import Pool from "../partials/cmx/_openshift-pool.mdx"
This topic lists the supported cluster and VM types that you can provision with Compatibility Matrix (CMX).
CMX can create VMs, VM-based clusters (such as kind, k3s, RKE2, and Red Hat OpenShift OKD), and cloud-managed clusters (such as EKS, GKE and AKS):
-
Cloud-based Kubernetes distributions are run in a Replicated managed and controlled cloud account to optimize and deliver a clusters quickly and reliably. The Replicated account has control planes ready and adds a node group when you request it, making the cluster available much faster than if you try to create your own cluster with your own cloud account.
-
VMs and VM-based clusters run on Replicated bare metal servers located in several data centers, including data centers physically in the European Union.
This section lists the supported VM cluster distributions for clusters created with CMX.
VM-based clusters refers to clusters that run on Hetzner servers with the CMX cluster provisioner. This allows for greater flexibility than with Cloud Clusters like AWS, EKS, etc. For example, with VM-based distributions, CMX offers warm pools to make Openshift startup times very fast.
CMX supports creating kind clusters.
| Type | Description |
|---|---|
| Supported Kubernetes Versions | {/* START_kind_VERSIONS */}1.30.13, 1.31.14, 1.32.11, 1.33.7, 1.34.3, 1.35.1{/* END_kind_VERSIONS */} |
| Supported Instance Types | See Replicated Instance Types |
| Node Groups | No |
| Node Auto Scaling | No |
| Nodes | Supports a single node. |
| IP Family | Supports `ipv4` or `dual`. |
| Limitations | See Limitations |
| Common Use Cases | Smoke tests |
CMX supports creating k3s clusters.
| Type | Description |
|---|---|
| Supported k3s Versions | The upstream k8s version that matches the Kubernetes version requested. |
| Supported Kubernetes Versions | {/* START_k3s_VERSIONS */}1.32.0, 1.32.1, 1.32.2, 1.32.3, 1.32.4, 1.32.5, 1.32.6, 1.32.7, 1.32.8, 1.32.9, 1.32.10, 1.32.11, 1.32.12, 1.32.13, 1.33.0, 1.33.1, 1.33.2, 1.33.3, 1.33.4, 1.33.5, 1.33.6, 1.33.7, 1.33.8, 1.33.9, 1.33.10, 1.34.1, 1.34.2, 1.34.3, 1.34.4, 1.34.5, 1.34.6, 1.35.0, 1.35.1, 1.35.2, 1.35.3{/* END_k3s_VERSIONS */} |
| Supported Instance Types | See Replicated Instance Types |
| Node Groups | Yes |
| Node Auto Scaling | No |
| Nodes | Supports multiple nodes. |
| IP Family | Supports `ipv4`. |
| Limitations | For additional limitations that apply to all distributions, see Limitations. |
| Common Use Cases |
|
CMX supports creating RKE2 clusters.
| Type | Description |
|---|---|
| Supported RKE2 Versions | The upstream k8s version that matches the Kubernetes version requested. |
| Supported Kubernetes Versions | {/* START_rke2_VERSIONS */}1.32.0, 1.32.1, 1.32.2, 1.32.3, 1.32.4, 1.32.5, 1.32.6, 1.32.7, 1.32.8, 1.32.9, 1.32.10, 1.32.11, 1.32.12, 1.32.13, 1.33.0, 1.33.1, 1.33.2, 1.33.3, 1.33.4, 1.33.5, 1.33.6, 1.33.7, 1.33.8, 1.33.9, 1.33.10, 1.34.1, 1.34.2, 1.34.3, 1.34.4, 1.34.5, 1.34.6, 1.35.0, 1.35.1, 1.35.2, 1.35.3{/* END_rke2_VERSIONS */} |
| Supported Instance Types | See Replicated Instance Types |
| Node Groups | No |
| Node Auto Scaling | No |
| Nodes | Supports a single node. |
| IP Family | Supports `ipv4`. |
| Limitations | For additional limitations that apply to all distributions, see Limitations. |
| Common Use Cases |
|
CMX supports creating Red Hat OpenShift OKD clusters, which is the community distribution of OpenShift, using CodeReady Containers (CRC).
