Completek8s Classroomnotes 02/Aug/2023

Kubernetes Pods

• There can be applications which are expected to run multiple processes. Docker goes with a golden rule of single process applications, In the case of k8s we can run each process in a container if we need
• Sometimes we might need extra functionality to inject log exports, network monitoring, tracing, then we can add one more container as side car in your pod
  
• In k8s there are two ways of creating resources
  • imperative:
    • This is good for one time jobs as we create a command
  • declartive
    • This is good for repetitive tasks as we create a manifest
    • we create a desired state by expressing in a yaml file called as manifest
• Composing YAML Files

apiVersion:
kind:
metadata:
spec:

• Versioning Refer Here
• kubernetes has resources which are grouped and there are stages for maturity levels and k8s apiVersion tries to express them all

apiVersion: <group>/<version>  # for all the other groups
apiVersion: <version> # if the group is core

• Api reference Refer Here
  
• kind: This represents the resource which we are trying to create
• metadata: This object represents information about the resources
  • name
  • labels
  • annotations
• spec: This is specification of resource
• Lets create a simple manifest file for nginx pod

apiVersion: v1
kind: Pod
metadata:
  name: nginx
spec:
  containers:
  - name: nginx
    image: nginx

• Create a nginx.yaml file and execute by using kubectl apply -f nginx.yaml
  
• The equivalent in declarative is kubectl run nginx --image=nginx
• Try to come up with a docker command
  • to run the alpine container with cmd overwritten to sleep 1d
    • docker command: docker container run -d --name alp alpine sleep 1d
  • to run the jenkins/jenkins container
• Manifest for alpine docker container run -d --name alp alpine sleep 1d

---
apiVersion: v1
kind: Pod
metadata:
  name: alp
spec:
  containers:
    - name: alp
      image: alpine
      command:
        - sleep
        - 1d

* Manifest for jenkins docker container run -d --name jenkinscicd jenkins/jenkins

---
apiVersion: v1
kind: Pod
metadata:
  name: jenkinscicd
spec:
  containers:
    - name: jenkinscicd
      image: jenkins/jenkins

Activity

• Write a Manifest for a Pod which runs two containers
  • name is first & image is alpine sleep 1d
  • name is second & image is ubuntu sleep 1d

---
apiVersion: v1
kind: Pod
metadata:
  name: mcpod
spec:
  containers:
    - name: first
      image: alpine
      command:
        - sleep
        - 1d
    - name: second
      image: ubuntu
      command:
        - sleep
        - 1d

* Login into alpine container

kubectl exec -it -c first  mcpod -- /bin/sh

• login into ubuntu container

kubectl exec -it -c second  mcpod -- /bin/bash

• Experiments:
  • Try creating a pod with
    • alpine container with sleep 10s
    • observe the failures after 10 s
  • Try creating a pod with 2 containers
    • one container put sleep 1d
    • other container put sleep 5s
    • observe failures
  • Try to run a jenkins container in docker and see the logs docker logs
  • Try running a container in docker with memory restrictions such as 256 MB RAM