The Wavefront Pivotal Container Service integration includes an overview and setup instructions.
On this page, we list predefined alerts and give other details about the integration.
Predefined Alerts for the PKS Integration
The PKS integration provides the following monitoring alerts for PKS. These alerts are predefined, no additional setup is required. You can create additional alerts from the Alerts browser or from charts.
|Name||Severity||Resolve After (min)|
|Node Memory Usage high||WARN||10|
|Node Memory Usage too high||SEVERE||10|
|Node CPU Usage high||WARN||5|
|Node CPU Usage too high||SEVERE||5|
|Node Storage Usage high||WARN||10|
|Node Storage Usage too high||SEVERE||10|
|Too many Pods crashing||SEVERE||5|
|Too many Containers not running||SEVERE||5|
You can create additional alerts using the Wavefront UI.
PKS Monitoring Dashboards
Wavefront includes several predefined dashboards for monitoring PKS. You can use these dashboards as is, or clone and customize them.
We support dashboards for major functionality including:
- PKS Home dashboard
- Nodes dashboard
- Namespaces dashboard
- Deployments dashboard
- Pods dashbord
- Pod containers dashboard
- Services and Replication Sets dashboard
The screenshots below show some examples. To see the full set of dashboards, sign in your Wavefront instance – or sign up for a free trial.
Pods Containers Dashboard
The Wavefront proxy pod includes four containers.
- Heapster monitors your worker kubelets and sends the result to the proxy.
- Telegraf receives metrics about the node, pod, and container status from kube-state-metrics, and sends those metrics to the Wavefront proxy as well.
Wavefront runs a Wavefront proxy pod inside each PKS-created Kubernetes cluster.
There are four containers within the Wavefront proxy pod:
Troubleshooting the PKS Integration
401 Unauthorized Error
If PKS tile deployment fails at running the
wavefront-alert-creation errand with
401 Unauthorized, the
wavefront-access-token is invalid.
No such host/no route to host Error
If you receive a no
such host/no route to host error, check that the wavefront-api-server can connect to the Internet, as follows:
curl -s 'https://vmware.wavefront.com/api/v2/source‘ \ -H 'Authorization: Bearer 1d23d456-XXXX-XXXX-XXXX-f123f12b1c21'| jq .
If you see no metrics in the Wavefront dashboard, possible causes are:
- 401 unauthorized: check wavefront-access-token validity
- Network connectivity to wavefront-api-server
Check the status of the wavefront-proxy pod:
kubectl get pods --all-namespaces kubectl describe pod wavefront-proxy-pod-name -n kube-system
Check wavefront-proxy pod logs:
kubectl logs wavefront-proxy-pod-name -n kube-system -c wavefront-proxy kubectl logs wavefront-proxy-pod-name -n kube-system -c heapster kubectl logs wavefront-proxy-pod-name -n kube-system -c kube-state-metrics