chore(w8): refresh module upgrade readiness inventory #501

New issue

Closed

opened 2026-05-26 08:51:34 +02:00 by codex · 1 comment

codex commented

2026-05-26 08:51:34 +02:00

Collaborator

Spec sources (whitelist)

state/cycle/W8-module-upgrade-prep.md §Current Readiness Snapshot / Wave Plan / Issue Handling
state/roadmap/current-platform-roadmap.md §Operator wave map / 09 - Module upgrade waves
decisions/0022-module-source-and-release-boundaries.md §Metadata direction / Update workflow / Pilot rollout
modules/minio-init/module.yaml

Extracted context

From state/cycle/W8-module-upgrade-prep.md:

W8.0 readiness | Fill enough metadata and classify risk before real upgrades.
Repo-only, no runtime mutation.
Snapshot on 2026-05-25:
Total modules 88; modules with spec.source 6; modules with spec.artifact 6;
missing source metadata 82; missing artifact metadata 82; propose-pr candidates 5; auto-pr candidates 0.
Child issues should be opened only when a candidate has enough metadata to be ready for an agent.

From state/roadmap/current-platform-roadmap.md:

W8 | M09 | Module upgrade waves using ADR-0022 metadata and rollback evidence.
Gate: M02 restore confidence and M05 metadata foundation are accepted.

From decisions/0022-module-source-and-release-boundaries.md:

Future schema work should add explicit source and artifact metadata to modules/<id>/module.yaml.
Desired artifact metadata answers "what should production run?"
Observed metadata answers "what did production actually run when audited?"
Only after the report is useful, add automation that opens update PRs.

Do NOT read (unless escape hatch fires)

Runtime host files under /opt/pdurlej-platform — this is repo-only readiness.
Forgejo upgrade plan details — Forgejo is W8.final, not W8.0.
Module source repos — this issue classifies platform metadata, it does not upgrade sources.

Allowed touched paths

state/cycle/W8-module-upgrade-prep.md (modify)
state/cycle/W8-readiness-*.md (create)
state/roadmap/current-platform-roadmap.md (modify only if the current W8 status needs a one-line update)

Why this exists (product-first)

W8 should not become "upgrade everything because there are newer tags." The operator needs a small, visible readiness packet that says which modules are safe candidates, which are blocked, and what evidence is missing before agents start upgrade PRs.

Why this matters now

Patchwarden and W6d reduced review friction, so the next bottleneck is safe upgrade selection. Without W8.0, agents will either avoid upgrades or pick risky targets without enough rollback evidence.

What "done" looks like

A fresh repo-only update/readiness report is run or reproduced from platform metadata.
The output records source/artifact coverage, propose-pr candidates, auto-pr candidates, and missing metadata counts.
Candidate modules are classified by risk class: stateless, stateful, sidecar-bound, public-edge, auth/security-sensitive, core/workbench.
minio-init is explicitly accepted or rejected as the W8.1 first pilot based on metadata, not vibes.
The artifact states that no runtime mutation, deploy, recreate, backup, or service restart was performed.
PR links back to #389.

Scope

In scope:

Read platform repo metadata.
Produce/update W8 readiness artifact.
Identify exact blockers before W8.1 upgrade work.

Out of scope:

Upgrading any image/tag/digest.
Running platformctl apply or host-agent commands.
Touching production compose/env/runtime.
Forgejo upgrade execution.

Suggested approach

Start from PYTHONPATH=control-plane python3 -m platformctl.cli update-report --json if available locally; otherwise derive the same fields from module manifests and say why.
Compare current numbers against the 2026-05-25 snapshot in state/cycle/W8-module-upgrade-prep.md.
Write a compact readiness artifact and update #389 with the result in the PR body or follow-up comment.

Escape hatch

If the report tooling is broken, do not fake the numbers. Open a blocker comment on #389 with the failing command, exit code, and whether the issue should become tooling repair first.

Unknowns / owner questions

None for repo-only readiness. Any production mutation belongs to a later operator-gated issue.

Risk class

risk/process — workflow/review/orchestration may produce bad outcomes

Class of service

class/security-sensitive

Trace

Original source: #389, Oracle review platform-pr488-wave-status, and operator approval on 2026-05-26.
Migrated by: codex, 2026-05-26.
Related issues / PRs: #389, #500.

