hub (unreleased): the end dates of the hub's own public names are read once a day; 14 days before, the operator is told (D85, R-1379)
The names are the setting HUB_OWN_NAMES; with no setting the hub makes no call. The setting is NOT in the manifest yet. Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
This commit is contained in:
1 parent
c616727d84
commit
c72ecf52b4
8 files changed
+616
-5
No files matched your search
@@ -822,7 +822,7 @@ another date or new names each hour. **Not proven:** on a box or a deployed hub;
|
||||
|
||||
| End date (block) | Who knows the date today | Could the product read it? | Proposed |
|
||||
|---|---|---|---|
|
||||
| The hub address's certificate, and the download, ISO, Gitea and website names (DH20) | anybody: readable from outside with no login | yes — the hub can open a TLS connection to its own five names once a day | **build next:** the same checker, a third source. Small. Not built tonight: it is a call from the hub to outside names, and the name list needs the operator's yes |
|
||||
| The hub address's certificate, and the download, ISO, Gitea and website names (DH20) | anybody: readable from outside with no login | yes — the hub can open a TLS connection to its own five names once a day | **BUILT, not released (2026-11-15; D85, `09` §3 decision 310): §12.3** |
|
||||
| A node's Tailscale key (DH30) | the Tailscale admin page; one date is in `operations/tailscale.md` | not from a box (the service does not run on two of three) | the operator's list |
|
||||
| The package signing keys (DB16) | measured: the soonest key in use ends 2029-01-15 | yes, by the root wrapper's facts read | no warning now; read again in the yearly end-dates pass |
|
||||
| The 30-day restore copy (DB24) | the box (the copy's own date) | yes | not an alarm: a page fact. The household's restore page shows the date already; nothing for the operator |
|
||||
@@ -835,3 +835,38 @@ another date or new names each hour. **Not proven:** on a box or a deployed hub;
|
||||
**A third source that is small and was left out on purpose:** the hub's own five public names (first row). It needs
|
||||
no secret, and its cost is one outbound connection a day; it waits for a yes because tonight's brief forbids outside
|
||||
calls and the list of names is a promise about what the hub watches.
|
||||
|
||||
### 12.3 The hub's own public names `[FACT — built, not released: hub `main` after v0.182.0, 2026-11-15]` (D85, `09` §3 decision 310, R-1379)
|
||||
|
||||
Once a day (06:40 Budapest, the daily job `own-names-check`) the hub opens one TLS connection to each of its own
|
||||
public names and reads the end date of the certificate that is served (`monitor.OwnNameChecker`). The third source
|
||||
beside an operator token and a box's web certificate; the same 14 days.
|
||||
|
||||
| | |
|
||||
|---|---|
|
||||
| The names | **the setting `HUB_OWN_NAMES`** (the hub's manifest), never code and never a box's word: `host` or `host:port`, at most 16; an entry that is not a plain host name is refused and named in the start log. **No setting → no call.** The operator's list (D85): the hub address, the download address, the ISO address, the Gitea address, the website |
|
||||
| 14 days before the end | `hub_name_cert_ends_soon`, warning, operator-only, once per name; the text names the name, the days and the day |
|
||||
| The date passed and the certificate is still served | the same event once more, as an error |
|
||||
| A renewal | nothing to clear: the renewed certificate has a later date and is told only when that one comes near |
|
||||
| **A name that does not answer** | its OWN line, `hub_name_unreachable` (warning, operator-only, once while it lasts; `hub_name_reachable` clears it). **Never an end-date alarm:** „did not answer” says nothing about the certificate |
|
||||
| The run was skipped (the hub was down at 06:40) | the daily ledger tells it at the next start (`hub_daily_job_skipped`, `05` §16.6) |
|
||||
| Two names the same day | two mails: the cooldown is per name |
|
||||
| The read | no chain check, on purpose: a certificate that has ENDED must still give its date. Nothing is trusted from the connection and nothing is sent over it |
|
||||
|
||||
**For the release (not done tonight — no manifest change, no outside call):** add to `manifests/hub.yaml`, with the
|
||||
image tag, the lines
|
||||
|
||||
```
|
||||
- name: HUB_OWN_NAMES
|
||||
value: "hub.felhom.eu,files.felhom.eu,iso.felhom.eu,gitea.dooplex.hu,felhom.eu"
|
||||
```
|
||||
|
||||
and read the start line „own-names check: 5 name(s) are read daily…”. The five names are CC's reading of the list in
|
||||
the end-dates evidence (`audits/end-dates-list-2026-11-11/`, DH20: the hub, the download, the ISO, Gitea, the
|
||||
website); **the operator confirms the five at the deploy.** From inside the cluster the hub's own address may answer
|
||||
from the LAN side — the certificate is the same one.
|
||||
|
||||
**Not proven:** on a deployed hub; no outside name was read. Tests with a real TLS server on the test machine and with
|
||||
a certificate that has ended (`TestOwnNames_*`, `TestD85_*`); 18 mutations seen red
|
||||
(`audits/build-release-4-2026-11-15/C/redproof-hub-own-names.txt`).
|
||||
|
||||
Reference in new issue
Block a user