Network Explorer isn’t showing metrics as expected

When Network Explorer experiences data issues, you might see the following:

• Very limited or no TCP metrics and HTTP metrics.

• No workloads, or only workloads for containerd and kubelet.

Causes

Your kubelet and containerd cgroup drivers might be not be the same value.

Solution

Check the cgroup drivers of your kubelet and containerd to make sure that they match. Both of them have to be either cgroupfs or systemd. For more information, see the Kubernetes documentation on cgroup drivers.

You don’t see Kubernetes metadata in Network Explorer metrics

Your Network Explorer metrics are not generated with Kubernetes metadata.

Causes

• Your Kubernetes collector is turned off.

  The Kubernetes collector is turned off if you see the following:

  k8sCollector:
    enabled: false

• If your Kubernetes collector is turned on, you can determine the root cause based on the logs for the k8s-watcher and k8s-relay containers in the k8s-collector pod.

  1. Run the following command to find your k8s-collector pod name.

     kubectl get pods

  2. Run the following commands to retrieve the logs for the containers.

     kubectl logs <POD_NAME> -c k8s-watcher
     kubectl logs <POD_NAME> -c k8s-relay

     This is an example error message from the logs. In this case, the service account token is not available.

     Error: open /var/run/secrets/kubernetes.io/serviceaccount/token: no such file or directory

  Note: On initial startup, the k8s-watcher tries to connect to the k8s-relay. If the k8s-relay has not yet come up, you might see the following set of error messages.

  Error: rpc error: code = Unavailable desc = connection error: desc = "transport: Error while dialing dial tcp [::1]:8712: connect: connection refused"

  These messages are expected and you can ignore them.

Solution

• If your Kubernetes collector is turned off, you need to turn it on by setting k8sCollector.enabled to true in the Network Explorer values file.

  k8sCollector:
      enabled: true

• For other errors, see the Kubernetes documentation for more information on configuring the service account for the pod to turn on communication with the API server.