Kubernetes :7

DevOps Classroom Series – 10/Apr/2021

Introducing the problem

• Consider we have replicaset where we deployed the application with 4 replicas 
• Now when we get a new release from dev team, we build a new docker image and upload it to the docker hub or any private registry
• How can we apply the new image to replicaset?
  • one approach is to change the replicaset and apply new changes (new docker image) but this will have downtime
• We need a solution where we can apply the new docker images with replications without having downtime
• This can be achieved by kubernetes Deployment controller object

Deployment

• A deployment is a k8s object that acts a wrapper around Replica Set and makes it easier to use.
• The major motivation for deployments is to maintain the history of revisions.
• Every time a change is made to the replica set or the underlying pods a new version of the ReplicaSet is recorded by Deployment. This way using a Deployment makes it easy to rollback to previous state or version
• Hierarchy of Deployment, ReplicaSet, Pods and container 
• Strategy: In this strategy section we can define strategy to replace old pods with new ones when you are updating . There are two kinds of strategies
  • RollingUpdate
  • Recreate
• RollingUpdate:
  • We use this strategy to update deployment without having a downtime
  • In Rolling update we specify two important values
    • maxUnavailable: This is max number of Pods that can be unavailable during the update. default value for maxUnavailable is 25%
    • maxSurge: this is number of Pods that can be scheduled/created above the desired number of Pods. default value for maxUnavailable is 25%
• Recreate: In this strategy all the existing pods will be killed b4 creating new pods, so there will be some down time
• Lets create a simple deployment
• Refer Here for the changeset
• Lets apply the template  
• Labels and annotations on the deployment
  • k8s adds an annotation with the deployment.kubernetes.io/revision key which contains information about how many revisions have been there for particular deployment 
• Added a simple service to access jenkins Refer Here for spec
• Now make some changes to simulate new application version Refer Here for the changes
• Now lets apply the changes using kubectl apply with –record flag. This flag esures that the the update to the deployment is recorded in the history of deployment   
• Lets undo the deployment  

DaemonSets

• DaemonSets are used to manage the creation of particular pod on all or set of selected nodes in a cluster
• Use cases for DaemonSet
  • Logging: To manage running a log collection pod on all nodes and then process the logs in a log processing pipeline (sending logs to central logging server)
  • Local data caching
  • Monitoring
• Lets write one daemon set
  • Refer Here for the changeset
  • lets apply the template 

Jobs

• A job is supervisor in k8s that can be used to manage pods to run some script/task and exit gracefully
• The pods created by the job arenot deleted following the completion of job. The pods run to completion and stay with a Completed status
• Lets create a k8s job configuration which will run the alpine pod to sleep for 2 minutes (sleep 2m) with restartPolicy OnFailure
• Refer Here for the changeset  
• Exercise: Try to create a cron job spec which run every 15 minutes using same pod spec