Installing Stackdriver monitoring support enables you to easily monitor GKE clusters, debug logs, and analyze your cluster performance using advanced profiling and tracing capabilities. In this recipe, we will enable Stackdriver Kubernetes Engine Monitoring support to collect cluster metrics from our GKE cluster:
- After following the Configuring a workspace on Stackdriver recipe, open the Stackdriver console at https://app.google.stackdriver.com:
- From the Resources menu, click on the Kubernetes Engine option:
- The Kubernetes Engine view will show the list of clusters that are Stackdriver Kubernetes Engine Monitoring-enabled. In our example, you can see that we have one cluster available:
- On the Infrastructure tab, click on the expand icon next to the cluster name. Stackdriver will expand the list with the individual worker nodes. In the Ready column, you can see the number of pods deployed and in a ready state in each node. In the CPU Utilization column, the value on the left-hand side shows the total available CPUs and the right-hand value shows the current utilization percentage. Similarly, in the Memory Utilization column, the value on the left-hand side shows the total available memory (GiB), and the right-hand value shows the current utilization percentage:
- Click on the expand icon next to a node name, and the list will expand to display the pods deployed on that specific node:
- Click on one of the pods on your cluster. Stackdriver will show a detailed view of pod metrics, including pod restarts, CPU, memory, storage, and network utilization for the pods. In our example, we can see metrics for the Prometheus pod:
- Click on the Logs tab to switch to the log summary view. This view will only show the most recent logs:
- Click on the Go to console button to open a detailed log view where you can see older logs and use filters to create metrics:
Now you know how to use Stackdriver to monitor health, performance metrics, and logs for GKE clusters and resources deployed on the GKE clusters.