prometheus.yml

global:
  scrape_interval: 15s
  evaluation_interval: 15s
  external_labels:
    project: openchainbench

# OpenChainBench is a federation: each harness is hosted by whoever wrote it
# (Mobula, an independent contributor, a provider that wants to be benchmarked).
# This single shared Prometheus scrapes every harness's public /metrics
# endpoint and aggregates everything into one queryable instance. the URL
# referenced by every YAML spec's `prom_url`.
#
# To onboard a new harness:
#   1. The contributor exposes /metrics on a publicly reachable URL
#      (Railway, Fly, Cloud Run, a VPS. anywhere with HTTPS).
#   2. Append a `- job_name: <slug>` block below pointing at that URL.
#   3. Open a PR. Once merged, redeploy the prometheus service so it picks
#      up the new target (or hit the /-/reload endpoint).
#
# All metrics receive a `benchmark: <slug>` label automatically; YAML specs
# use it to scope queries.

scrape_configs:
  # ─ Aggregator head lag ────────────────────────────────────────────────
  # Currently hosted by the maintainers on Railway. Replace the URL if you fork.
  - job_name: aggregator-head-lag
    metrics_path: /metrics
    scheme: https
    static_configs:
      - targets:
          - aggregator-head-lag-production.up.railway.app
        labels:
          benchmark: aggregator-head-lag
          host: openchainbench

  # ─ Bridge quote latency + cost ────────────────────────────────────────
  # Currently hosted by the maintainers on Railway. Replace the URL if you fork.
  - job_name: bridge-monitor
    metrics_path: /metrics
    scheme: https
    static_configs:
      - targets:
          - bridge-monitor-production.up.railway.app
        labels:
          benchmark: bridge-monitor
          host: openchainbench

  # ─ Wallet labels coverage ────────────────────────────────────────────
  # Currently hosted by the maintainers on Railway. Replace the URL if you fork.
  - job_name: wallet-labels
    metrics_path: /metrics
    scheme: https
    static_configs:
      - targets:
          - upstream-monorepo-production-1f1b.up.railway.app
        labels:
          benchmark: wallet-labels
          host: openchainbench

  # ─ L1 finality ────────────────────────────────────────────────────────
  # Currently hosted by the maintainers on Railway. Replace the URL if you fork.
  - job_name: l1-finality
    metrics_path: /metrics
    scheme: https
    static_configs:
      - targets:
          - l1-finality-production.up.railway.app
        labels:
          benchmark: l1-finality
          host: openchainbench

  # ─ Metadata coverage ──────────────────────────────────────────────────
  # Currently hosted by the maintainers on Railway. Replace the URL if you fork.
  - job_name: metadata-coverage
    metrics_path: /metrics
    scheme: https
    static_configs:
      - targets:
          - metadata-coverage-production.up.railway.app
        labels:
          benchmark: metadata-coverage
          host: openchainbench

  # ─ Network coverage ───────────────────────────────────────────────────
  # Currently hosted by the maintainers on Railway. Replace the URL if you fork.
  - job_name: network-coverage
    metrics_path: /metrics
    scheme: https
    static_configs:
      - targets:
          - network-coverage-production.up.railway.app
        labels:
          benchmark: network-coverage
          host: openchainbench

  # ─ Perp fees ──────────────────────────────────────────────────────────
  # Currently hosted by the maintainers on Railway. Replace the URL if you fork.
  - job_name: perp-fees
    metrics_path: /metrics
    scheme: https
    static_configs:
      - targets:
          - perp-fees-production.up.railway.app
        labels:
          benchmark: perp-fees
          host: openchainbench

  # ─ PM data freshness ─────────────────────────────────────────────────
  # Cross-correlates Polymarket CLOB WebSocket (T0) vs Mobula PM WS vs
  # Codex (defined.fi) GraphQL WS for the same trade. Reports per-provider
  # millisecond lag.
  #
  # Scraped over Railway internal DNS because the harness lives in the
  # same `Dashboard OpenChainBench` Railway project as this Prometheus.
  # Public *.up.railway.app routes go through the edge and Railway does
  # not always allow same-project traffic to loop back via the public
  # gateway. Internal DNS is HTTP on port 2112 (the harness's exposed
  # /metrics port) — never goes through Cloudflare / edge.
  - job_name: pm-freshness
    metrics_path: /metrics
    scheme: http
    static_configs:
      - targets:
          - pm-freshness-bench-production.railway.internal:2112
        labels:
          benchmark: pm-freshness
          host: openchainbench

  # ─ PM venue rate limits + latency ────────────────────────────────────
  # Probes the public APIs of 5 prediction-market venues (Polymarket,
  # Kalshi, Limitless, Manifold, Myriad): warm/cold latency on book,
  # price and list classes, plus a daily rate-limit ramp. One Railway
  # service per region, scraped over internal DNS like pm-freshness.
  # Service names must match these targets when deploying.
  - job_name: pm-rate-limits
    metrics_path: /metrics
    scheme: http
    static_configs:
      - targets:
          - pm-rate-limits-us.railway.internal:2112
          - pm-rate-limits-eu.railway.internal:2112
          - pm-rate-limits-sgp.railway.internal:2112
        labels:
          benchmark: pm-rate-limits
          host: openchainbench

