Debug School: Ankit Aggarwal

List of resources in K8s

Ankit Aggarwal — Fri, 11 Nov 2022 05:54:00 +0000

Namespaces: Namespace is something like an isolated environment in cluster keep resource within it like PODs, PVC and few others. By default we have default namespace created kubectl get ns
POD: Pod is atomic unit of k8s where we run one or more container which actually runs application within it kubectl get pod -n=yournamespace
ReplicationController: this resource will help us create replica of a POD. kubectl get rs
Deployments: This will help us to deploy pod with more features replication+controller+versioning+rollingoutupdate + rollback.

kubectl get deploy -n=namespace
PV: it is referred as persistent volume in K8S which is usually created by admin. It is a permanent volume which is not deleted even if pod is deleted
PVC: it is referred as persistent volumne claim which can be claimed by user or admin and then attached to pod. It is for claiming persistent volume available
Secrets: It is a kind of volume in etcd within cluster storage to store meta or config or data or file in encoded form. It will be in key value pair format
ConfigMap: It is kind of volume in etcd which stores information which is available to all the Nodes. It will be in key value form format. It can also be used for storing environment variables
Service account: These are the sa for process runs in PODs
Daemonsets:It also creates a pod. It is mainly for stateless application like application which never changes like apache. It is mainly used by admin where one pod each worker kind of requirement is there. like prometheous or monitoring or logging etc
Stateful sets: As the name suggests it is for creating PODS in stateful manner. It also create pods sequentially like Pod1, Pod2, Pod2 1 by 1.
role: It is there for authorization purpose so that role based access can be given to user.
rolebindings: It will do user, role and namespace binding for authorization of user.
Clusterroles: It is a cluster level roles created
Clusterrolesbinding: It will binding users to cluster level roles so that user will have cluster level access to resources
jobs: These are process which runs at one for specific purpose
cronjobs: These are process which runs recurrently or repeatedly
services: There are like a internal load balancers for pods so that if requests can be redirected to pods randomly

What is POD

Ankit Aggarwal — Wed, 09 Nov 2022 07:56:09 +0000

Pod is not a physical entity. It just like isolate containers.
Pod is like handlers for resources like CPU, RAM, Container etc
Pod is wrapper component where container resides
Pod can never be restarted or start/stop.
Pod can always be instantiated
If container inside the pod stop/failed Pod will be not be available
Pod has it's own network and uses POD network to communicate to each other.
We can have multiple containers inside POD and all will have same ip address
Life Cycle of POD is pending, running , success/failed
Anything shared at POD level, containers can access it
Pod is considered as atomic unit of scheduling
It is the duty of kubernetes to keep running POD

How Kubernetes works? All components and roles of their components?

Ankit Aggarwal — Wed, 09 Nov 2022 07:47:43 +0000

Everything in K8s is on K8s cluster
It has two main components Master Node and Worker Nodes
Master Node act as a controller while worker nodes will have container where actual application will be running
Master Node have mainly 4 components API-server, Cluster Store, Controller Manager & Scheduler
Api server will act as communication mechanism for whole components to communicate to achieve all the task
Api server will keep talking to Cluster store to fetch or feed the latest information which it can send to other components to master nodes,
Cluster store will store all the information about pods, configurations and APIs
Controller manager have various components like node controller, endpoint controller, namespace controller. It watches for the changes and tried to manage the desired state to keep the application running
Scheduler will actually commands kubelet to create, deploy and manage PODs & container 10.In the worker nodes we have three components Kublete, Kube Proxy and container engine to perform actions

Why we need kubernetes?

Ankit Aggarwal — Wed, 09 Nov 2022 07:35:21 +0000

It is very difficult to deploy & scale application manually as it might require loads of resources and time.
To handle the scalability challenges we face
To eliminate human error we might face on multiple deployment
To deploy application very fast at any scale without hustle
To Orchestrate the whole process of deployment of microservices
To tackle network problem like load balancing
To Manage 100% availability of Application all the time
To re-deploy changes easily and fastly

What is Kubernetes?

Ankit Aggarwal — Wed, 09 Nov 2022 07:24:00 +0000

Kubernetes considered to be like an Orchestrator to orchestrate containers, microservices etc.
Kubernetes is developed by google
Kubernetes is written in golang language
Kubernetes is inspired from Borg & Omega
It is dedicated to CNCF in year 2014
You can say kubernetes is a deployment tool which helps to scale your application in very less time.
It is a open source tool
It also be referred as K8S in short form
It follows master to worker nodes architecture to deploy applications in microservices manner.
It came into picture to handle container scalability challenges
It do workload management, infrastructure abstraction and desired state recovering(auto healing) to keep application up and running through PODs
It works internally by communicating through APIs managed in api server
Four important components in kubernetes are Kube-Api server, Cluster Store, Scheduler and Controller

Diff between docker stop and docker kill?

Ankit Aggarwal — Mon, 07 Nov 2022 12:10:00 +0000

Docker stop command stops the container gracefully and safely. It is not abruptly shut the containers. In this case four events triggers kill, die, disconnect network and stop container. It will send SIGTERM signal(28). If in case if it couldn't stop container in certain time then it kills the container.

docker stop containerid

Docker kill command kills the container abruptly and in unsafe manner. It will directly kills the container and send 3 events which are kill, die and disconnect network. It actually send SIGKILL(9) signal to kill container immediately

docker kill containerid

All events can be tracked by running

docker events command