Difference between 'kubectl apply' and 'kubectl create'

Question

I created a pod with kubectl create -f pod.xml and kubectl apply -f pod.xml using the below yaml and I don't see any difference, a pod gets created with both the commands. The K8S document, mentions imperative and declarative commands. But, still the create and apply behave the same way.

apiVersion: v1
kind: Pod
metadata:
  name: myapp-pod
  labels:
    app: myapp
spec:
  containers:
  - name: myapp-container
    image: busybox
    command: ['sh', '-c', 'echo Hello Kubernetes! && sleep 3600']

What's the difference? Also, how is kubectl apply declarative and kubectl create imperative? Both of them take one or multiple yaml files with the object details in it.

Answer 1

There is a subtle difference between kubectl create and kubectl apply commands.

The kubectl create command creates a new resource. So, if the command is run again it will throw an error as resource names should be unique in a namespace.

kubectl get pods
No resources found.

kubectl create -f pod.xml 
pod/myapp-pod created

kubectl create -f pod.xml 
Error from server (AlreadyExists): error when creating "pod.xml": pods "myapp-pod" already exists

2) The kubectl apply command applies the configuration to a resource. If the resource is not there then it will be created. The kubectl apply command can be run the second time as it simply applies the configuration as shown below. In this case, the configuration hasn't changed. So, the pod hasn't changed.

kubectl delete pod/myapp-pod
pod "myapp-pod" deleted

kubectl apply -f pod.xml 
pod/myapp-pod created

kubectl apply -f pod.xml 
pod/myapp-pod unchanged

In the kubectl create, we specify a certain action, in this case create and so it is imperative. In the kubectl apply command we specify the target state of the system and don't specify a certain action and so declarative. We let the system decide what action to take. If the resource is not there it will create it, if the resource is there then it will apply the configuration to the existing resource.

From an execution perspective, there is no difference when a resource is created for the first time between kubectl create and kubectl apply as shown above. But, the second time the kubectl create will throw an error.

It took me some time to get around it, but it makes sense now.

Answer 2

In plain terms, create and apply are essentially the same if you run the operation on a single file to create resources. However, apply lets you create and patch at the same time on multiple files under a directory.

There's also an apply to remove resources from a directory but it's in alpha as of this writing:

kubectl apply -f <directory/> --prune -l your=label)

There are more insights on this question too: Kubectl apply vs kubectl create?

Answer 3

Those are two different approaches. kubectl create is what we call Imperative Management. On this approach you tell the Kubernetes API what you want to create, replace or delete, not how you want your K8s cluster world to look like.

kubectl apply is part of the Declarative Management approach, where changes that you may have applied to a live object (i.e. through scale) are maintained even if you apply other changes to the object.

You can read more about imperative and declarative management in the Kubernetes Object Management documentation.