railiance-apps/workplans/RAPPS-WP-0018-wave2-resource-right-sizing.md

---
id: RAPPS-WP-0018
type: workplan
title: "Wave-2 resource right-sizing from the 2026-10-21 allocation review"
domain: financials
repo: railiance-apps
status: finished
flavor: implementation
owner: claude
topic_slug: railiance
created: "2026-10-21"
updated: "2026-10-21"
related:
  - CUST-WP-0071
  - RESOURCE-WP-0008
state_hub_workstream_id: "c3f7c479-7477-5d5e-865d-a5d5b9e79ab8"
---

# RAPPS-WP-0018 — Wave-2 resource right-sizing from the 2026-10-21 allocation review

Receiving workplan in `railiance-apps` for the Vergabe memory request (replay-backed) and the latent Inter-Hub request.

Origin: the 2026-10-21 node allocation review (`RESOURCE-WP-0008-T03`, feeding `CUST-WP-0071-T03`). Per the founder's 2026-10-21 delivery interview (`the-custodian/docs/decisions/2026-10-21-cust-wp-0071-delivery-interview.md`) the small adjustments form wave 2: one receiving task per owner repo, applied later in a batch. Applying wave 2 needs a separate founder approval, so every cluster-write task here is a human-gated `wait`. This file changes no manifest.

## Raise the Vergabe memory request and clear the latent inter-hub request

```task
id: RAPPS-WP-0018-T01
status: done
priority: medium
state_hub_task_id: "48a3c858-c65c-575c-8a97-4122245a1eca"
```

Values are pod-aggregate figures from the budget file (7-day window ending 2026-10-21T06:08Z). Limits stay unchanged unless stated.

| Workload (ns/name) | Pods | CPU request | CPU limit | CPU p95/peak 7d (m) | Memory request | Memory limit | Mem p95/peak (Mi) | Throttle | Action |
|---|---:|---:|---:|---:|---:|---:|---:|---:|---|
| `vergabe-demo-company/vergabe-teilnahme` | 1 | 60m → **60m** | 1000m | 0.5/20.3 | 256Mi → **512Mi** | 1024Mi | 171.5/171.6 | 0.0 | keep CPU (replay-backed) + memory up |

Source and container detail:

- `vergabe-demo-company/vergabe-teilnahme`: Deployment `vergabe-teilnahme` (container `vergabe-teilnahme`), `resources` in `helm/vergabe-demo-company-values.proposed.yaml` (60m/256Mi, limits 1000m/1Gi); the chart default in `charts/vergabe-teilnahme/values.yaml` (100m/256Mi) is not what is deployed. Keep the 60m CPU request and the 1000m/1Gi limits (replay-backed). Raise memory 256Mi → 512Mi: the target replay (`docs/evidence/2026-10-21-vergabe-target-replay/`, commit `1af6d82`, RAPPS-WP-0014-T04) showed a loaded working set of 320–480MiB.

**Prepared 2026-10-21 (wave-2 worker W1a; not applied).** Change: railiance-apps `b7fb7b2`.
- Vergabe: live Helm release `vergabe-teilnahme` in `vergabe-demo-company` (rev 3) matches `helm/vergabe-demo-company-values.proposed.yaml`. Memory request 256Mi → 512Mi for the app and the `migrate` init container, which share `resources`. CPU stays 60m and limits stay 1000m/1Gi. The full render against live adds **drift beyond resources**: the label `railiance.io/release-unit: vergabe` (from commit 64530e1, 2026-10-21) is missing on Service `vergabe-teilnahme` and both PVCs; the Deployment already has it. Because of this the apply comes to the founder. Apply: `helm upgrade vergabe-teilnahme charts/vergabe-teilnahme -n vergabe-demo-company -f helm/vergabe-demo-company-values.proposed.yaml --wait --timeout 5m`. Strategy is Recreate (short outage).
- Inter-Hub: the owner chose removal. Deleted `charts/inter-hub`, `helm/inter-hub-values.yaml`, the `inter-hub-*` Make targets, `tools/inter-hub-smoke.sh` and `.gitea/workflows/inter-hub-production-deploy.yaml`; the inter-hub part of `tools/k8s-server-dry-run.sh` is removed and the docs updated. Live: Deployment (0 replicas, 250m/512Mi), Service and 8 old ReplicaSets, all from Helm release `inter-hub` (rev 2); it has no Ingress. Apply: `helm uninstall inter-hub -n inter-hub`. Afterwards the founder checks for leftover credential objects, which agents do not inspect, before deleting the namespace. A stale rollback artifact `core-hub/k8s/railiance-production/rollback-inter-hub-ingress.yaml` (archived CORE-WP-0005) still points at the release.

Evidence: resource-control `docs/evidence/2026-10-21-node-allocation-review/2026-10-21-node-budget.md` (+ `.json`, commit `3d9b1bd`). Rule: CPU = max(floor, p95x1.5, peakx0.5) rounded up to 5m; memory request = 7d p95 working set x1.15 (only when the change is at least 64Mi).

Steps:

1. Map the pod-aggregate proposal onto the named container(s) in the source file; where a workload has several containers or pods, split by the per-pod figures in `2026-10-21-workload-usage.json` and keep the sum at the proposed value.
2. Prepare the source change as a reviewed commit (no push, no sync) and record the before values here.
3. Apply only inside the batched wave-2 rollout, after founder approval, through the owner's normal deploy path (ArgoCD sync or documented helm/kubectl apply).

Verification: `kubectl -n <ns> get pod -o jsonpath='{..resources}'` shows the new requests; pods Ready with no new restarts or OOMKilled over 24h; for CPU decreases the throttle ratio and request latency stay at or below the 7d baseline; node request total drops by the expected delta.

Rollback: revert the source commit and re-sync/re-apply; the before values in the table above are the rollback target. Keep before/after `kubectl get -o yaml` captures as the receipt.

Inter-Hub: `charts/inter-hub/values.yaml` still defaults to `requests: {cpu: 250m, memory: 512Mi}` (limits 1000m/2Gi), and `helm/inter-hub-values.yaml` does not override it. The release runs at 0 replicas, so the 250m is latent: it is reserved again the moment anything scales it up or the deploy workflow (`.gitea/workflows/inter-hub-production-deploy.yaml`) runs. Inter-Hub is retired (CORE-WP-0007; production `hub.coulomb.social` is Core Hub only). Preferred: remove the inter-hub release and retire its chart and deploy workflow; otherwise lower the default request to the 5m floor so a scale-up cannot reserve 250m. Record which option the owner chose.

## Applied — 2026-10-22 (RAPPS-WP-0018-T01)

Applied under the founder's wave-2 authorization (the-custodian docs/decisions/2026-10-21-cust-wp-0071-delivery-interview.md), after a full-render diff against live that showed resources only, plus the founder-approved `railiance.io/release-unit` label. Vergabe memory 512Mi, Helm revision 4. The retired inter-hub release was uninstalled; namespace `inter-hub` is kept until the founder checks what is left in it. Evidence: the-custodian docs/evidence/2026-10-22-wave2-apply.md.