Collecting Metrics for TCP services

This task shows how to configure Istio to automatically gather telemetry for TCP services in a mesh. At the end of this task, a new metric will be enabled for calls to a TCP service 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 the BookInfo sample will be deployed in the default namespace. If you use a different namespace, you will need to update the example configuration and commands.

  • Install the Prometheus add-on. Prometheus will be used to verify task success.

    kubectl apply -f install/kubernetes/addons/prometheus.yaml
    

    See Prometheus for details.

Collecting new telemetry data

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

    Save the following as tcp_telemetry.yaml:

     # Configuration for a metric measuring bytes sent from a server
     # to a client
     apiVersion: "config.istio.io/v1alpha2"
     kind: metric
     metadata:
     name: mongosentbytes
     namespace: default
     spec:
       value: connection.sent.bytes | 0 # uses a TCP-specific attribute
       dimensions:
         source_service: source.service | "unknown"
         source_version: source.labels["version"] | "unknown"
         destination_version: destination.labels["version"] | "unknown"
       monitoredResourceType: '"UNSPECIFIED"'
     ---
     # Configuration for a metric measuring bytes sent from a client
     # to a server
     apiVersion: "config.istio.io/v1alpha2"
     kind: metric
     metadata:
     name: mongoreceivedbytes
     namespace: default
     spec:
       value: connection.received.bytes | 0 # uses a TCP-specific attribute
       dimensions:
         source_service: source.service | "unknown"
         source_version: source.labels["version"] | "unknown"
         destination_version: destination.labels["version"] | "unknown"
       monitoredResourceType: '"UNSPECIFIED"'
     ---
     # Configuration for a Prometheus handler
     apiVersion: "config.istio.io/v1alpha2"
     kind: prometheus
     metadata:
     name: mongohandler
     namespace: default
     spec:
       metrics:
       - name: mongo_sent_bytes # Prometheus metric name
         instance_name: mongosentbytes.metric.default # Mixer instance name (fully-qualified)
         kind: COUNTER
         label_names:
         - source_service
         - source_version
         - destination_version
       - name: mongo_sent_bytes # Prometheus metric name
         instance_name: mongosentbytes.metric.default # Mixer instance name (fully-qualified)
         kind: COUNTER
         label_names:
         - source_service
         - source_version
         - destination_version
     ---
     # Rule to send metric instances to a Prometheus handler
     apiVersion: "config.istio.io/v1alpha2"
     kind: rule
     metadata:
     name: mongoprom
     namespace: default
     spec:
       match: context.protocol == "tcp"
              && destination.service = "mongodb.default.svc.cluster.local"
       actions:
       - handler: mongohandler.prometheus
         instances:
         - mongoreceivedbytes.metric
         - mongosentbytes.metric
    
  2. Push the new configuration.

    istioctl create -f tcp_telemetry.yaml
    

    The expected output is similar to:

    Created config metric/default/mongosentbytes at revision 3852843
    Created config metric/default/mongoreceivedbytes at revision 3852844
    Created config prometheus/default/mongohandler at revision 3852845
    Created config rule/default/mongoprom at revision 3852846
    
  3. Setup BookInfo to use MongoDB.

    1. Install v2 of the ratings service:

      kubectl apply -f <(istioctl kube-inject -f samples/bookinfo/kube/bookinfo-ratings-v2.yaml)
      

      Expected output:

      deployment "ratings-v2" configured
      
    2. Install the mongodb service:

      kubectl apply -f <(istioctl kube-inject -f samples/bookinfo/kube/bookinfo-db.yaml)
      

      Expected output:

      service "mongodb" configured
      deployment "mongodb-v1" configured
      
    3. Add routing rules to send traffic to v2 of the ratings service:

      kubectl apply -f samples/bookinfo/kube/route-rule-ratings-db.yaml
      

      Expected output:

      routerule "ratings-test-v2" created
      routerule "reviews-test-ratings-v2" created
      
  4. 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
    
  5. 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].metadata.name}') 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 mongo_received_bytes metric. The table displayed in the Console tab includes entries similar to:

    mongo_received_bytes{destination_version="v1",instance="istio-mixer.istio-system:42422",job="istio-mesh",source_service="ratings.default.svc.cluster.local",source_version="v2"}	2317
    

    NOTE: Istio also collects protocol-specific statistics for MongoDB. For example, the value of total OP_QUERY messages sent from the ratings service is collected in the following metric: envoy_mongo_mongo_collection_ratings_query_total_counter (click here to execute the query).

Understanding TCP telemetry collection

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

Similar to the Collecting Metrics and Logs Task, the new configuration consisted of instances, a handler, and a rule. Please see that Task for a complete description of the components of metric collection.

Metrics collection for TCP services differs only in the limited set of attributes that are available for use in instances.

TCP Attributes

Several TCP-specific attributes enable TCP policy and control within Istio. These attributes are generated by server-side Envoy proxies and forwarded to Mixer at both connection establishment and connection close. Additionally, context attributes provide the ability to distinguish between http and tcp protocols within policies.

Attribute Generation Flow for TCP Services in an Istio Mesh.
TCP Attribute Flow

Cleanup

  • Remove the new telemetry configuration:

    istioctl delete -f tcp_telemetry.yaml
    
  • If you are not planning to explore any follow-on tasks, refer to the BookInfo cleanup instructions to shutdown the application.

Further reading