Documents for Highly Available Topology

[RainbowMango]

What documentation would you like to be added:

Add some documents about how to deploy the HA（highly available） Karmada.

Why do you think this document is needed:

I believe that Karmada is capable of high availability because I have heard that some users have already done so, but so far, there are no relevant docs in Karmada that describe how to deploy a high availability control plane.

Many users would need this document, such as :

• Control plane availability karmada#3773
• Karmada controll plane fail over support  karmada#3954

Where do you think the document should be placed:

• Highly Available Topology Overview (@Affan-7, Karmada Highly Available Topology Overview #443)
  • this document could be placed at https://github.com/karmada-io/website/tree/main/docs/installation/ha-installation.md
  • reference: Kuberentes Options for Highly Available Topology
• Creating Highly Available Karmada with karmadactl
  • this document could be placed at https://github.com/karmada-io/website/tree/main/docs/installation/ha-installation-with-cli.md
  • reference: Creating Highly Available Clusters with kubeadm
• Creating Highly Available Karmada with helm chart
  • this document could be placed at https://github.com/karmada-io/website/tree/main/docs/installation/ha-installation-with-helm-chart.md
• Creating Highly Available Karmada with karmada-operator
  • this document could be placed at https://github.com/karmada-io/website/tree/main/docs/installation/ha-installation-with-operator.md
• Does Karmada support high availability(HA) deployments?
  • add an item to FAQ at https://github.com/karmada-io/website/blob/main/docs/faq/faq.md

[Affan-7]

/assign

[Affan-7]

I have successfully created a karmada cluster and a member cluster on my local machine using kind. I also registered the member cluster to the karmada. But when I try to deploy karmada with HA using kubectl karmada init --karmada-apiserver-replicas 3 --etcd-replicas 3 command. It give me the following error.

error: for data security,when etcd storage mode is hostPath,etcd-replicas can only be 1

Is there a solution to create multiple etcd storages with hostPath storage mode?
Or any other solution to run karmada HA locally?

[samzong]

Hi @Affan-7

Thanks for your choose the Karmada.

I'm not an expert in this area. However, based on the error message, my suggestion is that you could try to resolve the problem by using a storageclass or an external etcd.

• https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/setup-ha-etcd-with-kubeadm/

Using a single replica of etcd in a production environment is dangerous.

If you plan to use Karmada in a production environment, please let us know at the earliest opportunity.

If you have any more questions, feel free to create an Issue or join our official Slack channel at any time.