Explore navigation capabilities allow you to quickly locate the object that needs debugging in a physical stack. This page walks you through a high-level troubleshooting scenario to illustrate the possibilities.
Step 1: Analyzing the cluster
We suspect there's a problem with a Kubernetes cluster but aren't sure where, so we start by analyzing the Cluster Overview dashboard. Everything that is running on the cluster is shown on this dashboard. The Terminated and Waiting by Namespace panel allows us to easily comprehend the failure states the namespaces are in. Here we can easily see if there are configuration issues or overall administration issues that need to be addressed.
Step 2: Exploring a namespace
To further pinpoint the problem with our cluster, we investigated the namespace by selecting kube-system in the navigation panel and switching to the Namespace Overview dashboard. This dashboard provides information on pods running in the deployment, failed pods, errors, CPU and memory usage, file system usage, terminated and waiting pods and containers. In this example, we're focusing on the CPU and memory usage panels of the dashboard in our attempt to find out where our application is running into problems.
Step 3: Drilling down into a pod
Once we've determined which pod is having problems, we can drill down into the pod for more granular data. For example, you can select the Details icon for a panel to view that data in a search. Or, you can review the actual logs in the Log Stream panel.