Metrics

Before Consul Release Controller increases the traffic to your Candidate deployment it first checks the health of your application by looking at the traffic metrics for the application. By default Consul Release Controller provides default queries for each of the supported platforms.

Default Queries

Prometheus / Kubernetes

EnvoyRequestSuccess

This query measures the HTTP response codes emitted from Envoy for your application and returns the percentage of requests (0-100) that do not result in a HTTP 5xx response.

sum(
    rate(
    envoy_cluster_upstream_rq{
      namespace="{{ .Namespace }}",
      pod=~"{{ .Name }}-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)",
      envoy_cluster_name="local_app",
      envoy_response_code!~"5.*"
    }[{{ .Interval }}]
  )
)
/
sum(
  rate(
    envoy_cluster_upstream_rq{
      namespace="{{ .Namespace }}",
      envoy_cluster_name="local_app",
      pod=~"{{ .Name }}-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"
    }[{{ .Interval }}]
  )
)
* 100

EnvoyRequestDuration

This query measures the 99 percentile duration for application requests in milliseconds.

histogram_quantile(
  0.99,
  sum(
    rate(
      envoy_cluster_upstream_rq_time_bucket{
        namespace="{{ .Namespace }}",
        envoy_cluster_name="local_app",
        pod=~"{{ .Name }}-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"
      }[{{ .Interval }}]
    )
  ) by (le)
)

Custom Queries

Custom queries can be defined by specifying the optional query parameter instead of the preset parameter.

Queries are specified as Prometheus queries and must return a numeric value that can be evaluated with the min, max criteria. To enable generic queries Go templates can be used to inject values such as the Name of the deployment or the Namespace where the deployment is running.

monitor:
  pluginName: "prometheus"
  config:
    address: "http://prometheus-kube-prometheus-prometheus.monitoring.svc:9090"
    queries:
      - name: "mycustom"
        min: 20
        max: 200
        query: |
          histogram_quantile(
            0.99,
            sum(
              rate(
                envoy_cluster_upstream_rq_time_bucket{
                  namespace="{{ .Namespace }}",
                  envoy_cluster_name="local_app",
                  pod=~"{{ .Name }}-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"
                }[{{ .Interval }}]
              )
            ) by (le)
          )

Parameters

Parameter	Type	Description
Name	string	Name of the candidate deployment
Namespace	string	Namespace where the candidate is running
Interval	duration	Interval from the Strategy config, specified as a prometheus duration (30s, etc)