „the owner still exists”.
//
// A BOX exists while it has a `hosts` row. A HOUSEHOLD exists while it has a config row OR a box: a „ghost” (the
// config row gone, residue left) is exactly what this sweep is for, but a household in the middle of its delete
// cascade (boxes gone, config still there) is not an orphan.
var orphanTables = []struct{ table, where string }{
{"os_host_settings", `host_id NOT IN (SELECT host_id FROM hosts)`},
{"os_reports", `host_id NOT IN (SELECT host_id FROM hosts)`},
{"os_kernel_notices", `host_id NOT IN (SELECT host_id FROM hosts)`},
{"offsite_keys", `customer_id NOT IN (SELECT customer_id FROM customer_configs) AND customer_id NOT IN (SELECT customer_id FROM hosts)`},
{"offsite_windows", `customer_id NOT IN (SELECT customer_id FROM customer_configs) AND customer_id NOT IN (SELECT customer_id FROM hosts)`},
{"offsite_abandon_requests", `customer_id NOT IN (SELECT customer_id FROM customer_configs) AND customer_id NOT IN (SELECT customer_id FROM hosts)`},
}
// PurgeOrphanRows removes, in ONE transaction, the rows of boxes and households that no longer exist from the six
// tables that had no delete statement (D73). It is the daily clean-up for what was left BEFORE DeleteHost and
// PurgeCustomerResidue learned those tables (Peti's box, the test boxes of 2026-11): from now on a delete leaves
// nothing for it. It never touches the delete journal (`host_deletions`, `customer_resets`), `events`,
// `notification_log` or `operator_actions` (those go by the one-year rule). Idempotent.
// Pinned by TestD73_PurgeOrphanRows_OnlyTheGone.
func (s *Store) PurgeOrphanRows() (OrphanRows, error) {
tx, err := s.db.Begin()
if err != nil {
return nil, err
}
defer tx.Rollback()
out := OrphanRows{}
for _, t := range orphanTables {
res, err := tx.Exec(`DELETE FROM ` + t.table + ` WHERE ` + t.where)
if err != nil {
return nil, fmt.Errorf("PurgeOrphanRows %s: %w", t.table, err)
}
if n, _ := res.RowsAffected(); n > 0 {
out[t.table] = n
}
}
return out, tx.Commit()
}
+11
View File
@@ -3322,6 +3322,11 @@ func (s *Store) DeleteHost(hostID string, deleteEscrow bool) error {
`DELETE FROM log_bundle_requests WHERE scope_id = ?`,
`DELETE FROM log_bundles WHERE scope_id = ?`,
`DELETE FROM wg_peers WHERE host_id = ?`,
// D73 (`09` §3 decision 293, R-1376): the three box-keyed tables that had no delete at all. A deleted box
// kept its update settings, every update report it ever sent, and its kernel notices — for ever.
`DELETE FROM os_host_settings WHERE host_id = ?`,
`DELETE FROM os_reports WHERE host_id = ?`,
`DELETE FROM os_kernel_notices WHERE host_id = ?`,
`DELETE FROM hosts WHERE host_id = ?`,
}
for _, q := range stmts {
@@ -3329,6 +3334,12 @@ func (s *Store) DeleteHost(hostID string, deleteEscrow bool) error {
return fmt.Errorf("DeleteHost %s: %q: %w", hostID, q, err)
}
}
// D73: the hub's own small alarm marks that NAME the box (`os_alarm:<kind>:<host>[:…]` — a clock alarm, a trust
// cause, a guest the watchdog gave up on, a signed job). instr, not LIKE: a host id may hold `_` or `%`.
if _, err := tx.Exec(`DELETE FROM hub_settings WHERE key LIKE 'os_alarm:%' AND
(instr(key, ':' || ?1 || ':') > 0 OR substr(key, -(length(?1) + 1)) = ':' || ?1)`, hostID); err != nil {
return fmt.Errorf("DeleteHost %s: alarm marks: %w", hostID, err)
}
return tx.Commit()
}