Skip to content

Latest commit

 

History

History
181 lines (120 loc) · 6.16 KB

Section2-CoreConcepts.md

File metadata and controls

181 lines (120 loc) · 6.16 KB
Raw

Section 2: Core Concepts

9: Recap - Kubernetes Architecture

Node

  • Machine - Physical or virtual on which k8s is installed.
  • Worker machine where containers will be launched by k8s.
  • Previously referred to as Minions.
  • Need to have more than one node. If your only node fails, our app goes down.

Master Node

  • Type of k8s node which is configured as "Master".
  • Watches over the nodes in cluster.
  • Responsible for orchestration of nodes.
  • Has the kube-apiserver, which makes it the master.
  • Also has the controller/control manager and scheduler.

Worker Node

  • Type of k8s node which is controlled by the master node.
  • Has the kubelet agent, which makes it the worker.

Cluster

  • Number of nodes grouped together.

10: Recap - Pods

Pods

  • Kubernetes does not deploy containers directly on the worker nodes.
  • Encapsulated within a pod - which is a single instance of an application.
  • Smallest object you can create in Kubernetes.
  • We don't spin up a new instance of the app within an existing pod.
    • We will create a new pod entirely and bring up our new app instance in it.
  • Multiple pods can be within the same node.
  • Pods have a 1:1 relationship with containers running your application.

Multi-container Pods

  • We can have multiple containers within the same pod.
    • However, these containers aren't the same.
    • One container will be our app, the other container will be a "helper" container.
  • All containers within a pod will be killed if a pod goes down.
  • This is a rare use case.

12: Recap - Pods with YAML

YAML Basics

  • YAML denotes nesting through indentations.
  • Adding a hyphen - before a line indicates that the line is an element in a list/array.

21: Recap - ReplicaSets

Replica Set allows us to have High Availability. It is also used for Load Balancing and Scaling.

ReplicaSet MUST have a selector definition (required). selector definition helps identify which pods fall under a replica set. This is important because ReplicaSet can also manage pods which were not created as part of the replica set creation. Replica set uses labels and selectors to determine which pods to monitor.

ReplicationController does not need this (optional).

24: Recap - Deployments

Pod deploys a single instance of our application. Container is encapsulated in a pod. Multiple pods are deployed using replication controller/replica set. Deployment is a Kubernetes object that comes higher in this hierarchy.

Deployment provides us the capability to upgrade underlying instances seamlessly. Deployment YAML file is similar to the Replica Set definition file. Only the kind needs to be changed to Deployment.

30: Recap - Namespaces

  • default namespace is created automatically by Kubernetes when the cluster is first setup.
  • kube-system is another namespace created at cluster startup. It contains a set of pods and services for internal purpose. This is isolated from users to prevent accidental changes/deletions.
  • kube-public contains resources that should be made available to all users.

Limit resources in a namespace by creating a resource quota.

Noteworthy Commands

kubectl get pods will give us a list of pods currently available.

kubectl describe pod <pod-name> will give us more details about a particular pod.

kubectl run nginx deploys a docker container by creating a pod. First creates a pod automatically and deploys an instance of the nginx docker image.

kubectl get pods -o wide gives us slightly more information about the pods.

kubectl get pods --namespace=<ns> gives us pods from the specified namespace.

kubectl get pods --all-namespaces lists pods from all namespaces.

kubectl delete pod <pod-name> deletes the pod with name <pod-name>.

kubectl create -f <filename> creates kind specified in the <filename>.

kubectl create -f <filename> --namespace=<ns> creates kind specified in the <filename> in the <ns> namespace.

kubectl apply -f <filename> updates existing kind and applies changes.

kubectl edit pod <pod-name> edits <pod-name> config. Changes are applied on file save/exit.

kubectl get pod <pod-name> -o yaml > pod-definition.yaml extracts definition. <pod-name> pod YAML definition will be extracted into pod-definition.yaml.

kubectl get replicaset gives us a list of replica sets currently available.

kubectl delete replicaset myapp-replicaset deletes the replica set. It will also delete all underlying Pods.

kubectl replace -f <filename> will replace/update kind specified in filename.

kubectl scale --replicas=<number> -f <filename> will scale kind specified in filename. This will scale the kind from the command line without changing the contents of filename.

kubectl get deployments shows us the newly created deployment.

kubectl get all show us all the created objects.

Using --dry-run tells us whether the resource can be created. It will also tell us if our command is right.

kubectl create namespace dev creates a new namespace called dev.

kubectl config set-context $(kubectl config current context) --namespace=<ns> changes the current namespace to <ns>.

Imperative Commands

kubectl run nginx --image=nginx --dry-run -o yaml generates pod manifest YAML. Doesn't create the pod, only dry run.

kubectl create deployment --image=nginx nginx --dry-run=client -o yaml generates deployment YAML file. Doesn't create the deployment, only dry run. kubectl create deployment does not have a --replicas option. We can first create it and then scale it using the kubectl scale command. Or we can save the deployment YAML to a file and then update the YAML file with the replicas or any other field before creating the deployment.

kubectl expose pod redis --port=6379 --name redis-service --dry-run=client -o yaml creates a service named redis-service of type ClusterIP to expose pod redis on port 6379. This will automatically use the pod's labels as selectors.

OR

kubectl create service clusterip redis --tcp=6379:6379 --dry-run=client -o yaml will not use the pods labels as selectors. Instead it will assume selectors as app=redis. However, we cannot pass in selectors as an option.