coulomb/activity-core
4
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

docs(WP-0015): record railiance01 beachhead deploy evidence
All checks were successful
CI Smoke / host-smoke (push) Successful in 0s
Details
CI Smoke / container-smoke (push) Successful in 2s
Details

Browse source 
Note Forgejo image main-1cf949b live on cluster, relay health/cache warm-up,
and bridge retirement. Overnight triage proof remains the open stabilization check.
This commit is contained in:
tegwick 2026-07-09 01:34:04 +02:00
parent cdbe4de2bf
commit 92a5408156
1 changed files with 11 additions and 13 deletions
Download patch file Download diff file Expand all files Collapse all files

24
workplans/ACTIVITY-WP-0015-adopt-statehub-beachhead-endpoint.md
View file

@ -18,11 +18,9 @@ Carries the **blocked remainder** of [[ACTIVITY-WP-0014]] T05. The in-repo half
(idempotency-keyed State Hub writes) shipped in WP-0014; this workplan adopts the
state-hub-owned **edge relay** (write outbox v1) for the railiance01 runtime.
**Scope note (2026-07-09):** Path B — beachhead v1 shipped with write outbox.
**2026-07-09 update:** state-hub edge relay read cache landed for allowlisted
`GET` routes; activity-core consumes stale cache transparently. Operator deploy
of a state-hub image containing the read-cache code is still required on
railiance01.
**Scope note (2026-07-09):** Path B — beachhead v1 shipped with write outbox and
read cache for allowlisted `GET` routes. Deployed on railiance01 via Forgejo CI
image `forgejo.coulomb.social/coulomb/state-hub:main-1cf949b` (commit `1cf949b`).
## Point STATE_HUB_URL at the beachhead
@ -39,11 +37,11 @@ relay image with upstream `state-hub.state-hub.svc.cluster.local:8000` and a PVC
for the SQLite outbox. activity-core sinks accept HTTP 202 queued receipts as
successful delivery pending replay (`parse_state_hub_write_response`).
**Operator verification still required on railiance01:** apply the manifest with a
state-hub image that includes edge read cache, confirm `GET /edge/health` shows
`upstream_reachable: true` and `read_cache.entry_count > 0` after warm-up,
simulate upstream outage to confirm stale `GET /state/summary` and queued progress
writes, then replay with `statehub outbox replay`.
**Deployed (2026-07-09):** `actcore-statehub-edge-relay` 1/1 Ready;
`GET /edge/health` reports `upstream_reachable: true`; warm-up showed
`read_cache.entry_count: 1` and `X-StateHub-Edge-Cache: hit` on `/state/summary`.
Optional outage drill (stale GET + queued write + outbox replay) and overnight
triage proof remain open stabilization checks.
## Retire the bespoke actcore-state-hub-bridge proxy
@ -58,6 +56,6 @@ Removed `actcore-state-hub-bridge` Deployment/Service from
`k8s/railiance/20-runtime.yaml`. Ops inventory and tests now reference
`actcore-statehub-edge-relay`.
**Operator verification still required:** one overnight `daily-statehub-wsjf-triage`
run after deploy (original failure condition). If upstream was briefly down,
check the relay outbox and replay before closing the stabilization window.
**Deployed (2026-07-09):** legacy `actcore-state-hub-bridge` Deployment/Service
deleted on railiance01. Overnight `daily-statehub-wsjf-triage` proof (original
failure condition) is the remaining stabilization check.