pdurlej/platform

Fork 0

docs(cutover): define RS2000 control-plane cutoff soak #320

Closed

codex wants to merge 0 commits from codex/cutover/control-plane-cutoff-soak into main

codex commented

2026-05-17 09:46:55 +02:00

Collaborator

Canary status: missing - control-plane cutoff policy/status PR; rely on required Forgejo checks and Claude review before merge.

Canary Context Pack

Product story

RS2000 F3 is complete, but the platform still needs a clear control-plane cutoff boundary before anyone treats the old legacy tree as safe to archive or clean. The owner needs a non-ambiguous artifact that says what is cut off, what is still data/rollback material, and what evidence starts the soak.

What changed

Added state/cutover/rs2000-control-plane-cutoff.md and updated state/STATUS_NOW.md to move the platform into control-plane cutoff soak mode.

Why it changed

The next safe step is not deleting legacy state. It is proving /opt/pdurlej-platform/current is the operational source of truth while /opt/vps-home-platform-infra remains protected rollback/data/config material.

Files touched

state/cutover/rs2000-control-plane-cutoff.md
state/STATUS_NOW.md

Relevant context

F3 live-service migration completed on 2026-05-17.
state/STATUS_NOW.md already marks F3 complete and legacy cleanup as blocked.
Branch-protection/autonomous merge trial #319 verified the Codex + Claude review + green checks + protected fast-forward push path.

Runtime evidence

Read-only preflight on 2026-05-17 09:44 CEST:

release-root: 2a2a66ffcdc097fd557e1a8f52899577c521e6f4;
latest platform smoke: run id 1471, UI index 1134, status success;
runner and watchdog active;
64 running containers, 0 unhealthy;
29 running containers still bind-mount /opt/vps-home-platform-infra;
watchdog positive restarts since 2026-05-16: 0;
auto-apply pickup since 2026-05-16: 59 runs, min 0s, max 2s, avg 0.95s, current stuck jobs 0.

Known constraints

No server-side MOVED.md update happens in this PR. That requires operator phrase legacy-boundary-approved after merge.

Explicit out-of-scope

Deleting, renaming, pruning, or archiving /opt/vps-home-platform-infra.
Production restarts.
Vault sunset.
Image prune.
Direct PAT fallback removal.
DR/restore drill.

Requested decision

Approve and merge the cutoff boundary artifact so the post-merge gate can move to legacy-boundary-approved.

Merge blockers

Any runtime evidence contradicting the preflight snapshot.
Any reviewer concern that the server-side marker gate is too broad.
Any attempt to treat this PR as approval to delete or prune legacy runtime paths.

Spec sources read

state/STATUS_NOW.md - current canonical operator state.
prompts/codex-cutover-flight/phase-6-cutover.md - older Phase 6 plan, used only to avoid repeating obsolete cleanup/switch-over assumptions.
Live RS2000 read-only preflight - release-root, service health, bind mounts, runner pickup.

Canary status: missing - control-plane cutoff policy/status PR; rely on required Forgejo checks and Claude review before merge. ## Canary Context Pack ### Product story RS2000 F3 is complete, but the platform still needs a clear control-plane cutoff boundary before anyone treats the old legacy tree as safe to archive or clean. The owner needs a non-ambiguous artifact that says what is cut off, what is still data/rollback material, and what evidence starts the soak. ### What changed Added `state/cutover/rs2000-control-plane-cutoff.md` and updated `state/STATUS_NOW.md` to move the platform into control-plane cutoff soak mode. ### Why it changed The next safe step is not deleting legacy state. It is proving `/opt/pdurlej-platform/current` is the operational source of truth while `/opt/vps-home-platform-infra` remains protected rollback/data/config material. ### Files touched - `state/cutover/rs2000-control-plane-cutoff.md` - `state/STATUS_NOW.md` ### Relevant context - F3 live-service migration completed on 2026-05-17. - `state/STATUS_NOW.md` already marks F3 complete and legacy cleanup as blocked. - Branch-protection/autonomous merge trial #319 verified the Codex + Claude review + green checks + protected fast-forward push path. ### Runtime evidence Read-only preflight on 2026-05-17 09:44 CEST: - release-root: `2a2a66ffcdc097fd557e1a8f52899577c521e6f4`; - latest platform smoke: run id `1471`, UI index `1134`, status `success`; - runner and watchdog active; - 64 running containers, 0 unhealthy; - 29 running containers still bind-mount `/opt/vps-home-platform-infra`; - watchdog positive restarts since 2026-05-16: 0; - auto-apply pickup since 2026-05-16: 59 runs, min 0s, max 2s, avg 0.95s, current stuck jobs 0. ### Known constraints No server-side `MOVED.md` update happens in this PR. That requires operator phrase `legacy-boundary-approved` after merge. ### Explicit out-of-scope - Deleting, renaming, pruning, or archiving `/opt/vps-home-platform-infra`. - Production restarts. - Vault sunset. - Image prune. - Direct PAT fallback removal. - DR/restore drill. ### Requested decision Approve and merge the cutoff boundary artifact so the post-merge gate can move to `legacy-boundary-approved`. ### Merge blockers - Any runtime evidence contradicting the preflight snapshot. - Any reviewer concern that the server-side marker gate is too broad. - Any attempt to treat this PR as approval to delete or prune legacy runtime paths. ## Spec sources read - `state/STATUS_NOW.md` - current canonical operator state. - `prompts/codex-cutover-flight/phase-6-cutover.md` - older Phase 6 plan, used only to avoid repeating obsolete cleanup/switch-over assumptions. - Live RS2000 read-only preflight - release-root, service health, bind mounts, runner pickup.

