[Kubernetes] Managing Application with Kubernetes - Rolling Updates

Explain what a rolling update is and how it works:

A rolling update is an automated update process in Kubernetes that allows for the seamless and controlled deployment of changes to applications across pods. It ensures minimal to no downtime by gradually replacing old instances of an application with new ones, one at a time, ensuring that the application remains available throughout the update process. This process involves deploying new pods with updated configurations while gradually terminating the old ones until the entire deployment has been updated.

List the pre-steps before a rolling update can be applied:

1. Add liveness and readiness probes: Ensure that deployments have liveness and readiness probes configured appropriately. These probes determine the health of the application within pods, allowing Kubernetes to manage the deployment effectively.

2. Define a rolling update strategy: Specify the rolling update strategy in the YAML file of the deployment. Parameters such as maxSurge, maxUnavailable, and minReadySeconds define how the update process should be executed.

Demonstrate how to roll back a rolling update:

1. Scenario 1: all-at-once rollout

   • In an all-at-once rollout, all v1 objects are removed before v2 objects become active.
   • New pods are created for version 2 while version 1 pods are marked for deletion and removed.
   • Once all version 1 pods are removed, version 2 pods become active, restoring user access.

2. Scenario 2: all-at-once rollback

   • In an all-at-once rollback, all v2 objects are removed before v1 objects become active.
   • New pods are created for version 1 while version 2 pods are marked for deletion and removed.
   • Once all version 2 pods are removed, version 1 pods become active, restoring user access.

3. Scenario 3: one-at-a-time rollout

   • In a one-at-a-time rollout, the update is staggered to prevent interruption of user access.
   • New pods are created for version 2, and one version 1 pod is marked for deletion and removed.
   • This process repeats until all version 1 pods have been replaced by version 2 pods.

4. Scenario 4: one-at-a-time rollback

   • In a one-at-a-time rollback, the rollback process is staggered to prevent interruption of user access.
   • New pods are created for version 1, and one version 2 pod is marked for deletion and removed.
   • This process repeats until all version 2 pods have been replaced by version 1 pods.

In summary, rolling updates in Kubernetes facilitate seamless application updates while maintaining availability. Pre-steps such as configuring probes and defining update strategies are essential for a successful rolling update. Additionally, different rollback scenarios ensure that deployments can revert changes if necessary without disrupting user access.