feat(obs): Prometheus metrics on /metrics + golden-signal Grafana dashboard (closes #124) (#129)

## What & why

S-16c, the last of the S-16 (#17) split, on top of the backplane (#122) and distributed tracing (#123). The five .NET services now expose OpenTelemetry **metrics** in Prometheus format at `/metrics`; Prometheus scrapes each (one job per service); and Grafana ships a pre-built **Request path — golden signals** dashboard (traffic / errors / latency / saturation), split by service.

Closes #124

### How

- Each service adds `.WithMetrics(AddAspNetCoreInstrumentation + AddHttpClientInstrumentation + AddMeter("System.Runtime") + AddPrometheusExporter)` and maps `/metrics`. Same shape as the S-16b tracing wiring already in these `Program.cs` files.
- `infra/observability/prometheus/prometheus.yml`: one scrape job per service (`acl`, `domain`, `bff`, `event-subscriber`, `projection-api`), reached by compose service name.
- `infra/observability/grafana/provisioning/dashboards/`: dashboard provider + `golden-signals.json` (baked into the Grafana image by the existing `COPY provisioning/`).
- `verify-metrics` (new CI verify-stack step + Makefile target): generates BFF traffic and asserts Prometheus scraped the golden-signal metric from every service. Mirrors `verify-tracing`.

### Dependency (CLAUDE.md §13/§14)

Adds `OpenTelemetry.Exporter.Prometheus.AspNetCore` `1.17.0-beta.1` (matched to the `1.17.0` core already in use). It gives the OTel-native `/metrics` pull endpoint; replacing it would mean hand-rolling Prometheus exposition over a `MeterListener`; the risk is that it is a **prerelease** package (the whole OTel .NET Prometheus line is `-beta`) — pinned, wired only in `Program.cs`, and gated by `verify-metrics`. Recorded in **ADR-0024**.

## Definition of Done

- [x] Linked Gitea issue (#124).
- [x] Failing test committed before the implementation (`test(bff): /metrics exposes http-server request duration`).
- [x] Implementation makes the test pass.
- [ ] CI green — pending Gitea Actions run.
- [x] `docker compose up` reaches green health within 3 min (backplane images unchanged in shape; not on the health gate, ADR-0023).
- [x] Docs updated — demo-script S-16c entry.
- [x] ADR added — ADR-0024.
- [x] Demo note in `docs/demo-script.md`.

## Notes for reviewers

- `/health` polls are counted as traffic (metrics aren't path-filtered, unlike traces). Fine for a demo dashboard and honest — real load stacks on top.
- `projection-api` has no Stryker config (unchanged); the four mutated services carry the metrics wiring in `Program.cs`, same as the merged S-16b tracing code.
- Metric names verified against a live service: `http_server_request_duration_seconds{,_bucket,_count}`, label `http_response_status_code`, `dotnet_process_cpu_time_seconds_total`.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Reviewed-on: #129

infra/metrics-check.py

@@ -0,0 +1,75 @@
+#!/usr/bin/env python3
+"""S-16c (#124): prove the golden-signal metrics pipeline works end to end.
+
+Generate anonymous BFF traffic (GET /openbaar/register — no auth, no OpenZaak egress),
+then query Prometheus and assert (1) every .NET service's scrape target is UP, and (2)
+the http.server.request.duration histogram is actually being scraped — i.e. the services
+expose /metrics AND Prometheus collects it, which is exactly what the golden-signal
+dashboard reads.
+
+Stdlib only (urllib/json) so it runs in a bare python:3-slim container in-network.
+"""
+import json
+import os
+import sys
+import time
+import urllib.error
+import urllib.parse
+import urllib.request
+
+BFF = os.environ["BFF"]          # http://<bff-ip>:8080
+PROM = os.environ["PROMETHEUS"]  # http://<prometheus-ip>:9090
+TIMEOUT = int(os.environ.get("METRICS_TIMEOUT", "90"))
+SERVICES = {"acl", "domain", "bff", "event-subscriber", "projection-api"}
+
+
+def _get(url):
+    with urllib.request.urlopen(url, timeout=10) as r:
+        return r.read()
+
+
+def generate_traffic():
+    for _ in range(3):
+        try:
+            _get(f"{BFF}/openbaar/register")
+        except urllib.error.HTTPError:
+            pass  # a non-2xx still records an http.server metric
+
+
+def query(promql):
+    q = urllib.parse.quote(promql)
+    try:
+        data = json.loads(_get(f"{PROM}/api/v1/query?query={q}"))
+    except Exception:
+        return []
+    return data.get("data", {}).get("result", [])
+
+
+def jobs_up():
+    return {r["metric"].get("job") for r in query("up == 1")}
+
+
+def jobs_with_request_metric():
+    return {r["metric"].get("job")
+            for r in query("http_server_request_duration_seconds_count")}
+
+
+def main():
+    deadline = time.time() + TIMEOUT
+    while time.time() < deadline:
+        generate_traffic()
+        up = jobs_up()
+        scraped = jobs_with_request_metric()
+        if SERVICES.issubset(up) and SERVICES.issubset(scraped):
+            print(f"OK — targets up: {sorted(up & SERVICES)}; "
+                  f"request metric scraped from: {sorted(scraped & SERVICES)}")
+            return 0
+        time.sleep(3)
+    print(f"FAIL — up: {sorted(jobs_up() & SERVICES)}; "
+          f"request metric from: {sorted(jobs_with_request_metric() & SERVICES)}; "
+          f"expected all of {sorted(SERVICES)}", file=sys.stderr)
+    return 1
+
+
+if __name__ == "__main__":
+    sys.exit(main())