OpenShift clusters are provisioned with two users:
- (Default) A
kubeadminuser withcluster-adminpriviledges. Use thekubeadminuser only for administrative tasks such as creating new users or setting roles. - A
developeruser with namespace-scoped priviledges. Thedeveloperuser can be used to better simulate access in end-customer environments.
By default, kubeconfig context is set to the kubeadmin user. To switch to the developer user, run the command oc login --username developer.
| Type | Description |
|---|---|
| Supported OpenShift Versions | {/* START_openshift_VERSIONS */}4.16.0-okd, 4.17.0-okd, 4.18.0-okd, 4.19.0-okd, 4.20.0-okd, 4.21.0-okd{/* END_openshift_VERSIONS */} |
| Supported Instance Types | See Replicated Instance Types |
| Node Groups | Yes |
| Node Auto Scaling | No |
| Nodes | Supports multiple nodes for versions 4.13.0-okd and later. |
| IP Family | Supports `ipv4`. |
| Limitations |
For additional limitations that apply to all distributions, see Limitations. |
| Common Use Cases | Customer release tests |
CMX supports creating clusters with Replicated Embedded Cluster. For more information, see Embedded Cluster Overview.
| Type | Description |
|---|---|
| Supported Embedded Cluster Versions | Any valid release sequence that has previously been promoted to the channel where the customer license is assigned. Version is optional and defaults to the latest available release on the channel. |
| Supported Instance Types | See Replicated Instance Types |
| Node Groups | Yes |
| Nodes | Supports multiple nodes (alpha). |
| IP Family | Supports `ipv4`. |
| Limitations |
For additional limitations that apply to all distributions, see Limitations. |
| Common Use Cases | Customer release tests |
CMX supports creating kURL clusters.
| Type | Description |
|---|---|
| Supported kURL Versions | Any promoted kURL installer. Version is optional. For an installer version other than "latest", you can find the specific Installer ID for a previously promoted installer under the relevant **Install Command** (ID after kurl.sh/) on the **Channels > kURL Installer History** page in the Vendor Portal. For more information about viewing the history of kURL installers promoted to a channel, see [Installer History](/vendor/installer-history). |
| Supported Instance Types | See Replicated Instance Types |
| Node Groups | Yes |
| Node Auto Scaling | No |
| Nodes | Supports multiple nodes. |
| IP Family | Supports `ipv4`. |
| Limitations | Does not work with the Longhorn add-on. For additional limitations that apply to all distributions, see Limitations. |
| Common Use Cases | Customer release tests |
This section lists the supported cloud clusters for compatibility testing.
CMX supports creating AWS EKS clusters.
| Type | Description |
|---|---|
| Supported Kubernetes Versions | {/* START_eks_VERSIONS */}1.30, 1.31, 1.32, 1.33, 1.34, 1.35{/* END_eks_VERSIONS */} Extended Support Versions: 1.30, 1.31, 1.32 |
| Supported Instance Types | m6i.large, m6i.xlarge, m6i.2xlarge, m6i.4xlarge, m6i.8xlarge, m7i.large, m7i.xlarge, m7i.2xlarge, m7i.4xlarge, m7i.8xlarge, m5.large, m5.xlarge, m5.2xlarge, m5.4xlarge, m5.8xlarge, m7g.large (arm), m7g.xlarge (arm), m7g.2xlarge (arm), m7g.4xlarge (arm), m7g.8xlarge (arm), c5.large, c5.xlarge, c5.2xlarge, c5.4xlarge, c5.9xlarge, g4dn.xlarge (gpu), g4dn.2xlarge (gpu), g4dn.4xlarge (gpu), g4dn.8xlarge (gpu), g4dn.12xlarge (gpu), g4dn.16xlarge (gpu) g4dn instance types depend on available capacity. After a g4dn cluster is running, you also need to install your version of the NVIDIA device plugin for Kubernetes. See [Amazon EKS optimized accelerated Amazon Linux AMIs](https://docs.aws.amazon.com/eks/latest/userguide/eks-optimized-ami.html#gpu-ami) in the AWS documentation. |
| Node Groups | Yes |
| Node Auto Scaling | Yes. Cost will be based on the max number of nodes. |
| Nodes | Supports multiple nodes. |
| IP Family | Supports `ipv4`. |
| Limitations | You can only choose a minor version, not a patch version. The EKS installer chooses the latest patch for that minor version. For additional limitations that apply to all distributions, see Limitations. |
| Common Use Cases | Customer release tests |
CMX supports creating Google GKE clusters.
