129 lines
4.2 KiB
Markdown
129 lines
4.2 KiB
Markdown
---
|
|
id: REEF-RAILIANCE-WP-0002
|
|
type: workplan
|
|
title: "Admit rail-knative and rapp-qonto with production evidence"
|
|
domain: financials
|
|
repo: reef-railiance
|
|
status: finished
|
|
owner: codex
|
|
topic_slug: railiance
|
|
created: "2026-07-26"
|
|
updated: "2026-07-26"
|
|
state_hub_workstream_id: "5b31c13e-ddc9-495f-8325-88fa770982c5"
|
|
---
|
|
|
|
# REEF-RAILIANCE-WP-0002 - Knative and Qonto production admission
|
|
|
|
## Goal
|
|
|
|
Host a derived `rail-knative` slice while keeping `rail-kubernetes` primary,
|
|
and bind `rapp-qonto` only through explicit readiness evidence.
|
|
|
|
## T01 - Publish the admission schema and baseline
|
|
|
|
```task
|
|
id: REEF-RAILIANCE-WP-0002-T01
|
|
status: done
|
|
priority: high
|
|
state_hub_task_id: "a20cc59f-bb93-4d1b-bf46-0d67c2270b98"
|
|
```
|
|
|
|
Add declared, installed, verified, production-approved, and deprecated states
|
|
plus source-linked evidence requirements.
|
|
|
|
2026-07-26: Added the admission schema and a baseline record for the existing
|
|
Kubernetes rail. The baseline intentionally records unresolved capacity,
|
|
recovery, and failure-domain evidence rather than inferring it from topology.
|
|
|
|
## T02 - Collect the Knative substrate baseline
|
|
|
|
```task
|
|
id: REEF-RAILIANCE-WP-0002-T02
|
|
status: done
|
|
priority: high
|
|
state_hub_task_id: "833f6246-23bc-43ac-802d-7595b18f77a2"
|
|
```
|
|
|
|
Collect Kubernetes version, capacity/headroom, ingress, DNS/TLS, network,
|
|
identity/secret delivery, observability, backup/restore, and maintenance
|
|
evidence before installation.
|
|
|
|
2026-07-26: The first read-only probe reached the configured context but the
|
|
API endpoint timed out. Recorded machine-readable failure evidence and kept
|
|
installation blocked. `railiance-cluster` now owns a reusable preflight probe
|
|
and the access/install workplan.
|
|
|
|
2026-07-26: Server-side preflight passed after the upgrade. Kubernetes is
|
|
v1.35.1 and the Ready node has 4 CPU/16 GiB. The below-recommendation CPU count
|
|
and single-node topology remain explicit production risks.
|
|
|
|
## T03 - Install and verify the derived rail
|
|
|
|
```task
|
|
id: REEF-RAILIANCE-WP-0002-T03
|
|
status: done
|
|
priority: high
|
|
state_hub_task_id: "ff070df6-8ffa-40e6-bbaa-2dcff20f64f2"
|
|
```
|
|
|
|
Install only after the baseline matches the derived rail prerequisites. Record
|
|
versioned smoke and rollback evidence.
|
|
|
|
Blocked by failed API reachability preflight; no install was attempted.
|
|
|
|
2026-07-26: Installed and verified checksum-locked Serving and Kourier
|
|
v1.22.0. Scale-to-zero, cold activation, revision creation, private routing,
|
|
and previous-revision rollback passed. Public exposure remains disabled.
|
|
|
|
## T04 - Bind and verify rapp-qonto
|
|
|
|
```task
|
|
id: REEF-RAILIANCE-WP-0002-T04
|
|
status: cancel
|
|
priority: high
|
|
state_hub_task_id: "485e757b-6519-4f2e-a678-d66ef14206f5"
|
|
```
|
|
|
|
Add the binding at `declared`, then promote it only from machine-readable
|
|
identity, egress, cold-start, audit, revocation, rollback, and dependency
|
|
failure evidence.
|
|
|
|
2026-07-26: Added topology-only declared bindings for `rail-knative` and
|
|
`rapp-qonto`. Baseline evidence keeps unmeasured runtime checks `unknown`; no
|
|
installed, verified, or production-approved state is claimed.
|
|
|
|
2026-07-26: The declared binding is retained without false promotion. Live
|
|
Qonto verification moved to `REEF-RAILIANCE-WP-0003` because identity,
|
|
restricted egress, and workload evidence are not yet established.
|
|
|
|
## T05 - Decide production residual risk
|
|
|
|
```task
|
|
id: REEF-RAILIANCE-WP-0002-T05
|
|
status: done
|
|
priority: high
|
|
state_hub_task_id: "f1ac627d-1ec8-445b-acd9-a65eeb068d3c"
|
|
```
|
|
|
|
Reserve human interaction for explicit acceptance or mitigation of the
|
|
single-server/shared-control-plane failure domain. Automated evidence must be
|
|
complete first.
|
|
|
|
Decision: do not accept the current failure domain for production. The
|
|
verified rail may operate privately for non-production verification. Reassess
|
|
only after workload gates pass or the failure domain is mitigated.
|
|
|
|
## T06 - Automate evidence freshness and drift
|
|
|
|
```task
|
|
id: REEF-RAILIANCE-WP-0002-T06
|
|
status: done
|
|
priority: medium
|
|
state_hub_task_id: "82685f27-575e-4956-8c46-433a2ab2b9ea"
|
|
```
|
|
|
|
Make probes idempotent and suitable for scheduled reconciliation. Expired or
|
|
failed evidence must prevent false production-ready state.
|
|
|
|
2026-07-26: Added a deterministic freshness gate that fails stale evidence and
|
|
rejects production approval with any non-passing check or open residual risk.
|