chore: import upstream snapshot with attribution

docker/Dockerfile.dev

@@ -0,0 +1,39 @@
+# Use a Python image with uv pre-installed
+FROM ghcr.io/astral-sh/uv:python3.12-bookworm
+
+# Setup a non-root user
+RUN groupadd --system --gid 999 nonroot \
+ && useradd --system --gid 999 --uid 999 --create-home nonroot
+
+# Install the project into `/app`
+WORKDIR /app
+
+# Enable bytecode compilation
+ENV UV_COMPILE_BYTECODE=1
+
+# Copy from the cache instead of linking since it's a mounted volume
+ENV UV_LINK_MODE=copy
+
+# Ensure installed tools can be executed out of the box
+ENV UV_TOOL_BIN_DIR=/usr/local/bin
+
+# Install the project's dependencies using the lockfile and settings
+RUN --mount=type=cache,target=/root/.cache/uv \
+    --mount=type=bind,source=uv.lock,target=uv.lock \
+    --mount=type=bind,source=pyproject.toml,target=pyproject.toml \
+    uv sync --locked --no-install-project --group dev --extra mongo --group core-stable
+
+# Then, add the rest of the project source code and install it
+# Installing separately from its dependencies allows optimal layer caching
+COPY . /app
+RUN --mount=type=cache,target=/root/.cache/uv \
+    uv sync --locked --group dev --extra mongo --group core-stable
+
+# Place executables in the environment at the front of the path
+ENV PATH="/app/.venv/bin:$PATH"
+
+# Reset the entrypoint, don't invoke `uv`
+ENTRYPOINT []
+
+# Use the non-root user to run our application
+USER nonroot

docker/compose.grafana.yml

@@ -0,0 +1,29 @@
+services:
+  prometheus:
+    image: prom/prometheus:latest
+    command:
+      - "--config.file=/etc/prometheus/prometheus.yml"
+      - "--storage.tsdb.path=/prometheus"
+    volumes:
+      - ./prometheus/prometheus.base.yml:/etc/prometheus/prometheus.yml:ro
+      - ${AGL_MONITORING_DATA_PATH:?Set AGL_MONITORING_DATA_PATH to the metrics directory}/prometheus:/prometheus
+    ports:
+      - "9090:9090"
+
+  grafana:
+    image: grafana/grafana:latest
+    depends_on:
+      - prometheus
+    ports:
+      - "9091:3000"
+    volumes:
+      - ./data/grafana:/var/lib/grafana
+      - ./grafana/datasource.yml:/etc/grafana/provisioning/datasources/datasource.yml
+      - ./grafana/dashboard-provider.yml:/etc/grafana/provisioning/dashboards/provider.yml
+      - ./grafana/dashboards:/var/lib/grafana/dashboards
+    environment:
+      - GF_INSTALL_PLUGINS=grafana-piechart-panel
+      - GF_AUTH_ANONYMOUS_ENABLED=true
+      - GF_AUTH_ANONYMOUS_ORG_ROLE=Admin
+      - GF_AUTH_DISABLE_LOGIN_FORM=true
+      - GF_DASHBOARDS_DEFAULT_HOME_DASHBOARD_PATH=/var/lib/grafana/dashboards/agentlightning.json

docker/compose.mongo.yml

@@ -0,0 +1,22 @@
+# Docker-compose file to launch a MongoDB server for development.
+# It's used to test the MongoDB store implementation.
+
+services:
+  mongo:
+    image: mongo:8.2
+    ulimits:
+      nofile:
+        soft: 65535
+        hard: 65535
+    ports:
+      - "27017:27017"
+    command: ["mongod", "--bind_ip_all", "--replSet", "rs0"]
+    volumes:
+      - ../scripts/mongodb_init_rs_host.js:/docker-entrypoint-initdb.d/init-rs.js:ro
+      - ./data/mongo-host:/data/db
+    healthcheck:
+      test: ["CMD", "mongosh", "--eval", "db.adminCommand('ping')"]
+      interval: 10s
+      timeout: 5s
+      retries: 5
+      start_period: 30s