  # ─ Fiat on-ramp all-in cost (bench 262) ───────────────────────────────
  # One EU vantage (persona FR/EUR), container ocb-fiat-onramp-cost on the
  # Paris VPS next to Prom. Harness quotes every 5 min; the scrape only
  # refreshes gauges, sample counts come from onramp_quote_samples_total.
  - job_name: fiat-onramp-cost
    metrics_path: /metrics
    scheme: http
    static_configs:
      - targets:
          - ocb-fiat-onramp-cost:2113
        labels:
          benchmark: fiat-onramp-cost
          host: openchainbench

  # ─ Perp DEX daily volume (bench 266) ──────────────────────────────────
  # Container ocb-perp-volume-history on the Paris VPS. One number per
  # venue per closed UTC day, backfilled 400 days; the scrape refreshes
  # the last-day / 7d / 30d gauges, the full series is served as JSON by
  # Caddy at kv.openchainbench.com/aggregate/perp-volume/history.json.
  - job_name: perp-volume-history
    metrics_path: /metrics
    scheme: http
    static_configs:
      - targets:
          - ocb-perp-volume-history:2112
        labels:
          benchmark: perp-daily-volume
          host: openchainbench

  # ─ Trading app daily volume (bench 267) ───────────────────────────────
  # Container ocb-trading-app-volume on the Paris VPS. Cross-chain swap
  # volume per trading app / Telegram bot per closed UTC day from
  # DeFiLlama's free dexs summary; the full series with the per-chain
  # split is served by Caddy at
  # kv.openchainbench.com/aggregate/trading-apps/history.json.
  - job_name: trading-app-volume
    metrics_path: /metrics
    scheme: http
    static_configs:
      - targets:
          - ocb-trading-app-volume:2112
        labels:
          benchmark: trading-app-daily-volume
          host: openchainbench

  # ─ Terminal activity (benches 203, 206, 207, 232) ─────────────────────
  # Container ocb-terminal-activity on the OCB VPS. Daily volume,
  # transactions, fees and distinct wallets per trading terminal per chain,
  # from the public tehcscreener API (Allium-backed, no key). Replaces the
  # dune-platform-volume source for these four benches, which were gated to
  # staging on 2026-10-02 when the Dune trial ended. No secret to inject.
  - job_name: terminal-activity
    metrics_path: /metrics
    scheme: http
    static_configs:
      - targets:
          - ocb-terminal-activity:2116
        labels:
          benchmark: trading-platform-wallets
          host: openchainbench

  # ─ USDC corridor flows (bench 281) ────────────────────────────────────
  # Container ocb-bridge-flows on the Paris VPS. Circle CCTP burns
  # (DepositForBurn v1 and v2) read off seven EVM chains over public RPCs,
  # bucketed per destination; Wormholescan outbound volume per chain per
  # UTC day; L2Beat TVS deltas. State under /data/state/bridge-flows.
  - job_name: bridge-flows
    metrics_path: /metrics
    scheme: http
    static_configs:
      - targets:
          - ocb-bridge-flows:2112
        labels:
          benchmark: usdc-corridor-flows
          host: openchainbench

  # Container ocb-terminal-fill-quality on the Paris VPS. Cost per swap on
  # each Solana trading terminal from sampled user transactions read
  # on-chain (bench 268); per-terminal stats and recent samples served by
  # Caddy at kv.openchainbench.com/aggregate/terminal-fills/fills.json.
  - job_name: terminal-fill-quality
    metrics_path: /metrics
    scheme: http
    static_configs:
      - targets:
          - ocb-terminal-fill-quality:2112
        labels:
          benchmark: terminal-fill-quality
          host: openchainbench

  # Container ocb-rpc-cost. What each RPC provider bills for a fixed set of
  # workload profiles (bench 282), plus the freshness and drift of every
  # upstream pricing artifact the catalogue was verified against. Holds no
  # API keys and sends no RPC traffic: its only outbound requests are HTTP
  # GETs of public pricing pages.
  - job_name: rpc-cost
    metrics_path: /metrics
    scheme: http
    scrape_interval: 5m
    static_configs:
      - targets:
          - ocb-rpc-cost:2112
        labels:
          benchmark: rpc-cost
          host: openchainbench

  # Container ocb-price-api-accuracy. How far the USD price each market-data
  # API publishes sits from the Coinbase/Kraken/Binance mid of the same
  # moment (bench 305). Keys in /run/ocb/.env.price-api-accuracy; CoinGecko
  # goes through the metadata-coverage proxy. Recording rules in
  # infrastructure/monitoring/prometheus/recording_rules/ocb_price_api_accuracy.yml.
  - job_name: price-api-accuracy
    metrics_path: /metrics
    scheme: http
    static_configs:
      - targets:
          - ocb-price-api-accuracy:2112
        labels:
          benchmark: price-api-accuracy
          host: openchainbench

  # ─ Self-scrape ────────────────────────────────────────────────────────
  # Surfaces Prometheus's own health (scrape error rates, target up/down)
  # in the same UI used to debug harnesses.
  - job_name: prometheus
    static_configs:
      - targets:
          - localhost:9090