Nodes are the machines in your cluster which actually run containers. Kubernetes stores information about nodes in its database and you can query it using the command line.
Kubectl is the command line to work with a Kubernetes cluster. It has commands to deploy applications and work with objects.
This is a simple lab to get you used to working with Kubectl, and to see some of the interesting things Kubernetes knows about the machines in the cluster.
Two of the most common Kubectl commands are get and describe.
You can use them with different objects, by adding the object type to the command:
kubectl get nodes
This shows all the servers in the cluster.
The get command prints a table with basic information. You use it to find out what objects exist.
To see the next level of detail for an object, use the describe command:
kubectl describe nodes
There's a lot more information to see and
describegives it to you in a readable format.
If you have several nodes in your cluster (which you will do for all production Kubernetes deployments), you can print the details for one node by adding its name to the decribe command:
# replace <node-name> with the name of one of your nodes:
kubectl describe node <node-name>
# e.g. on Docker Desktop:
# kubectl describe node docker-desktop
📋 Can you use the output to see how hard your node is working?
Kubernetes knows the compute capacity of each node - how many CPU cores and how much memory it has. That's shown in the output, along with the workloads currently running on the node and the amount of CPU and memory which has been allocated.
Kubectl has built-in help, you can use it to list all commands or list the details of one command:
kubectl --help
kubectl get --help
And you can learn about resources by asking Kubectl to explain them:
kubectl explain node
This doesn't tell you much, but it gives you a brief description of the resource. Nodes are just like any other resource in Kubernetes, their status is monitored and recorded using a fixed data schema.
You will spend a lot of time with Kubectl. You'll want to get familiar with some features early on, like querying.
Kubectl can print information in different formats, try showing your node details in JSON:
kubectl get node <node-name> -o json
# e.g. on Docker Desktop
# kubectl get node docker-desktop -o json
📋 Check the help to see what other output formats you can use.
kubectl --help gives you the top-level information, showing you the commands you can run.
To get the options for the get command run:
kubectl get --help
That shows the output formats available with -o (or --output), something like this:
-o, --output='': Output format. One of:
json|yaml|name|go-template|go-template-file|template|templatefile|jsonpath|jsonpath-as-json|jsonpath-file|custom-columns-file|custom-columns|wide
One output option is JSON Path, which is a query language you can use to print specific fields. This is great for automation, when you want to query some details in a script:
kubectl get node <your-node> -o jsonpath='{.status.capacity.cpu}'
# e.g.
# kubectl get node docker-desktop -o jsonpath='{.status.capacity.cpu}'
This tells you the number of CPU cores Kubernetes sees for that node.
📋 Can you run a similar command to show which container runtime your node is using?
You can show all the fields by using -o json and then work out the query to see the details you want:
kubectl get node <node-name> -o jsonpath='{.status.nodeInfo.containerRuntimeVersion}'
The output will show which version of Docker or containerd your node is running.
Every object in Kubernetes can have labels - they are key-value pairs used to record additional information about the object.
Nodes have a set of labels which are provisioned by the environment, including some standard ones which all Kubernetes platforms provide.
Use Kubectl to find labels for your node, which will show you (among other things) the CPU architecture and operating system it's using.