Are the 3 nodes recommended to Enable HA Control Plane sufficient to achieve the 99.99% uptime?
My question is whether having two nodes with the enabled HA Control Plane is sufficient to achieve the 99.99% uptime, or if it is necessary to have three nodes as recommended?
The 3 worker nodes are recommended though having 2 nodes with the enabled HA Control Plane is sufficient.
In an LKE cluster, there's the Managed control plane (Kubernetes API server, etcd, the Kubernetes scheduler, the cloud control manager, and the Kubernetes controller manager) and your worker nodes. The worker nodes are the nodes that reside in a Node Pool. These nodes will contain all of your workloads, services, deployments, pods, etc. You can select 1, 2, 3, or even up to 100 nodes. The more nodes you deploy, the more options you have in terms of fault tolerance for your workloads.
By default, the control plane components are not replicated. If any of them fail, it’s possible that there may be issues with your cluster. Enabling HA (High Availability) Control Plane adds an additional layer of redundancy by replicating these control plane components. In doing so, this feature can ensure that the cluster experiences maximum uptime (99.99% guaranteed) and is recommended for all production applications running on LKE.
You can read more about the control plane in the below guide:
Please know that while upgrading to an HA cluster is always possible, downgrading your cluster is not currently supported. Enabling HA is an irreversible change for your cluster.