Collecting Metrics

This task shows how to configure Istio to automatically gather telemetry for services in a mesh. At the end of this task, a new metric will be enabled for calls to services within your mesh.

The Bookinfo sample application is used as the example application throughout this task.

Before you begin

  • Install Istio in your cluster and deploy an application. This task assumes that Mixer is setup in a default configuration (--configDefaultNamespace=istio-system). If you use a different value, update the configuration and commands in this task to match the value.

Collecting new metrics

  1. Create a new YAML file to hold configuration for the new metric and log stream that Istio will generate and collect automatically.

    Save the following as new_metrics.yaml:

    # Configuration for metric instances
    apiVersion: ""
    kind: metric
      name: doublerequestcount
      namespace: istio-system
      value: "2" # count each request twice
        reporter: conditional((context.reporter.kind | "inbound") == "outbound", "client", "server")
        source: | "unknown"
        destination: | "unknown"
        message: '"twice the fun!"'
      monitored_resource_type: '"UNSPECIFIED"'
    # Configuration for a Prometheus handler
    apiVersion: ""
    kind: prometheus
      name: doublehandler
      namespace: istio-system
      - name: double_request_count # Prometheus metric name
        instance_name: doublerequestcount.metric.istio-system # Mixer instance name (fully-qualified)
        kind: COUNTER
        - reporter
        - source
        - destination
        - message
    # Rule to send metric instances to a Prometheus handler
    apiVersion: ""
    kind: rule
      name: doubleprom
      namespace: istio-system
      - handler: doublehandler.prometheus
        - doublerequestcount.metric
  2. Push the new configuration.

    $ kubectl apply -f new_metrics.yaml
    Created configuration metric/istio-system/doublerequestcount at revision 1973035
    Created configuration prometheus/istio-system/doublehandler at revision 1973036
    Created configuration rule/istio-system/doubleprom at revision 1973037
  3. Send traffic to the sample application.

    For the Bookinfo sample, visit http://$GATEWAY_URL/productpage in your web browser or issue the following command:

    $ curl http://$GATEWAY_URL/productpage
  4. Verify that the new metric values are being generated and collected.

    In a Kubernetes environment, setup port-forwarding for Prometheus by executing the following command:

    $ kubectl -n istio-system port-forward $(kubectl -n istio-system get pod -l app=prometheus -o jsonpath='{.items[0]}') 9090:9090 &

    View values for the new metric via the Prometheus UI.

    The provided link opens the Prometheus UI and executes a query for values of the istio_double_request_count metric. The table displayed in the Console tab includes entries similar to:

    istio_double_request_count{destination="details-v1",instance="",job="istio-mesh",message="twice the fun!",reporter="client",source="productpage-v1"}   8
    istio_double_request_count{destination="details-v1",instance="",job="istio-mesh",message="twice the fun!",reporter="server",source="productpage-v1"}   8
    istio_double_request_count{destination="istio-policy",instance="",job="istio-mesh",message="twice the fun!",reporter="server",source="details-v1"}   4
    istio_double_request_count{destination="istio-policy",instance="",job="istio-mesh",message="twice the fun!",reporter="server",source="istio-ingressgateway"}   4

    For more on querying Prometheus for metric values, see the Querying Istio Metrics task.

Understanding the metrics configuration

In this task, you added Istio configuration that instructed Mixer to automatically generate and report a new metric for all traffic within the mesh.

The added configuration controlled three pieces of Mixer functionality:

  1. Generation of instances (in this example, metric values) from Istio attributes

  2. Creation of handlers (configured Mixer adapters) capable of processing generated instances

  3. Dispatch of instances to handlers according to a set of rules

The metrics configuration directs Mixer to send metric values to Prometheus. It uses three stanzas (or blocks) of configuration: instance configuration, handler configuration, and rule configuration.

The kind: metric stanza of configuration defines a schema for generated metric values (or instances) for a new metric named doublerequestcount. This instance configuration tells Mixer how to generate metric values for any given request, based on the attributes reported by Envoy (and generated by Mixer itself).

For each instance of doublerequestcount.metric, the configuration directs Mixer to supply a value of 2 for the instance. Because Istio generates an instance for each request, this means that this metric records a value equal to twice the total number of requests received.

A set of dimensions are specified for each doublerequestcount.metric instance. Dimensions provide a way to slice, aggregate, and analyze metric data according to different needs and directions of inquiry. For instance, it may be desirable to only consider requests for a certain destination service when troubleshooting application behavior.

The configuration instructs Mixer to populate values for these dimensions based on attribute values and literal values. For instance, for the source dimension, the new configuration requests that the value be taken from the attribute. If that attribute value is not populated, the rule instructs Mixer to use a default value of "unknown". For the message dimension, a literal value of "twice the fun!" will be used for all instances.

The kind: prometheus stanza of configuration defines a handler named doublehandler. The handler spec configures how the Prometheus adapter code translates received metric instances into Prometheus-formatted values that can be processed by a Prometheus backend. This configuration specified a new Prometheus metric named double_request_count. The Prometheus adapter prepends the istio_ namespace to all metric names, therefore this metric will show up in Prometheus as istio_double_request_count. The metric has three labels matching the dimensions configured for doublerequestcount.metric instances.

For kind: prometheus handlers, Mixer instances are matched to Prometheus metrics via the instance_name parameter. The instance_name values must be the fully-qualified name for Mixer instances (example: doublerequestcount.metric.istio-system).

The kind: rule stanza of configuration defines a new rule named doubleprom. The rule directs Mixer to send all doublerequestcount.metric instances to the doublehandler.prometheus handler. Because there is no match clause in the rule, and because the rule is in the configured default configuration namespace (istio-system), the rule is executed for all requests in the mesh.


  • Remove the new metrics configuration:

    $ kubectl delete -f new_metrics.yaml
  • Remove any kubectl port-forward processes that may still be running:

    $ killall kubectl
  • If you are not planning to explore any follow-on tasks, refer to the Bookinfo cleanup instructions to shutdown the application.