## Spec sources (whitelist) - `state/cycle/W8-module-upgrade-prep.md` §Current Readiness Snapshot / Wave Plan / Issue Handling - `state/roadmap/current-platform-roadmap.md` §Operator wave map / 09 - Module upgrade waves - `decisions/0022-module-source-and-release-boundaries.md` §Metadata direction / Update workflow / Pilot rollout - `modules/minio-init/module.yaml` ## Extracted context > From `state/cycle/W8-module-upgrade-prep.md`: > ``` > W8.0 readiness | Fill enough metadata and classify risk before real upgrades. > Repo-only, no runtime mutation. > Snapshot on 2026-05-25: > Total modules 88; modules with spec.source 6; modules with spec.artifact 6; > missing source metadata 82; missing artifact metadata 82; propose-pr candidates 5; auto-pr candidates 0. > Child issues should be opened only when a candidate has enough metadata to be ready for an agent. > ``` > From `state/roadmap/current-platform-roadmap.md`: > ``` > W8 | M09 | Module upgrade waves using ADR-0022 metadata and rollback evidence. > Gate: M02 restore confidence and M05 metadata foundation are accepted. > ``` > From `decisions/0022-module-source-and-release-boundaries.md`: > ``` > Future schema work should add explicit source and artifact metadata to modules/<id>/module.yaml. > Desired artifact metadata answers "what should production run?" > Observed metadata answers "what did production actually run when audited?" > Only after the report is useful, add automation that opens update PRs. > ``` ## Do NOT read (unless escape hatch fires) - Runtime host files under `/opt/pdurlej-platform` — this is repo-only readiness. - Forgejo upgrade plan details — Forgejo is W8.final, not W8.0. - Module source repos — this issue classifies platform metadata, it does not upgrade sources. ## Allowed touched paths - `state/cycle/W8-module-upgrade-prep.md` (modify) - `state/cycle/W8-readiness-*.md` (create) - `state/roadmap/current-platform-roadmap.md` (modify only if the current W8 status needs a one-line update) ## Why this exists (product-first) W8 should not become "upgrade everything because there are newer tags." The operator needs a small, visible readiness packet that says which modules are safe candidates, which are blocked, and what evidence is missing before agents start upgrade PRs. ## Why this matters now Patchwarden and W6d reduced review friction, so the next bottleneck is safe upgrade selection. Without W8.0, agents will either avoid upgrades or pick risky targets without enough rollback evidence. ## What "done" looks like - [ ] A fresh repo-only update/readiness report is run or reproduced from platform metadata. - [ ] The output records source/artifact coverage, `propose-pr` candidates, `auto-pr` candidates, and missing metadata counts. - [ ] Candidate modules are classified by risk class: stateless, stateful, sidecar-bound, public-edge, auth/security-sensitive, core/workbench. - [ ] `minio-init` is explicitly accepted or rejected as the W8.1 first pilot based on metadata, not vibes. - [ ] The artifact states that no runtime mutation, deploy, recreate, backup, or service restart was performed. - [ ] PR links back to #389. ## Scope **In scope:** - Read platform repo metadata. - Produce/update W8 readiness artifact. - Identify exact blockers before W8.1 upgrade work. **Out of scope:** - Upgrading any image/tag/digest. - Running `platformctl apply` or host-agent commands. - Touching production compose/env/runtime. - Forgejo upgrade execution. ## Suggested approach 1. Start from `PYTHONPATH=control-plane python3 -m platformctl.cli update-report --json` if available locally; otherwise derive the same fields from module manifests and say why. 2. Compare current numbers against the 2026-05-25 snapshot in `state/cycle/W8-module-upgrade-prep.md`. 3. Write a compact readiness artifact and update #389 with the result in the PR body or follow-up comment. ## Escape hatch If the report tooling is broken, do not fake the numbers. Open a blocker comment on #389 with the failing command, exit code, and whether the issue should become tooling repair first. ## Unknowns / owner questions - None for repo-only readiness. Any production mutation belongs to a later operator-gated issue. ## Risk class - [x] `risk/process` — workflow/review/orchestration may produce bad outcomes ## Class of service - [ ] `class/security-sensitive` ## Trace - Original source: #389, Oracle review `platform-pr488-wave-status`, and operator approval on 2026-05-26. - Migrated by: codex, 2026-05-26. - Related issues / PRs: #389, #500.

codex added this to the 09 - Module upgrade waves milestone

2026-05-26 08:51:34 +02:00

codex added the

tier/full

ready-for-agent

risk/process

labels

2026-05-26 08:51:34 +02:00

codex referenced this issue

2026-05-26 08:52:26 +02:00

chore(minio-init): prepare W8.1 pilot upgrade packet #502

codex referenced this issue