Kubernetes can function with master and slave nodes.

Nodes

A node is a Kubernetes worker machine that may be physical or virtual.

Nodes are managed by the control plane.

Every node must have 3 processes installed, being:

• Container runtime (Docker, Rocket, containerd, etc)
  • Downloads images and runs the containers.
• Kubelet
  • Communicates with the Kubernetes API Server.
  • It's a node process that interacts with the container and the node.
  • Kubelet is responsible for starting the pod and container using the defined container runtime and also ensure that they are running.
  • Reports health and status to the control plane.
• Kube proxy
  • It's a network proxy that forwards the requests and also can apply defined rules.
  • When a container makes a request to another service the kube proxy is responsible to forward the request to an available service.

Master node (Control Plane)

There can be multiple master nodes, with the api-server being load balanced.

Every master node needs to have 4 processes, being:

• Api server (kube-apiserver)

  • It is a cluster gateway
  • Also functions as a gatekeeper for authentication

• Scheduler

  • Assigns pods to nodes according to constraints and available resources on each node in the cluster.
  • After the scheduler chooses the most suitable node the given node forwards the execution to it's own kubelet.

• kube-controller-manager

  • Runs controller processes like:
    • Node controller: Responsible for noticing and responding when nodes go down.
    • Job controller: Watches for Job objects that represent one-off tasks, then creates Pods to run those tasks to completion.
    • Endpoints controller: Populates the Endpoints object (that is, joins Services & Pods).
    • Service Account & Token controllers: Create default accounts and API access tokens for new namespaces.

• etcd

  • It's the "cluster brain"
  • Key value store for all cluster data.
  • All cluster changes are stored there.

  <aside> ❗ The etcd does not store application data, only data relative to the cluster is being stored there.

  </aside>

Example Cluster Set-Up

Layers of abstraction

• Deployment manages a ReplicaSet which manages a Pod which is an abstraction of a container.
• Everything below a Deployment is usually managed automatically by Kubernetes.