docker/compose.prometheus-memory-store.yml

@@ -0,0 +1,81 @@
+services:
+  app:
+    extends:
+      file: compose.store.yml
+      service: app
+
+    depends_on:
+      - app-exporter  # Wait for the exporter to be ready first
+
+    command: agl store --host 0.0.0.0 --port 4747 --tracker console prometheus --backend memory
+    environment:
+      - PROMETHEUS_MULTIPROC_DIR=/tmp/prometheus_multiproc
+    volumes:
+      - prometheus_multiproc:/tmp/prometheus_multiproc
+
+  app-exporter:
+    build:
+      context: ../
+      dockerfile: docker/Dockerfile.dev
+
+    command: agl prometheus --host 0.0.0.0 --port 4748
+    ports:
+      - "4748:4748"
+    environment:
+      - PROMETHEUS_MULTIPROC_DIR=/tmp/prometheus_multiproc
+    volumes:
+      - prometheus_multiproc:/tmp/prometheus_multiproc
+
+  node-exporter:
+    image: prom/node-exporter:latest
+    # In CI you might not have full /proc, but this is OK for container-level stats
+    pid: "host"
+    command:
+      - "--path.rootfs=/host"
+    volumes:
+      - "/:/host:ro,rslave"
+
+  prometheus:
+    image: prom/prometheus:latest
+    command:
+      - "--config.file=/etc/prometheus/prometheus.yml"
+      - "--storage.tsdb.path=/prometheus"
+      - "--storage.tsdb.retention.time=1h"
+    volumes:
+      - ./prometheus/prometheus.base.yml:/etc/prometheus/prometheus.yml:ro
+      - ./data/prometheus:/prometheus
+    depends_on:
+      - app
+      - app-exporter
+      - node-exporter
+    ports:
+      - "9090:9090"
+
+  grafana:
+    image: grafana/grafana:latest
+    ports:
+      - "9091:3000"
+    depends_on:
+      - prometheus
+    volumes:
+      - ./data/grafana:/var/lib/grafana
+      # 1. Mount the Datasource Config
+      - ./grafana/datasource.yml:/etc/grafana/provisioning/datasources/datasource.yml
+      # 2. Mount the Dashboard Provider Config
+      - ./grafana/dashboard-provider.yml:/etc/grafana/provisioning/dashboards/provider.yml
+      # 3. Mount the folder containing the actual JSON files
+      - ./grafana/dashboards:/var/lib/grafana/dashboards
+
+    environment:
+      - GF_INSTALL_PLUGINS=grafana-piechart-panel
+      - GF_AUTH_ANONYMOUS_ENABLED=true
+      - GF_AUTH_ANONYMOUS_ORG_ROLE=Admin
+      - GF_AUTH_DISABLE_LOGIN_FORM=true
+      - GF_DASHBOARDS_DEFAULT_HOME_DASHBOARD_PATH=/var/lib/grafana/dashboards/agentlightning.json
+
+volumes:
+  prometheus_multiproc:
+    driver: local
+    driver_opts:
+      type: tmpfs
+      device: tmpfs

