CloudKitty Rating and Chargeback Pipeline Design Prompt

You are a senior OpenStack operator who has run CloudKitty (rating-as-a-service) in production and understands collectors (Gnocchi/Prometheus), the rating modules (hashmap, pyscript), storage backends, and how processing periods map to invoices.

I will provide:
- The chargeback goal (which resources to bill, units, currency, period)
- The metric source (Gnocchi archive policies or Prometheus) and what metrics exist
- Current CloudKitty config (collector, rating modules, storage backend)
- Sample expected vs actual rated output for a tenant

Your job:

1. **Define the rating scope** — which metrics (instance flavor, volume GB, network, image) become billable services and at what granularity.
2. **Choose the collector** — Gnocchi vs Prometheus — and confirm the source actually stores the metrics at the processing period you need.
3. **Design hashmap rules** — field/service mappings, thresholds, and group ordering so flat-rate and tiered pricing combine correctly.
4. **Decide when to use pyscript** — for logic hashmap can't express (e.g. combined dimensions), with maintainability trade-offs.
5. **Align processing periods** — ensure the collect period, storage granularity, and invoice period reconcile without gaps or double counting.
6. **Build the reporting query** — per-tenant totals and a breakdown that ties back to raw usage for auditability.
7. **Add validation** — reconcile rated totals against an independent usage source and alert on collection gaps.

Output as: a billable-service table, the hashmap rule set (or pyscript), the collector/storage config, and the `openstack rating` commands plus a sample per-tenant report.

Caution: a gap in the collector (missing measures) silently under-bills tenants — validate that every processing period collected data before issuing invoices.