Debug School: rkumar0810@gmail.com

3rd day summary Rajesh Kumar

rkumar0810@gmail.com — Wed, 18 Oct 2023 12:00:11 +0000

How to deploy a POD

API Resource POD
ReplicationController
ReplicaSets
Deployment

DaemonSets
StatefulSets

DaemonSets

Min -Max ---1 POD per Node

Make sure that only one POD on every worker

ConfigMap is data that is stored in cluster and that (key-value) can be attached to POD and mounted to container.

kubectl exec helloworld-nginx -n=raj -- ls /etc/nginx/conf.d
kubectl exec helloworld-nginx -n=raj -- more /etc/nginx/conf.d/myconfo.conf

Service

Network loadbalancer -- Network layer
Application loadbalancer -- Application layer

Service is network loadbalancer when selector label of POD matches

Matching PODS with service are loadbalanced by the service (conditions apply)

healthy pods
configured for session affinity -- one user session goes to one pod
can point to things outside the cluster
random load balancing
uses TCP by default

watch curl http:// to check loadbalancing

Type NodePort creates a service that is exposed to node port and at the same time giving a port to outside

kubectl expose deployment my-dep --port=80 --target-port=80 -n=raj --type=NodePort

Loadbalancer is outside the cluster - it is connected to DNS through which the user can access.

User-->DNS-->LoadBalancer-->NodePort-->Service-->PODs

NAME based Rule based certificate based - INGRESS rules

Ingress is application loadbalancer

Loadbalancer (outside the cluster) --> Ingress Controller (inside the controller) --> Read the rules in the Ingress --> Redirects traffic based on rules

2nd day learning summary Rajesh

rkumar0810@gmail.com — Tue, 17 Oct 2023 11:47:41 +0000

1) Namespace

Namespaces allow for a logical separation between different pods within the cluster. This is useful when you are running different products, often managed by different teams, within the same cluster. Using namespaces ensures that one product's pods does not accidentally hamper another product's pods

2) Troubleshoot POD

You access container but thru POD IP

Type kubectl to get help on commands available

kubectl logs podName

curl http://IP for logs

kubectl option-picked -h --> help command

kubectl exec -it POD /bin/bash --> inside container

kubectl port-forward --address 0.0.0.0 pod/rajesh 8888:80 ----> 8888 is POD port and can be any available port

kubectl cp pod.yaml rajesh:/tmp ----> copy file

kubectl exec rajesh ls /tmp

kubectl auth can-i create pod ---> check access

port is set up in image and the port number in pod.yaml is for information only

3) Deployment

Replication. ------ From ReplicaSets
Controller ------ From ReplicaSets
Versioning
Rollout
Rollback

Strategy
Recreate
RollingUpdate

Versioning is clubbed with Rollout

What is POD by Rajesh Kumar

rkumar0810@gmail.com — Mon, 16 Oct 2023 11:46:59 +0000

1) A pod can't be created
2) A pod is instantiated
3) A pod can have multiple containers
4) A worker can have multiple pods
5) Multiple containers is a bad design
6) A pod has IP address
7) A pod has three states - instantiating, running, passed/error
8) A pod is a logical unit
9) Containers within pod can share resources
10) A pod is an atomic unit - it is either there or not
11) Pods communicate on the pods network

What is Kubernetes by Rajesh Kumar

rkumar0810@gmail.com — Mon, 16 Oct 2023 07:06:09 +0000

Assignment # 1 -

What is Kubernetes?

Kubernetes is the orchestrator for containers. Based on the command received it maintains a the requested number of containers in a cluster and maintains the state of the cluster.

Why Do we need Kubernetes? Explain in 10 lines

I will start with a contrast to virtual machines.
For virtual machines we can manage the following very easily as there are a lot of software options available -

Memory
Disk
Security
Load Balancer
Compute
Database

However, it is almost impossible to do so for containers. The only option that is available is Kubernetes. As mentioned earlier, Kubernetes is the orchestrator for containers.
Kubernetes can bring containers up and down and maintain a certain number of them running within a blink of an eye.

It is an open source software that was initially developed by Google and then it was handed over to the open source community which has maintained and added features since then.

It is written in Go/GOLANG.

How Kubernetes Works?

Kubernetes Architecture. Explain each component with 1 one line.

Kubernetes has the below 4 components that run on a VM -

1) API Server

This is a collection of APIs that receives requests and provides response.

2) Controller Manager

This runs in a loop and keeps checking the containers in the cluster

3) Scheduler

Scheduler schedules the tasks and keeps a state of requested number of containers running

4) etdc

This is the database and is the source of truth for the cluster

5) Kubelet runs on each worker node

6) For the cluster there is a networking component that doesn't use NAT

6) Kubectl is the application that is used on the laptop to provide commands

7) There can be multiple controller planes (first 4 components together) and it is recommended to run at least 3 instance of them