Usage.

Hook approach

Build with -DWITH_METRICS for a prometheus exporter hook. Alternatively use image soullessblob/villas-node:prometheus the same way one would use the official image. Metrics follow the prometheus best practices, and has the form <metric_name>{node=<node_short_name>, accumulator=<stat_type_name>}, with stat_type_name originating from Stats::types. They are exposed by adding a hook of metrics type to the in/out field of a node entry in the node config. Exposing more than one port will not work because the first constructed hook sets the exposer to the corresponding port, next ones will not attempt to redeclare the exposer if a port is set.

Example docker compose.

services:
  node:
    image: soullessblob/villas-node:prometheus
    hostname: node-prom
    container_name: node-prom
    command: ["node","-d","debug","/configs/test-conf.conf"]
    volumes:
      - ./test-conf.conf:/configs/test-conf.conf
    ports:
      - 9096:9096
      - 80:80

  prom:
    image: prom/prometheus
    hostname: prom
    container_name: prom
    volumes:
      - ./prometheus.yml:/etc/prometheus/prometheus.yml
    ports:
      - 9090:9090

With node config:

{
  "http": {
    "port": 8080
  },
  "hugepages": 0,
  "nodes": {
    "siggen": {
      "type": "signal",
      "signal": [
        "sine",
        "sine"
      ],
      "rate": 10,
      "values": 2
    },
    "relay": {
      "type": "websocket",
      "destinations": [
        "https://s-dev.k8s.eonerc.rwth-aachen.de/ws/relay/hyperride-test"
      ],
       "in":{
          "hooks":[
            {
              "type":"metrics",
              "port":9096
            }
          ]
            
        }
    }
  },
  "paths": [
    {
      "in": "siggen",
      "out": "relay"
    }
  ],
  "logging": {
    "level": "info"
  },
  "uuid": "b3b45caf-c560-4d9c-9f66-4b9538a0a07a"
}

And a static prometheus config for testing:

global:
  scrape_interval:     15s

  external_labels:
    monitor: 'codelab-monitor'

scrape_configs:
  - job_name: 'prometheus'

    scrape_interval: 5s

    static_configs:
      - targets: ['node-prom:9096']

API approach

Found in the no-extern-deps branch. Build and configure node according to the official documentation. Metrics are exposed by adding a hook of type "stats" to the node config from which metrics are required. They are then found under the /metrics path. The scrape_configs of prometheus.yml become:

global:
  scrape_interval:     15s

  external_labels:
    monitor: 'codelab-monitor'

scrape_configs:
  - job_name: 'prometheus'

    scrape_interval: 5s
    metrics_path: '/api/v2/metrics' #<=== add this
    static_configs:
      - targets: ['node-prom:8080'] #<== metrics are now exposed under the default hostname:port of the villasnode http config