[Kubernetes] Pod and Kubernetes Run

---

Pod and Kubernetes Run

• Pod
• Kubernetes Run
• Diagram for kubectl run

---

Pod

• A unique k8s concept
• Layer of abstraction
• That is, a resource type that wraps around one or more containers
  • that all share the same IP address
  • and the same deployment mechanism.
• all containers in a pod will deploy together on the same node, having access to each other over localhost and a few things(e.g. IP addresses)
• Unlike Docker, a container cannot be created directly, but can create pods(via CLI, YAML, or API), and k8s then created the container inside the pods
  • More specifically kubelet tells the container runtime(e.g. Docker, containerd, cri-o, etc.) to create the containers

Kubernetes Run

# kube version
kubectl version
kubectl version --short

# create a pod(kubectl run {pod_name} --image {image_name})
# k8s API receives the request and stored the entry(that my-nginx pod needs to be created) in its db(etcd)
kubectl run my-nginx --image nginx

# list the pods
kubectl get pods

# see all objects
# it doesn't list namely all objects, but the ones that a 'normal' developer will be interested in
kubectl get all

kubectl delete pod my-nginx

Diagram for kubectl run

• 1) request a pod to the control plane(api server + controller manager + etcd + more) == kubectl run my-nginx --image nginx
• 2) api server receives the request, store in the database as a pod request
• 3) the scheduler will assign it to a node
• 4) the kubelet agent, which has a persistent connection to the API, will see a new request for a pod on its node
• 5) the kubelet agent interprets how many containers to create
• 6) kubelet then tells the container runtime(e.g. docker, containerd, etc.)
• 7) container runtime creates the container

---

ref

• 🔗 kubectl Cheat Sheet
• 🔗 kubectl for Docker Users