docker/compose.prometheus-mongo-store.yml

@@ -0,0 +1,116 @@
+services:
+  mongo:
+    extends:
+      file: compose.mongo.yml
+      service: mongo
+
+    hostname: mongo
+    volumes:
+      - ./data/mongo-container:/data/db
+      - ../scripts/mongodb_init_rs_profiling.js:/docker-entrypoint-initdb.d/init-rs.js:ro
+
+    # This forces "mongo" to resolve to localhost ONLY inside this container.
+    # This allows rs.initiate to succeed using the hostname "mongo".
+    extra_hosts:
+      - "mongo:127.0.0.1"
+
+  app:
+    extends:
+      file: compose.store.yml
+      service: app
+
+    depends_on:
+      - mongo
+      - app-exporter  # Wait for the exporter to be ready first
+
+    command:
+      - /bin/bash
+      - -c
+      - |
+          agl store --host 0.0.0.0 --port 4747 \
+            --tracker console prometheus --backend mongo \
+            --mongo-uri mongodb://mongo:27017/?replicaSet=rs0 \
+            --n-workers ${AGL_STORE_N_WORKERS:-32}
+    environment:
+      - PROMETHEUS_MULTIPROC_DIR=/tmp/prometheus_multiproc
+    volumes:
+      - prometheus_multiproc:/tmp/prometheus_multiproc
+
+  app-exporter:
+    build:
+      context: ../
+      dockerfile: docker/Dockerfile.dev
+
+    command: agl prometheus --host 0.0.0.0 --port 4748
+    ports:
+      - "4748:4748"
+    environment:
+      - PROMETHEUS_MULTIPROC_DIR=/tmp/prometheus_multiproc
+    volumes:
+      - prometheus_multiproc:/tmp/prometheus_multiproc
+
+  mongodb-exporter:
+    image: percona/mongodb_exporter:0.47.1
+    command:
+      - "--mongodb.uri=mongodb://mongo:27017/"
+      - "--collect-all"
+      - "--mongodb.collstats-colls=agentlightning.rollouts,agentlightning.attempts,agentlightning.spans,agentlightning.resources,agentlightning.workers,agentlightning.rollout_queue,agentlightning.span_sequence_ids"
+    depends_on:
+      - mongo
+    ports:
+      - "9216:9216"
+
+  node-exporter:
+    image: prom/node-exporter:latest
+    # In CI you might not have full /proc, but this is OK for container-level stats
+    pid: "host"
+    command:
+      - "--path.rootfs=/host"
+    volumes:
+      - "/:/host:ro,rslave"
+
+  prometheus:
+    image: prom/prometheus:latest
+    command:
+      - "--config.file=/etc/prometheus/prometheus.yml"
+      - "--storage.tsdb.path=/prometheus"
+      - "--storage.tsdb.retention.time=1h"
+    volumes:
+      - ./prometheus/prometheus.mongo.yml:/etc/prometheus/prometheus.yml:ro
+      - ./data/prometheus:/prometheus
+    depends_on:
+      - app
+      - app-exporter
+      - mongodb-exporter
+      - node-exporter
+    ports:
+      - "9090:9090"
+
+  grafana:
+    image: grafana/grafana:latest
+    ports:
+      - "9091:3000"
+    depends_on:
+      - prometheus
+    volumes:
+      - ./data/grafana:/var/lib/grafana
+      # 1. Mount the Datasource Config
+      - ./grafana/datasource.yml:/etc/grafana/provisioning/datasources/datasource.yml
+      # 2. Mount the Dashboard Provider Config
+      - ./grafana/dashboard-provider.yml:/etc/grafana/provisioning/dashboards/provider.yml
+      # 3. Mount the folder containing the actual JSON files
+      - ./grafana/dashboards:/var/lib/grafana/dashboards
+
+    environment:
+      - GF_INSTALL_PLUGINS=grafana-piechart-panel
+      - GF_AUTH_ANONYMOUS_ENABLED=true
+      - GF_AUTH_ANONYMOUS_ORG_ROLE=Admin
+      - GF_AUTH_DISABLE_LOGIN_FORM=true
+      - GF_DASHBOARDS_DEFAULT_HOME_DASHBOARD_PATH=/var/lib/grafana/dashboards/agentlightning.json
+
+volumes:
+  prometheus_multiproc:
+    driver: local
+    driver_opts:
+      type: tmpfs
+      device: tmpfs

docker/compose.store.yml

@@ -0,0 +1,30 @@
+services:
+  app:
+    build:
+      context: ../
+      dockerfile: docker/Dockerfile.dev
+
+    ports:
+      - "4747:4747"
+
+    command: agl store --host 0.0.0.0 --port 4747
+
+    ulimits:
+      nofile:
+        soft: 65535
+        hard: 65535
+
+    develop:
+      watch:
+        # Sync the working directory with the `/app` directory in the container
+        - action: sync
+          path: ..
+          target: /app
+          # Exclude the project virtual environment — it could be for a
+          # different platform in the container
+          ignore:
+            - .venv/
+
+        # Rebuild the image if dependencies change by checking uv.lock
+        - action: rebuild
+          path: ../uv.lock

docker/grafana/dashboard-provider.yml

@@ -0,0 +1,12 @@
+apiVersion: 1
+
+providers:
+  - name: "default"
+    orgId: 1
+    folder: ""
+    type: file
+    disableDeletion: false
+    updateIntervalSeconds: 10
+    options:
+      # This tells Grafana to look for JSON files in this directory inside the container
+      path: /var/lib/grafana/dashboards

