roybay / metrics-client

Hapi-centric Prometheus Plugin and optional endpoint

metrics-client

Hapi-centric Prometheus Plugin and optional endpoint

Use this plugin to:

• Add an endpoint for metrics (defaults to /metrics)
• Add custom metrics in routes (server binds .metrics in request contexts)
• Run a server for worker that reports metrics from an endpoint (defaults to /metrics on port 3000)

Integration

Modules

Once metric-client is a Hapi-centric Prometheus Plugin. Hapi and metrics-client modules need to be installed.

npm install -S @hapi/hapi
npm install -S @pager/metrics-client

Usage

Sample usage for metrics-client server

'use strict';

const Hapi = require('@hapi/hapi');
const Metrics = require('@pager/metrics-client');

const startServer = async () => {

    const server = await new Hapi.Server({
        port: 3000
    });
    await server.register([
        {
            plugin: Metrics
        }
    ]);
    await server.start();
    console.log( `Server started at ${ server.info.uri }`);
    return server;
};

module.exports = startServer();

Annotations

annotations:
  prometheus.io/scrape: "true"
  prometheus.io/port: {{ .Values.service.internalPort | quote }}

Readiness and Liveness

Readiness probes are designed to let Kubernetes know when your app is ready to serve traffic. Similarly, liveness probes let Kubernetes know if your app is alive or dead. Metric-client plugin is also support health check and needs to be added to the deployment file under container details.

livenessProbe:
  httpGet:
    path: /health
    port: {{ .Values.service.internalPort }} 
  initialDelaySeconds: 15
  timeoutSeconds: 1
readinessProbe:
  httpGet:
    path: /health
    port: {{ .Values.service.internalPort }}
  initialDelaySeconds: 15
  timeoutSeconds: 1

Internal Ports

Prometheus server Kubernetes service discovery configuration is going to search running server on kubernetes cluster. It will use default server port 3000. Thus, metrics-client must start at default port 3000. If the server is not already running on port 3000, it should be added on containers details.

env:
  - name: PORT
  value: "3000"

References

Health vs Metrics endpoint

Health endpoint is designed to provide information about a readiness and liveness of the server. On the other hand, metrics endpoint provides information about the performance of the server such as load time, response time, memory usage, CPU usage etc.

Metrics List
cpu_user_time
cpu_user_utilization
cpu_system_time
cpu_system_utilization
event_loop_usage
event_loop_wait
memory_physical
memory_heap_used
memory_heap_free
memory_heap_max
memory_nonheap_used
http_request_duration_milliseconds
http_request_buckets_milliseconds
gc_pause_time_min
gc_pause_time_max
gc_pause_time_total
gc_scavenge_min
gc_scavenge_max
gc_scavenge_total
gc_marksweepcompact_min
gc_marksweepcompact_max
gc_marksweepcompact_total
gc_incrementalmarking_min
gc_incrementalmarking_max
gc_incrementalmarking_total