About managed Kubernetes
What an Akasha managed Kubernetes cluster is, what Akasha operates for you, and what you control.
Akasha provisions and operates Kubernetes clusters inside your project. Each cluster is a group of virtual machines running Talos Linux — an immutable, API-managed operating system built for Kubernetes — so you get a working cluster with full kubectl access without installing or maintaining node software yourself.
Cluster anatomy
- Control plane — one replica for development, or three for high availability. The control plane size is chosen at creation.
- Worker pools — groups of identically sized worker nodes, named
<cluster>-md-<n>. Pools can be scaled, autoscaled, labeled, and tainted independently. See Manage worker pools. - API endpoint — the Kubernetes API server is published on a virtual IP allocated from your network's uplink address ranges, on port
6443. Akasha keeps this endpoint pointed at the cluster's control-plane nodes.
Node sizes use flavors in the format c<vCPU>-m<GB RAM>, for example c2-m4 for 2 vCPUs and 4 GB of RAM.
Kubernetes version
Clusters run Kubernetes v1.35.0. The version is fixed by the platform: there is no version selector when you create a cluster, and there is currently no upgrade flow in the console or the API.
What Akasha manages, and what you manage
Akasha manages:
- The node operating system. Nodes run Talos Linux; there is no SSH access to nodes and no packages to patch.
- Provisioning, scaling, and replacement of node virtual machines.
- The Kubernetes API endpoint and the load balancing behind it.
- Installation of the add-ons you select (as queued operations you can track).
You manage:
- Your workloads, namespaces, and everything you deploy with
kubectl, Helm, or GitOps. - Which add-ons are installed and how they are configured.
- Ingress routes, external IPs, backup schedules, and worker pool sizing.
Project scoping
Every cluster belongs to a project. Cluster nodes are virtual machines that run in your project and draw on your project's networks and storage pools. Anyone with access to the project can see the cluster and download its admin credentials — see Access clusters and Access control.
The cluster detail page
Open a cluster from Kubernetes → Clusters to reach its detail page. It is organized into tabs:
| Tab | What it covers |
|---|---|
Overview | Status, endpoint, node counts, quick pool scaling, node list, and health conditions |
Configuration | YAML editor for the cluster's labels and annotations |
Nodes | Control plane and worker pool management |
Resource | The add-on catalog and installed add-ons |
Ingress | External IPs, HTTP routes, and TCP routes |
GitOps | Flux sources, bootstrap, and export |
Etcd | Scheduled etcd snapshot backups |
Terminal | A k9s session in your browser |