Rollouts and Versioning in deployments
Table of contents
When you first create a deployment, it triggers a rollout, a new rollout creates a new deployment.
Understanding Updates and Rollbacks in Kubernetes Deployments
Introduction
Kubernetes deployments provide a mechanism to manage application updates and rollbacks effectively. This document explains key concepts, strategies, and commands related to rollouts, updates, and rollbacks in a clear and detailed manner.
Key Concepts
1. Rollouts and Versions:
- A rollout occurs when a deployment is created or updated. Each rollout creates a deployment revision, enabling tracking of changes and rollbacks.
2. Deployment Strategies:
- Recreate Strategy: Stops all old versions and starts the new ones, causing downtime.
- Rolling Update Strategy (default): Gradually replaces old versions with new ones, ensuring application availability.
Commands and Their Usage
Creating a Deployment
Starts a new deployment.
Command:
kubectl create -f <deployment-file.yaml>
Example:
kubectl create -f my-deployment.yaml
Check Rollout Status
Shows the progress of a deployment rollout.
Command:
kubectl rollout status deployment/<deployment-name>
Example:
kubectl rollout status deployment/my-app
View Rollout History
Lists all revisions of a deployment, showing changes made.
Command:
kubectl rollout history deployment/<deployment-name>
Example:
kubectl rollout history deployment/my-app
Update the Deployment (Using Apply)
Apply changes from a new YAML file.
Command:
kubectl apply -f <updated-deployment-file.yaml>
Example:
kubectl apply -f updated-deployment.yaml
Update the Deployment (Using Set Image)
Set a new container image directly.
Command:
kubectl set image deployment/<deployment-name> <container-name>=<new-image>
Example:
kubectl set image deployment/my-app my-container=my-image:v2
Rollback the Deployment
Reverts to the previous version of the deployment.
Command:
kubectl rollout undo deployment/<deployment-name>
Example:
kubectl rollout undo deployment/my-app
Inspect Deployment Details
Provides detailed information about the deployment.
Command:
kubectl describe deployment/<deployment-name>
Example:
kubectl describe deployment/my-app
List Replica Sets
Shows all replica sets for the deployment.
Command:
kubectl get replicasets
Example:
kubectl get replicasets
Delete a Deployment
Removes the deployment entirely.
Command:
kubectl delete deployment/<deployment-name>
Example:
kubectl delete deployment/my-app
Deployment Lifecycle and Flow
1. Deployment Creation:
- Trigger rollout → Create ReplicaSet → Scale Pods.
2. Update Process:
- Modify configuration → Apply changes → Create new ReplicaSet → Rolling update.
3. Rollback:
- Undo rollout → Scale down current ReplicaSet → Scale up previous ReplicaSet.
| Aspect | kubectl set | kubectl apply (Update) |
| Purpose | Used for imperative changes (quick updates) | Used for declarative updates (updates the configuration in YAML files) |
| Use Case | Quick, temporary changes like updating image, replicas, etc. | Updating the resource definition, keeping it in sync with the YAML file |
| Changes to YAML | Does not modify the local YAML file. Updates are applied directly to the running resources. | Modifies the YAML file (if you update the file locally) and applies changes to the cluster. |
| Version Control | Not tracked in source control. Only affects live resources. | Tracked in source control as part of the YAML file. |
| When to Use | When you need a quick update or when working in a testing or development environment. | When you need to manage changes declaratively, track them over time, and maintain configuration in source control. |
| Examples of Changes | Changing container image, scaling replicas, setting environment variables directly in the cluster. | Modifying container image, replicas, or any other setting in the YAML file and applying the update using kubectl apply. |
| Reusability | Updates are not stored in YAML files, so it is not easy to reapply the same changes later. | The updated configuration is stored in YAML files, making it easy to apply changes across environments. |
| Consistency Across Clusters | Changes are applied to the current cluster only; they need to be manually replicated elsewhere. | Changes are part of version-controlled YAML files, which can be applied across multiple clusters for consistency. |
| Automation/Scripting | Ideal for use in scripts and automated processes where the local file does not need to be modified. | Better for managing configuration as code, especially in a CI/CD pipeline or a multi-cluster environment. |
| Example Command | kubectl set image deployment/my-app my-container=my-app:v2 | kubectl apply -f deployment.yaml (after editing the YAML file) |