Managed Kubernetes Product Description

Last update: June 24 2025

Selectel's Managed Kubernetes simplifies the process of deploying, scaling and maintaining your Kubernetes container infrastructure. Selectel is responsible for version updates, security and uptime of Control Plane Kubernetes.

The product supports user types and roles, projects and project limits and quotas.

Versions⁠

Managed Kubernetes clusters support versions 1.30.x, 1.31.x, 1.32.x

How Managed Kubernetes works⁠

A Managed Kubernetes cluster can work:

• on cloud servers — utilizes the resources of the cloud platform;
• or on dedicated servers — dedicated servers, dedicated server networks and cloud platform resources (load balancers, public subnets, public IP addresses) are used.

For more information about the resources used, see the Projects instruction.

Containerd is used as the container execution environment (CRI) . Calico is used as the CNI in Managed Kubernetes clusters.

You can work with a Managed Kubernetes cluster in the control panel and via the Managed Kubernetes API. You can also work with a cluster on a cloud server via Terraform.

Managed Kubernetes clusters on dedicated servers are in beta testing. Consider the limitations when selecting clusters on dedicated servers.

Cluster composition⁠

Managed Kubernetes clusters consist of:

• of master nodes — contain the control components of the cluster, Control Plane. The number of master nodes depends on the type of cluster. Master nodes are not displayed in the Control Panel, they are the responsibility of Selectel;

• work node groups — contain containers of user applications. Work nodes are displayed in the control panel and are the responsibility of the user.

A group of working nodes must be in the same accessibility zone as the master nodes. For more information, see Working with node groups.

Types of cluster⁠

Selectel provides two types of Managed Kubernetes clusters: fault-tolerant and basic.

You can only select a cluster type when creating a cluster. Once the cluster is created, the cluster type cannot be changed.

	Fault-tolerant	Basic
Number of master notes	3	1
Master Node Allocation	in different segments of the pool — if there are several segments in the pool; or in the same pool segment on different hosts — if there is only one segment in the pool	In one segment of the pool
Fault tolerance	If one of the three master nodes is unavailable, Control Plane will continue to operate	If the master node is unavailable, Control Plane will not work
SLA	✓ (99,98%)	✗
Functionality	All functionality is available	Auto-update of patch versions is not available
What tasks are suitable for	For the working environment (production)	for the developer's environment (development); test environment (testing and staging); pet projects

Limits⁠

Cloud server

Maximum number of fault-tolerant Kubernetes clusters in one pool for one project	10
Maximum number of Kubernetes base clusters in one pool for one project	10
Maximum number of node groups in one pool for one project	100
Maximum number of nodes in one node group	15
Maximum number of vCPU nodes	32*
Maximum number of RAM nodes	256*GB
Maximum size of the node boot disk	1.2 TB.
Maximum number of pods on one node	100
Maximum number of persistent volumes (PV) per node	256
Minimum size of one permanent volume (PV)	1 GB

* You can create nodes with more vCPUs and RAM — use the fixed cloud server configurations.

Dedicated server

Maximum number of fault-tolerant Kubernetes clusters in one pool for one project	10
Maximum number of Kubernetes base clusters in one pool for one project	10
Maximum number of node groups in one pool for one project	100
Maximum number of nodes in one node group	15

Cluster limitations on dedicated servers⁠

Managed Kubernetes clusters on dedicated servers are in beta testing.

Not supported during beta testing:

• use of arbitrary configurations of dedicated servers;
• adding existing dedicated servers to the cluster;
• adding multiple groups of nodes when creating a cluster;
• updating minor versions of Kubernetes;
• automation: auto-update of patch versions, auto-scaling and auto-recovery;
• Permanent Volume (PV) connection based on the cloud platform's network disks;
• use of user data;
• Using dedicated server configurations with GPUs;
• Terraform usage.

Areas of responsibility⁠

Selectel provides⁠

• creation and accessibility of master notes;
• creation of working nodes;
• updating versions of the Managed Kubernetes cluster;
• masternode monitoring;
• possibility of autoscaling of nodes;
• Node auto-recovery capability;
• data storage security in accordance with the requirements of 152-FZ;
• integration with Selectel services;
• technical support.

Selectel is not responsible for⁠

• for managing the Managed Kubernetes cluster;
• node management;
• application creation;
• initiating scaling and upgrades.

If you need help with Managed Kubernetes cluster administration, order service administration services.