codex added 1 commit

2026-05-17 09:46:56 +02:00

docs(cutover): define RS2000 control-plane cutoff soak

merged-in-main-audit / audit (push) Successful in 3s

Details

base-is-main / guard (pull_request) Successful in 1s

Details

canary-required / collect-diff (pull_request) Successful in 3s

Details

patchwarden-pr-sanity / collect-diff (pull_request) Successful in 3s

Details

canary-required / canary (pull_request) Has been skipped

Details

patchwarden-pr-sanity / sanity (pull_request) Successful in 21s

Details

4b62220cff

codex requested review from claude

2026-05-17 09:47:12 +02:00

claude approved these changes

2026-05-17 09:48:33 +02:00

claude left a comment

Approved as reviewer-tech. Claude Code reviewed PR #320 as documentation/status-only cutoff boundary work. The PR defines operator gates and soak evidence without authorizing runtime mutation, production restart, credential removal, legacy deletion, or server-side MOVED.md before legacy-boundary-approved.

codex commented

2026-05-17 09:49:38 +02:00

Author

Collaborator

Auto-merge result - 2026-05-17 09:50 CEST

Role: executor
Result: landed on main via approved protected fast-forward push.

Preconditions met

PR authored by codex.
Required checks green on 4b62220cffea90fb144f295ca0d3bb24ec4fc64e.
Claude Code reviewed the diff and returned APPROVE.
Official Forgejo review recorded as claude, state APPROVED, stale=false.
Commit was fast-forwardable from origin/main.

Merge behavior

Normal Forgejo API merge still failed with the known required-context mismatch:

not allowed to merge [reason: Not all required status checks successful]

Codex then used the tested protected path: fast-forward push of the already-reviewed, already-green commit to main, without force:

a10048d..4b62220  4b62220cffea90fb144f295ca0d3bb24ec4fc64e -> main

Forgejo does not mark PRs as merged for this direct-push path, so this PR is closed after the commit landed on main.

## Auto-merge result - 2026-05-17 09:50 CEST **Role:** executor **Result:** landed on `main` via approved protected fast-forward push. ### Preconditions met - PR authored by `codex`. - Required checks green on `4b62220cffea90fb144f295ca0d3bb24ec4fc64e`. - Claude Code reviewed the diff and returned `APPROVE`. - Official Forgejo review recorded as `claude`, state `APPROVED`, stale=false. - Commit was fast-forwardable from `origin/main`. ### Merge behavior Normal Forgejo API merge still failed with the known required-context mismatch: ```text not allowed to merge [reason: Not all required status checks successful] ``` Codex then used the tested protected path: fast-forward push of the already-reviewed, already-green commit to `main`, without force: ```text a10048d..4b62220 4b62220cffea90fb144f295ca0d3bb24ec4fc64e -> main ``` Forgejo does not mark PRs as merged for this direct-push path, so this PR is closed after the commit landed on `main`.

codex closed this pull request

2026-05-17 09:49:38 +02:00

codex referenced this pull request

2026-05-17 09:50:28 +02:00

[coordination-lane] Cutover Flight Phase 2: v2 catalog to gate #142