wiki: sync 2026-08-11 20:34 UTC — 5 file(s) updated

This commit is contained in:
2026-08-11 20:34:53 +00:00
parent 26c221085d
commit 6c8ffdf1b0
5 changed files with 68 additions and 11 deletions
+3 -1
View File
@@ -1,6 +1,6 @@
# 🧠 Knowledge Wiki — Brain Summary
_Last updated: 2026-08-11 04:15 UTC_
_Last updated: 2026-08-11 20:33 UTC_
## Workstreams & Status
@@ -55,6 +55,8 @@ _Last updated: 2026-08-11 04:15 UTC_
- Gitea: gitea.lego-cloud.eu (org: home-v1) ✅
- **General Gitea repository skill:** ✅ `gitea-repository-operations` — reusable API + SSH workflow for any repo; full wiki runbook at `wiki/infrastructure/gitea-repository-operations.md`
- **Gitea API credential (confirmed 2026-08-02):** use Bitwarden-provided `HL_V1_GITEA_ACCESS_TOKEN` for `gitea.lego-cloud.eu`; never use `GITHUB_TOKEN` for that instance or copy the Gitea token into `/opt/data/.env`. Bitwarden keys are exported without aliases and gateway restart is required after key changes.
- **Gitea recovery (confirmed 2026-08-11):** Gitea `1.27.0` crashed all SSH/Smart-HTTP Git-serving processes despite healthy web/API; upgrade to `1.27.1` restored API and both Git transports. The queued VSSA commit `6fc42ab` was pushed and remotely verified.
- **Gitea Pages pattern (verified 2026-08-11):** Actions runners build repositories and publish to `/gondor-v1-gitea-pages/<organization>/<repository>/`, exposed at `https://pages.apps.lego-cloud.eu/<organization>/<repository>/`. The Pages dataset must be allowed by `container.valid_volumes` and mounted into job containers. `skic-v1-playground/documentation` run 2326 and the live site were verified.
- **Gondor v1 (2026-08-02):** Hermes SSH access as `lego` is verified to Osgiliath 000 and both Minas Tirith workers; use the `gondor-v1-nodes` skill. Open ingress incident: Argo CD's hostPort 80/443 `ingress-nginx` conflicts with Osgiliath's system Nginx; repair through `home-v1/gondor-v1.git`, not a transient live patch.
- **Hermes outage (2026-08-04–11):** confirmed Codex quota exhaustion first blocked model responses while Discord/cron remained alive; a separate likely host/container-runtime failure then stopped all gateway, scheduler, log, and health progress after 2026-08-04 14:00:42 UTC. Physical restart at 2026-08-11 00:11:40 restored service without Hermes repair. Exact host-level trigger remains **UNKNOWN** because previous-boot TrueNAS/kernel/ZFS/runtime/hardware telemetry is unavailable inside the container.
- **world-v1/network-v1:** ✅ global Internet address book; IP-level mapping now separates registry holder, route origin, operator, and service. Initial records: `0.0.0.0` and `1.1.1.1`; wiki page at `wiki/infrastructure/network-v1.md`