Monitoring Application

If your application has exposed metrics, you can easily tell the platform how to collect the metrics data from your app with metrics capability.

Let’s run christianhxc/gorandom:1.0 as an example app. The app will emit random latencies as metrics.

  1. Prepare Appfile:
  1.   $ cat <<EOF > vela.yaml
  2.     name: metricapp
  3.     services:
  4.       metricapp:
  5.         type: webservice
  6.         image: christianhxc/gorandom:1.0
  7.         port: 8080
  9.         metrics:
  10.           enabled: true
  11.           format: prometheus
  12.           path: /metrics
  13.           port: 0
  14.           scheme: http
  15.   EOF
  1. Deploy the application:
  1. $ vela up
  1. Check status:
  1.   $ vela status metricapp
  2.   About:
  4.     Name:          metricapp
  5.     Namespace:     default
  6.     Created at:    2020-11-11 17:00:59.436347573 -0800 PST
  7.     Updated at:    2020-11-11 17:01:06.511064661 -0800 PST
  9.   Services:
  11.     - Name: metricapp
  12.       Type: webservice
  13.       HEALTHY Ready: 1/1
  14.       Traits:
  15.         - ✅ metrics: Monitoring port: 8080, path: /metrics, format: prometheus, schema: http.
  16.       Last Deployment:
  17.         Created at: 2020-11-11 17:00:59 -0800 PST
  18.         Updated at: 2020-11-11T17:01:06-08:00

The metrics trait will automatically discover port and label to monitor if no parameters specified. If more than one ports found, it will choose the first one by default.

(Optional) Verify that the metrics are collected on Prometheus

Expose the port of Prometheus dashboard:

  1. kubectl --namespace monitoring port-forward `kubectl -n monitoring get pods -l prometheus=oam -o name` 9090

Then access the Prometheus dashboard via http://localhost:9090/targets

Prometheus Dashboard