Before we look at the monitoring platform of GCP, the nodes of GKE cluster should admit being scanned for any applied status:

// add nodes with monitoring scope
$ gcloud container clusters create my-k8s-cluster --cluster-version 1.9.7-gke.0 - -machine-type f1-micro --num-nodes 3 --network k8s-network --zone us-central1-a - -tags private --scopes=storage-rw,compute-ro, https://www.googleapis.com/auth/monitoring

Google Stackdriver provides system monitoring in a hybrid cloud environment. Besides its own GCP, it can also cover your computing resources on AWS. To access its web console, you can find its section under the menu on the left-hand side. There are multiple service categories in Stackdriver. Select Monitoring to check related functionality.

As a new user, you will get a 30-day free trial. The initial configuration is simple; just enable an account and bind your project. You may avoid the agent installation and AWS account setup since we simply want to check the GKE cluster. Once you log in to Stackdriver successfully, click Resources on the left-side panel and choose Kubernetes Engine under infrastructure type:

There are several metrics set up already for computing resources from node to container. Take your time exploring and check the official introduction for more features: https://cloud.google.com/kubernetes-engine/docs/how-to/monitoring.