Installation

npm i adonis5-prometheus
node ace configure adonis5-prometheus

Usage

A configuration file has been added in config/prometheus.ts.

By default the system metrics are collected ( systemMetrics.enabled: true ), so now you can call the endpoint {{host}}/metrics to get the measured metrics.

Here is an example scrape_config to add to prometheus.yml:

scrape_configs:
  - job_name: my-adonis-app
    static_configs:
      - targets: [my-adonis-app.com]
    scrape_interval: 5s

Built-in Metrics

Metrics collected by Adonis5-prometheus middleware | Type | Name | Description | | --- | --- | --- | | Histogram | adonis_http_request_durations | Total time each HTTP requests takes. | | Gauge | adonis_uptime_metrics | Uptime performance of the application (1 = up, 0 = down) | | Counter | adonis_throughput_metrics | No. of request handled. |

To enable them, simply register the CollectPerformanceMetrics as the first item in the start/kernel.ts:

Server.middleware.register([
  // Make it first in the list for reliable metrics.
  () => import('@ioc:Adonis/Prometheus/Middlewares/CollectPerformanceMetrics'),
  () => import('@ioc:Adonis/Core/BodyParser'),
  ...
])

Verify if the metrics are enabled in the config/prometheus.ts file. You can also configure the metrics there.

Custom Metrics

// Register your custom metrics in the separate file you want.
export const OrderMetric = new Prometheus.Counter({
  name: 'sent_orders',
  help: 'Total Orders Sent',
})

// OrderController.ts
import { OrderMetric } from 'App/Metrics'

export default class OrderController {
  public async store({ request }: HttpContextContract) {
    const order = await request.validate({ schema: OrderSchema })

    // ...
    OrderMetric.inc()
    // ...
  }
}

When hitting {{host}}/metrics you will now get the following:

# HELP send_orders Total Orders Sent
# TYPE send_orders counter
sent_orders 2

Grafana Dashboard

A basic ready to use dashboard is available in the grafana folder. Grafana 8 is required to be able to import the dashboard correctly, otherwise you'll have to tinker a bit to make it work. It includes :

• Process CPU usage
• Event loop lag
• Node.JS version
• Requests by second
• Request volume rate by URL
• Average response time
• Response error rate by URL

To be fully functional, you need to enable systemMetrics, httpMetric and throughputMetric in the config/prometheus.ts file.

Documentation

This library is a wrapper for prom-client. The prom-client object can be imported with import Prometheus from '@ioc:Adonis/Prometheus'. Check out the documentation for more information.