infra/observability/grafana/Dockerfile

@@ -1,4 +1,4 @@
-# Grafana with datasources baked in via provisioning (S-16a, ADR-0023).
-# Dashboards (S-16c, #124) are added under provisioning/dashboards later.
+# Grafana with datasources + the golden-signals dashboard baked in via provisioning
+# (S-16a/S-16c, ADR-0023). Everything under provisioning/ is copied in below.
 FROM grafana/grafana:11.3.0
 COPY provisioning/ /etc/grafana/provisioning/

infra/observability/grafana/provisioning/dashboards/dashboards.yaml

@@ -0,0 +1,13 @@
+# Dashboard provider (S-16c, ADR-0023): Grafana loads every *.json in this folder as a
+# read-only, code-owned dashboard. The golden-signals board is versioned here, not
+# clicked together in the UI.
+apiVersion: 1
+
+providers:
+  - name: register-referentie
+    type: file
+    disableDeletion: true
+    allowUiUpdates: false
+    options:
+      path: /etc/grafana/provisioning/dashboards
+      foldersFromFilesStructure: false

infra/observability/grafana/provisioning/dashboards/golden-signals.json

@@ -0,0 +1,87 @@
+{
+  "uid": "golden-signals",
+  "title": "Request path — golden signals",
+  "tags": ["s-16c", "golden-signals"],
+  "timezone": "browser",
+  "schemaVersion": 39,
+  "version": 1,
+  "editable": true,
+  "refresh": "10s",
+  "time": { "from": "now-15m", "to": "now" },
+  "templating": {
+    "list": [
+      {
+        "name": "job",
+        "type": "query",
+        "datasource": { "type": "prometheus", "uid": "prometheus" },
+        "query": "label_values(http_server_request_duration_seconds_count, job)",
+        "includeAll": true,
+        "multi": true,
+        "current": { "text": "All", "value": "$__all" },
+        "refresh": 2
+      }
+    ]
+  },
+  "panels": [
+    {
+      "id": 1,
+      "title": "Traffic — requests/sec",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 0 },
+      "fieldConfig": { "defaults": { "unit": "reqps", "custom": { "drawStyle": "line", "fillOpacity": 10 } }, "overrides": [] },
+      "targets": [
+        {
+          "refId": "A",
+          "expr": "sum by (job) (rate(http_server_request_duration_seconds_count{job=~\"$job\"}[$__rate_interval]))",
+          "legendFormat": "{{job}}"
+        }
+      ]
+    },
+    {
+      "id": 2,
+      "title": "Errors — 5xx responses/sec",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 12, "y": 0 },
+      "fieldConfig": { "defaults": { "unit": "reqps", "custom": { "drawStyle": "line", "fillOpacity": 10 }, "color": { "mode": "fixed", "fixedColor": "red" } }, "overrides": [] },
+      "targets": [
+        {
+          "refId": "A",
+          "expr": "sum by (job) (rate(http_server_request_duration_seconds_count{job=~\"$job\",http_response_status_code=~\"5..\"}[$__rate_interval]))",
+          "legendFormat": "{{job}}"
+        }
+      ]
+    },
+    {
+      "id": 3,
+      "title": "Latency — p95 request duration",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 8 },
+      "fieldConfig": { "defaults": { "unit": "s", "custom": { "drawStyle": "line", "fillOpacity": 10 } }, "overrides": [] },
+      "targets": [
+        {
+          "refId": "A",
+          "expr": "histogram_quantile(0.95, sum by (job, le) (rate(http_server_request_duration_seconds_bucket{job=~\"$job\"}[$__rate_interval])))",
+          "legendFormat": "{{job}} p95"
+        }
+      ]
+    },
+    {
+      "id": 4,
+      "title": "Saturation — CPU cores in use",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 12, "y": 8 },
+      "fieldConfig": { "defaults": { "unit": "none", "custom": { "drawStyle": "line", "fillOpacity": 10 } }, "overrides": [] },
+      "targets": [
+        {
+          "refId": "A",
+          "expr": "sum by (job) (rate(dotnet_process_cpu_time_seconds_total{job=~\"$job\"}[$__rate_interval]))",
+          "legendFormat": "{{job}}"
+        }
+      ]
+    }
+  ]
+}

