How do you use Kubernetes for rolling updates?

In Kubernetes, rolling updates are a process that gradually upgrades applications to a new version during deployment updates while minimizing downtime. Kubernetes leverages its powerful scheduling and management capabilities to automatically handle rolling updates. The following are the steps and considerations for performing rolling updates:

1. Prepare the New Application Version

First, ensure that you have prepared the new version of the application and created a new container image. Typically, this includes application development, testing, and pushing the image to a container registry.

2. Update the Deployment Image

In Kubernetes, the most common method to update an application is to modify the container image referenced in the Deployment resource. You can update the image using the following command:

bash
kubectl set image deployment/<deployment-name> <container-name>=<new-image>:<tag>

Here, <deployment-name> is the name of your Deployment, <container-name> is the name of the container within the Deployment, and <new-image>:<tag> is the name and tag of the new image.

3. Rolling Update Process

After updating the Deployment's image, Kubernetes initiates the rolling update. During this process, Kubernetes gradually replaces old Pod instances with new ones. This process is automatically managed, including:

• Gradual creation and deletion of Pods: Kubernetes controls the speed and concurrency of updates based on the defined maxSurge and maxUnavailable parameters.
• Health checks: Each newly started Pod undergoes startup and readiness probes to ensure the health of the new Pod and service availability.
• Version rollback: If issues arise with the new version deployment, Kubernetes supports automatic or manual rollback to a previous version.

4. Monitor Update Status

You can monitor the status of the rolling update using the following command:

bash
kubectl rollout status deployment/<deployment-name>

This displays the progress of the update, including the number and status of updated Pods.

5. Configure Rolling Update Strategy

You can configure the rolling update strategy in the Deployment's spec section:

yaml
spec:
  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 0

• maxSurge defines the number of Pods that can exceed the desired count.
• maxUnavailable defines the maximum number of Pods that can be unavailable during the update.

Example: Practical Application of Rolling Updates

Suppose I have a backend service for an online e-commerce platform deployed on Kubernetes. To avoid disrupting users' shopping experience, I need to update the service. I will first fully test the new version in a test environment, then update the production Deployment's image, and monitor the progress of the rolling update to ensure sufficient instances are available to handle user requests at all times.

Through this approach, Kubernetes' rolling update functionality makes application updates flexible and reliable, significantly reducing the risk of disruptions and service interruptions.