docker/grafana/dashboards/agentlightning.json

@@ -0,0 +1,530 @@
+{
+  "annotations": {
+    "list": []
+  },
+  "editable": true,
+  "gnetId": null,
+  "graphTooltip": 0,
+  "id": null,
+  "links": [],
+  "panels": [
+    {
+      "collapsed": false,
+      "gridPos": {
+        "h": 1,
+        "w": 24,
+        "x": 0,
+        "y": 0
+      },
+      "id": 100,
+      "panels": [],
+      "title": "LightningStore HTTP API",
+      "type": "row"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "ops/s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 8,
+        "x": 0,
+        "y": 1
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "sum by (method, path) (rate(agl_http_latency_count[15s]))",
+          "legendFormat": "{{method}} {{path}}"
+        }
+      ],
+      "title": "HTTP Requests / Sec (by Path)",
+      "type": "timeseries"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 8,
+        "x": 8,
+        "y": 1
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "histogram_quantile(0.50, sum by (le) (rate(agl_http_latency_bucket[15s])))",
+          "legendFormat": "P50"
+        },
+        {
+          "expr": "histogram_quantile(0.95, sum by (le) (rate(agl_http_latency_bucket[15s])))",
+          "legendFormat": "P95"
+        },
+        {
+          "expr": "histogram_quantile(0.99, sum by (le) (rate(agl_http_latency_bucket[15s])))",
+          "legendFormat": "P99"
+        }
+      ],
+      "title": "HTTP Latency (P50, P95, P99)",
+      "type": "timeseries"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "ops/s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 8,
+        "x": 16,
+        "y": 1
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "sum by (method, path, status) (rate(agl_http_latency_count{status!~\"2..\"}[15s]))",
+          "legendFormat": "{{method}} {{path}} {{status}}"
+        }
+      ],
+      "title": "HTTP Errors / Sec (non-2xx)",
+      "type": "timeseries"
+    },
+    {
+      "collapsed": false,
+      "gridPos": {
+        "h": 1,
+        "w": 24,
+        "x": 0,
+        "y": 9
+      },
+      "id": 200,
+      "panels": [],
+      "title": "LightningStore RPC Metrics",
+      "type": "row"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "ops/s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 8,
+        "x": 0,
+        "y": 10
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "sum by (method) (rate(agl_store_latency_count[15s]))",
+          "legendFormat": "{{method}}"
+        }
+      ],
+      "title": "Store Calls / Sec (by Method)",
+      "type": "timeseries"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 8,
+        "x": 8,
+        "y": 10
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "histogram_quantile(0.50, sum by (le) (rate(agl_store_latency_bucket[15s])))",
+          "legendFormat": "P50"
+        },
+        {
+          "expr": "histogram_quantile(0.95, sum by (le) (rate(agl_store_latency_bucket[15s])))",
+          "legendFormat": "P95"
+        },
+        {
+          "expr": "histogram_quantile(0.99, sum by (le) (rate(agl_store_latency_bucket[15s])))",
+          "legendFormat": "P99"
+        }
+      ],
+      "title": "Store Latency (P50, P95, P99)",
+      "type": "timeseries"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "ops/s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 8,
+        "x": 16,
+        "y": 10
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "sum by (method, status) (rate(agl_store_latency_count{status!=\"OK\"}[15s]))",
+          "legendFormat": "{{method}} {{status}}"
+        }
+      ],
+      "title": "Store Errors / Sec (by Method)",
+      "type": "timeseries"
+    },
+    {
+      "collapsed": false,
+      "gridPos": {
+        "h": 1,
+        "w": 24,
+        "x": 0,
+        "y": 18
+      },
+      "id": 300,
+      "panels": [],
+      "title": "Collection Metrics",
+      "type": "row"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "ops/s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 12,
+        "x": 0,
+        "y": 19
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "sum by (store_pubmeth, operation) (rate(agl_collections_latency_count[15s]))",
+          "legendFormat": "{{store_pubmeth}} {{operation}}"
+        }
+      ],
+      "title": "Collection Ops / Sec (by Store Method)",
+      "type": "timeseries"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "ops/s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 12,
+        "x": 12,
+        "y": 19
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "sum by (store_pubmeth, collection, status) (rate(agl_collections_latency_count{status!=\"OK\"}[15s]))",
+          "legendFormat": "{{store_pubmeth}} {{collection}} {{status}}"
+        }
+      ],
+      "title": "Collection Errors / Sec",
+      "type": "timeseries"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 12,
+        "x": 0,
+        "y": 27
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "histogram_quantile(0.50, sum by (le) (rate(agl_collections_latency_bucket[15s])))",
+          "legendFormat": "P50"
+        },
+        {
+          "expr": "histogram_quantile(0.95, sum by (le) (rate(agl_collections_latency_bucket[15s])))",
+          "legendFormat": "P95"
+        },
+        {
+          "expr": "histogram_quantile(0.99, sum by (le) (rate(agl_collections_latency_bucket[15s])))",
+          "legendFormat": "P99"
+        }
+      ],
+      "title": "Collection Latency (P50, P95, P99)",
+      "type": "timeseries"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 12,
+        "x": 12,
+        "y": 27
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "histogram_quantile(0.95, sum by (le, store_pubmeth) (rate(agl_collections_latency_bucket[15s])))",
+          "legendFormat": "{{store_pubmeth}}"
+        }
+      ],
+      "title": "Collection Latency P95 (by Store Method)",
+      "type": "timeseries"
+    },
+    {
+      "collapsed": false,
+      "gridPos": {
+        "h": 1,
+        "w": 24,
+        "x": 0,
+        "y": 35
+      },
+      "id": 400,
+      "panels": [],
+      "title": "Rollout Outcomes",
+      "type": "row"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "ops/s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 12,
+        "x": 0,
+        "y": 36
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "sum by (status) (rate(agl_rollouts_duration_count[15s]))",
+          "legendFormat": "{{status}}"
+        }
+      ],
+      "title": "Rollout Completion Rate (by Status)",
+      "type": "timeseries"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "ops/s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 12,
+        "x": 12,
+        "y": 36
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "sum by (mode) (rate(agl_rollouts_duration_count[15s]))",
+          "legendFormat": "{{mode}}"
+        }
+      ],
+      "title": "Rollout Completion Rate (by Mode)",
+      "type": "timeseries"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 12,
+        "x": 0,
+        "y": 44
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "histogram_quantile(0.50, sum by (le) (rate(agl_rollouts_duration_bucket[15s])))",
+          "legendFormat": "P50"
+        },
+        {
+          "expr": "histogram_quantile(0.95, sum by (le) (rate(agl_rollouts_duration_bucket[15s])))",
+          "legendFormat": "P95"
+        },
+        {
+          "expr": "histogram_quantile(0.99, sum by (le) (rate(agl_rollouts_duration_bucket[15s])))",
+          "legendFormat": "P99"
+        }
+      ],
+      "title": "Rollout Duration (P50, P95, P99)",
+      "type": "timeseries"
+    },
+    {
+      "fieldConfig": {
+        "defaults": {
+          "unit": "s"
+        }
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 12,
+        "x": 12,
+        "y": 44
+      },
+      "options": {
+        "legend": {
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "expr": "histogram_quantile(0.95, sum by (le, status) (rate(agl_rollouts_duration_bucket[15s])))",
+          "legendFormat": "{{status}}"
+        }
+      ],
+      "title": "Rollout Duration P95 (by Status)",
+      "type": "timeseries"
+    }
+  ],
+  "schemaVersion": 36,
+  "style": "dark",
+  "tags": [
+    "agentlightning"
+  ],
+  "templating": {
+    "list": []
+  },
+  "time": {
+    "from": "now-10m",
+    "to": "now"
+  },
+  "title": "Agent Lightning Metrics",
+  "version": 2
+}

