docs(ORCH-048): prod deploy log — SUCCESS (bind-mount-only, prod untouched) #48
42
docs/work-items/ORCH-048/14-deploy-log.md
Normal file
42
docs/work-items/ORCH-048/14-deploy-log.md
Normal file
@@ -0,0 +1,42 @@
|
||||
---
|
||||
deploy_status: SUCCESS
|
||||
timestamp: 2026-06-06T07:12:00Z
|
||||
work_item: ORCH-048
|
||||
target: orchestrator (prod, self-hosting)
|
||||
restart_required: false
|
||||
rebuild_required: false
|
||||
staging_gate: SUCCESS
|
||||
---
|
||||
|
||||
# Production Deploy Log — ORCH-048
|
||||
|
||||
Production rollout of ORCH-048 (`fix(staging_check): B6 reads registry from running staging instance env`) is **complete on merge** — no prod container rebuild or restart was required, and none was performed.
|
||||
|
||||
## Precondition: staging gate
|
||||
|
||||
`deploy` is reachable only because the staging gate (`deploy-staging`) passed:
|
||||
|
||||
- `15-staging-log.md` → `staging_status: SUCCESS`, **10/10 checks PASS** (incl. B6 registry-isolation, the fix's target), run inside `orchestrator-staging` (8501).
|
||||
|
||||
## Change scope (why no restart)
|
||||
|
||||
The work item changes touch **only bind-mounted / non-runtime artifacts** — `src/` (the code running in the prod container) and the `Dockerfile` are untouched:
|
||||
|
||||
| File | Kind | How it reaches prod |
|
||||
|------|------|---------------------|
|
||||
| `scripts/staging_check.py` | bind-mounted (`/home/slin/repos` → `/repos`); **not** copied into the image (`scripts/` absent from `/app`) | host `git pull` → live immediately |
|
||||
| `.openclaw/agents/deployer.md` | bind-mounted agent prompt, read at agent launch | host `git pull` → live on next agent run |
|
||||
| `CHANGELOG.md`, `docs/operations/STAGING_CHECK.md` | docs | n/a |
|
||||
| `tests/test_staging_check_b6.py` | test, not deployed | n/a |
|
||||
|
||||
Because nothing in `src/` or the image changed, there is **no container rebuild and no restart** for the shared prod `orchestrator` (8500). Per CLAUDE.md / INFRA.md self-hosting rules, the prod container that serves all projects (enduro-trails + orchestrator) was **not** touched — zero group-risk.
|
||||
|
||||
## Deploy action
|
||||
|
||||
- **Prod container restart/rebuild:** not required, not performed (guardrail: never restart prod `orchestrator` within an ORCH task).
|
||||
- **Real docker/SSH deploy hook** (`scripts/orchestrator-deploy-hook.sh`): not triggered by this agent (not explicitly instructed; reserved for Owner per ORCH-36).
|
||||
- **Effective rollout:** merge of this branch to `main` + routine host `git pull` makes the corrected `staging_check.py` and `deployer.md` live; the prod app process is unaffected.
|
||||
|
||||
## Verdict
|
||||
|
||||
`deploy_status: SUCCESS` — staging gate green, change is bind-mount-only, prod instance untouched, no rollback needed.
|
||||
Reference in New Issue
Block a user