Rolling Updates and Rollbacks

Question 1

What is a rollout in the context of a Kubernetes deployment?

Answer 1

A rollout is the process triggered when a deployment is created or updated which creates or updates replica sets and pods for that deployment.

Question 2

What happens under the hood when you first create a new deployment?

Answer 2

Kubernetes creates a new replica set for that deployment and this is recorded as the first deployment revision.

Question 3

What does a new deployment revision represent?

Answer 3

A deployment revision represents a specific version of the deployment configuration such as a particular container image version or settings at the time of a rollout.

Question 4

What typically triggers a new rollout after the initial deployment is created?

Answer 4

Updating the deployment such as changing the container image version triggers a new rollout and creates a new deployment revision.

Question 5

Why is tracking deployment revisions useful?

Answer 5

Tracking revisions allows you to see what changes were made over time and enables you to roll back to a previous working version if needed.

Question 6

Which kubectl command shows you the status of an ongoing deployment rollout?

Answer 6

You use kubectl rollout status followed by the deployment name to see the rollout status.

Question 7

Which kubectl command shows the history of rollout revisions for a deployment?

Answer 7

You use kubectl rollout history followed by the deployment name to see the list of revisions and their history.

Question 8

What are the two main deployment strategies described in the lecture?

Answer 8

The two strategies are Recreate and RollingUpdate.

Question 9

What is the Recreate deployment strategy?

Answer 9

The Recreate strategy terminates all existing pods of the old version first and only then creates new pods of the updated version which causes downtime.

Question 10

Why is the Recreate strategy risky for production applications?

Answer 10

Because during the time after old pods are deleted and before new pods are up the application is unavailable to users.

Question 11

What is the RollingUpdate deployment strategy?

Answer 11

The RollingUpdate strategy gradually scales down old pods and scales up new pods so that some pods are always available and the application stays online during the update.

Question 12

Which deployment strategy is the default if you do not specify one in the deployment spec?

Answer 12

RollingUpdate is the default deployment strategy.

Question 13

How can you update a deployment using its definition file?

Answer 13

You modify the deployment yaml file with the desired changes and then run kubectl apply dash f with that file to trigger a new rollout.

Question 14

What kinds of changes in a deployment yaml file can trigger a rollout?

Answer 14

Changes such as updating the container image version changing labels or altering the number of replicas can trigger a new rollout.

Question 15

What kubectl command can you use to update the container image of a deployment without editing the yaml file?

Answer 15

You can use kubectl set image on the deployment specifying the container name and new image.

Question 16

What is a potential risk of using kubectl set image instead of updating the yaml file?

Answer 16

If you change the image with kubectl set image but do not update the yaml file the live configuration and the file configuration can drift out of sync and cause confusion later.

Question 17

How can you see detailed information about a deployment and its strategy behavior?

Answer 17

You can run kubectl describe deployment followed by the deployment name to view strategy type events and replica set changes.

Question 18

In the describe output how does the event sequence differ between Recreate and RollingUpdate strategies?

Answer 18

With Recreate you will see the old replica set scaled down to zero first and then the new replica set scaled up while with RollingUpdate you see old and new replica sets scaled down and up gradually side by side.

Question 19

When you first create a deployment with five replicas how many replica sets are created and what do they do?

Answer 19

One replica set is created which then creates and manages five pods to satisfy the replica count.

Question 20

When you upgrade a deployment to a new version how do replica sets change under the hood?

Answer 20

Kubernetes creates a new replica set for the new version and gradually moves pods from the old replica set to the new one according to the RollingUpdate strategy.

Question 21

After an upgrade what might kubectl get rs show for the old and new replica sets?

Answer 21

It might show the old replica set with zero pods and the new replica set with the full number of pods such as five.

Question 22

If you detect a problem after upgrading your deployment what Kubernetes feature helps you go back to the previous version?

Answer 22

You can use the rollback feature of deployments to return to an earlier revision.

Question 23

Which kubectl command do you use to roll back a deployment to a previous revision?

Answer 23

You run kubectl rollout undo followed by the deployment name to roll back.

Question 24

During a rollback how do the replica sets change?

Answer 24

Kubernetes scales down pods in the newest replica set and scales up pods in the previous replica set so the application returns to the older version.

Question 25

After a rollback how would the output of kubectl get rs differ from before the rollback?

Answer 25

Before rollback the new replica set had the active pods and the old replica set had zero pods after rollback this is reversed and the previous replica set again has active pods.

Question 26

Which kubectl command is used to initially create a deployment from a file?

Answer 26

You use kubectl create dash f followed by the deployment file.

Question 27

Which kubectl command lists existing deployments in a namespace?

Answer 27

You use kubectl get deployments to list deployments.

Question 28

Which kubectl commands are highlighted in the lecture for updating deployments?

Answer 28

The lecture highlights kubectl apply with the yaml file and kubectl set image for updating deployments.

Question 29

Which two rollout related kubectl commands are emphasized for checking and controlling deployment rollouts?

Answer 29

The two commands are kubectl rollout status to watch the rollout and kubectl rollout undo to roll back a deployment.