How does a pod get created in Kubernetes?

The best way to develop an understanding of the Kubernetes Architecture is by seeing how each component behaves when an action is performed on the cluster.

Even the simplest task such as - 'Creation of a pod' involves a lot of steps.

Understanding how this works would provide much-needed clarity about the functions of each component.

To fully understand the below steps, knowledge of different Kubernetes components is required.

Step 1

• User issues a kubectl command to create a pod.
• Kubectl passes this command to api-server.
• Api-server validates this command & acts on it.

Step 2

• Api-server passes the information to etcd.
• Etcd creates a record for this new pod. (which is yet to be created right now)

Step 3

• The scheduler takes cognizance of the requirement to create a new pod.
• It acquires details such as memory, CPU requirements of the pod from api-server.
• It then returns a worker node where the pod can be scheduled to the api-server.

Step 4

• Api-server sends the update to the etcd.
• Etcd updates the worker node name for the pod in its key-value store.

An important observation here is that - out of all the components, only api-server communicates with the etcd directly.

Step 5

• Api-server passes the pod specifications to the kubelet on the worker node.
• Kubelet provisions the pod & runs the pod on the node's container runtime engine.
• Kubelet sends updates about the pod's status to the api-server.
• The api-server sends the updated status to the etcd.

Step 6

• Once the pod gets into a running state, the api-server returns the status to the user after it's updated on the etcd.

And that's how each component performs its function to create a new pod.

I hope this added to your understanding of the Kubernetes.

If you found this useful, share it with your network!