Overview

Workloads running on Kubernetes do not make assumptions about the underlying infrastructure. However, it is still non-trivial to move them across different clusters. It is also non trivial to move them to another namespace within the same cluster scope such as to a different zone. KubeMove provides a set of tools and API to coordinate the orchestration and workflow of moving an application in production across cluster or namespace boundaries.

Architecture/Design proposal

The initial KEP (KubeMove Enhancement Proposal) is at KEP-1

Join the discussion through the issue here

Using Kubemove

• Install kubemove using helm - helm install kubemove.io/kubemove
• Annotate the application to be moved with kubemove.io/kubemove.enable="true"
• Use MoveEngine API with the required configuration to start the data movement or sync
• Wait for the movement or datasync to complete and switch the application on the target cluster using MoveSwitch API

CLI

Example move commands:

kubectl km move init <app> <kubemove-template.yaml>

kubectl km status <movepair-cr>

kubectl km move switch <movepair>

Use cases

• Hybrid and multi-cloud deployments - Move applications on-prem to cloud or vice-versa or from cloud-cloud
• Onramp to Kubernetes - KubeMove can help migrate the data from legacy volumes onto Kubernetes
• Kubernetes and/or application upgrades - Applications may need to be moved back and forth while following blue green strategy

Developer Guide

Kubemove uses Makefile based build process. This section will give you an overview on how to build and deploy Kubemove resources.

Setup Environment

Setup your development environment by the following steps.

• Use your own docker account for the docker images:

export REGISTRY=<your docker username>

• For AWS ECR

export REGISTRY=<accountid>.dkr.ecr.<Region>.amazonaws.com

• Changes in build/push

  Replace this sudo docker login "${REGISTRY}" -u "${DNAME}" -p "${DPASS}"; with aws ecr get-login-password --region <Region> | docker login --username AWS --password-stdin "${REGISTRY}"

• Build a developer image with all dependencies:

make dev-image

Code Generation

If you update any API or any gRPC protos, then generate respective codes by the following steps.

• Generate gRPC codes:

make gen-grpc

• Generate CRDs and respective codes.

make gen-crds

• Update respective controllers

make gen-k8s

Alternatively, you can run the following command that will run all the previous code generation commands.

make gen

Build the Binaries and Docker Images

If you update any codes, then re-build the project and the respective docker images.

• Run gofmt

make format

• Run linter

make lint

• Build binaries

make build

• Build docker images

make images

• Push docker images

make deploy-images

Deploy Kubemove

At first, create two different clusters. Make sure KUBECONFIG environment variable is pointing to the right cluster config file.

• Specify the source cluster, and the destination cluster

export SRC_CONTEXT=<source cluster context>
export DST_CONTEXT=<destination cluster context>

• Register the Kubemove CRDs

make register_crds

• Create the RBAC resources

make create_rbac_resources

• Deploy MovePair controller

make deploy_mp_ctrl

• Deploy MoveEngine controller

make deploy_me_ctrl

• Deploy DataSync controller

make deploy_ds_ctrl

You can also install all the Kubemove resources by using the following command:

make deploy_kubemove

If you are using two local kind cluster, you can create a MovePair using the following command:

make create_local_mp

Removing Kubemove

You can uninstall/remove all Kubemove resources created in the deploy section just by replacing deploy,create, or register word of the respective command with remove. For example:

make remove_mp_ctrl

Alternatively, you can use the following command to remove all the Kubemove resources created in the clusters using the following command,

make purge_kubemove

License

KubeMove is developed under Apache 2.0 license.

Contributing

We welcome participation from the community in defining more use cases, developing API spec and implementation. Please write new issues as you like.

If you would like to know the ongoing work and its status, you can join us at status meeting that happens every Monday / Wednesday / Friday at 2.30 PM IST.

Minutes of meeting are captured in this doc