docker/grafana/datasource.yml

@@ -0,0 +1,8 @@
+apiVersion: 1
+
+datasources:
+  - name: Prometheus
+    type: prometheus
+    access: proxy
+    url: http://prometheus:9090
+    isDefault: true

docker/prometheus/prometheus.base.yml

@@ -0,0 +1,13 @@
+global:
+  scrape_interval: 5s
+  evaluation_interval: 5s
+
+scrape_configs:
+  - job_name: app
+    static_configs:
+      - targets: ["app-exporter:4748"]
+    metrics_path: /v1/prometheus/
+
+  - job_name: node
+    static_configs:
+      - targets: ["node-exporter:9100"]

docker/prometheus/prometheus.mongo.yml

@@ -0,0 +1,17 @@
+global:
+  scrape_interval: 5s
+  evaluation_interval: 5s
+
+scrape_configs:
+  - job_name: app
+    static_configs:
+      - targets: ["app-exporter:4748"]
+    metrics_path: /v1/prometheus/
+
+  - job_name: node
+    static_configs:
+      - targets: ["node-exporter:9100"]
+
+  - job_name: mongodb
+    static_configs:
+      - targets: ["mongodb-exporter:9216"]

docker/setup.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+
+set -euo pipefail
+
+# Create data directories
+mkdir -p data/prometheus data/mongo-container data/mongo-host data/grafana
+
+# Change permissions
+chmod 777 data/prometheus data/mongo-container data/mongo-host data/grafana