infra/observability/prometheus/prometheus.yml

@@ -1,6 +1,7 @@
-# Prometheus scrape config (S-16a, ADR-0023). For the backplane slice it scrapes
-# only itself; the .NET services' /metrics scrape targets are added in S-16c
-# (#124) when the services expose metrics.
+# Prometheus scrape config (S-16c, ADR-0023). Each .NET service exposes OTel metrics
+# at /metrics (Prometheus text format); one scrape job per service, so the service is
+# identified by the `job` label in the golden-signal dashboard. Targets are reached by
+# compose service name on the shared `cg` network (internal port 8080).
 global:
   scrape_interval: 15s
 
@@ -8,3 +9,19 @@ scrape_configs:
   - job_name: prometheus
     static_configs:
       - targets: ['localhost:9090']
+
+  - job_name: acl
+    static_configs:
+      - targets: ['acl:8080']
+  - job_name: domain
+    static_configs:
+      - targets: ['domain:8080']
+  - job_name: bff
+    static_configs:
+      - targets: ['bff:8080']
+  - job_name: event-subscriber
+    static_configs:
+      - targets: ['event-subscriber:8080']
+  - job_name: projection-api
+    static_configs:
+      - targets: ['projection-api:8080']

infra/run-metrics-check.sh

@@ -0,0 +1,28 @@
+#!/usr/bin/env bash
+#
+# S-16c (#124): assert the golden-signal metrics pipeline works — the .NET services expose
+# /metrics and Prometheus scrapes them — against an ALREADY-RUNNING full stack. Runs the
+# driver in a python:3-slim container on the stack network (services reached by container IP;
+# the runner can't reach published ports — gitea-actions-gotchas.md §5/§6). Does NOT manage
+# the stack lifecycle.
+set -euo pipefail
+
+here="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+
+ip() { docker inspect -f '{{range .NetworkSettings.Networks}}{{.IPAddress}}{{end}}' "$1"; }
+
+bff="$(docker ps -q --filter 'name=[-_]bff[-_]' | head -1)"
+prom="$(docker ps -q --filter 'name=[-_]prometheus[-_]' | head -1)"
+[ -n "$bff" ] && [ -n "$prom" ] || { echo "ERROR: bff and/or prometheus not running — bring the stack up first" >&2; exit 1; }
+net="$(docker inspect -f '{{range $k,$_ := .NetworkSettings.Networks}}{{$k}}{{"\n"}}{{end}}' "$bff" | head -1)"
+bff_ip="$(ip "$bff")"; prom_ip="$(ip "$prom")"
+echo ">> network=$net bff=$bff_ip prometheus=$prom_ip"
+
+cid="$(docker create --network "$net" \
+  -e "BFF=http://$bff_ip:8080" -e "PROMETHEUS=http://$prom_ip:9090" \
+  -e "METRICS_TIMEOUT=${METRICS_TIMEOUT:-90}" \
+  python:3-slim python /metrics-check.py)"
+docker cp "$here/metrics-check.py" "$cid:/metrics-check.py" >/dev/null
+rc=0; docker start -a "$cid" || rc=$?
+docker rm -f "$cid" >/dev/null
+exit $rc