| Type | Description |
|---|---|
| Supported Kubernetes Versions | {/* START_gke_VERSIONS */}1.32, 1.33, 1.34, 1.35{/* END_gke_VERSIONS */} |
| Supported Instance Types | n2-standard-2, n2-standard-4, n2-standard-8, n2-standard-16, n2-standard-32, t2a-standard-2 (arm), t2a-standard-4 (arm), t2a-standard-8 (arm), t2a-standard-16 (arm), t2a-standard-32 (arm), t2a-standard-48 (arm), e2-standard-2, e2-standard-4, e2-standard-8, e2-standard-16, e2-standard-32, n1-standard-1+nvidia-tesla-t4+1 (gpu), n1-standard-1+nvidia-tesla-t4+2 (gpu), n1-standard-1+nvidia-tesla-t4+4 (gpu), n1-standard-2+nvidia-tesla-t4+1 (gpu), n1-standard-2+nvidia-tesla-t4+2 (gpu), n1-standard-2+nvidia-tesla-t4+4 (gpu), n1-standard-4+nvidia-tesla-t4+1 (gpu), n1-standard-4+nvidia-tesla-t4+2 (gpu), n1-standard-4+nvidia-tesla-t4+4 (gpu), n1-standard-8+nvidia-tesla-t4+1 (gpu), n1-standard-8+nvidia-tesla-t4+2 (gpu), n1-standard-8+nvidia-tesla-t4+4 (gpu), n1-standard-16+nvidia-tesla-t4+1 (gpu), n1-standard-16+nvidia-tesla-t4+2 (gpu), n1-standard-16+nvidia-tesla-t4+4 (gpu), n1-standard-32+nvidia-tesla-t4+1 (gpu), n1-standard-32+nvidia-tesla-t4+2 (gpu), n1-standard-32+nvidia-tesla-t4+4 (gpu), n1-standard-64+nvidia-tesla-t4+1 (gpu), n1-standard-64+nvidia-tesla-t4+2 (gpu), n1-standard-64+nvidia-tesla-t4+4 (gpu), n1-standard-96+nvidia-tesla-t4+1 (gpu), n1-standard-96+nvidia-tesla-t4+2 (gpu), n1-standard-96+nvidia-tesla-t4+4 (gpu) You can specify more than one node. |
| Node Groups | Yes |
| Node Auto Scaling | Yes. Cost will be based on the max number of nodes. |
| Nodes | Supports multiple nodes. |
| IP Family | Supports `ipv4`. |
| Limitations | You can choose only a minor version, not a patch version. The GKE installer chooses the latest patch for that minor version. For additional limitations that apply to all distributions, see Limitations. |
| Common Use Cases | Customer release tests |
CMX supports creating Azure AKS clusters.
| Type | Description |
|---|---|
| Supported Kubernetes Versions | {/* START_aks_VERSIONS */}1.33, 1.34, 1.35{/* END_aks_VERSIONS */} |
| Supported Instance Types | Standard_B2ms, Standard_B4ms, Standard_B8ms, Standard_B16ms, Standard_D2S_v5, Standard_D4S_v5, Standard_D8S_v5, Standard_D16S_v5, Standard_D32S_v5, Standard_D48S_v5, Standard_D2ps_v5 (arm), Standard_D4ps_v5 (arm), Standard_D8ps_v5 (arm), Standard_D16ps_v5 (arm), Standard_D32ps_v5 (arm), Standard_D48ps_v5 (arm), Standard_NC4as_T4_v3 (gpu), Standard_NC8as_T4_v3 (gpu), Standard_NC16as_T4_v3 (gpu), Standard_NC64as_T4_v3 (gpu) GPU instance types depend on available capacity. After a GPU cluster is running, you also need to install your version of the NVIDIA device plugin for Kubernetes. See [NVIDIA GPU Operator with Azure Kubernetes Service](https://docs.nvidia.com/datacenter/cloud-native/gpu-operator/latest/microsoft-aks.html) in the NVIDIA documentation. |
| Node Groups | Yes |
| Node Auto Scaling | Yes. Cost will be based on the max number of nodes. |
| Nodes | Supports multiple nodes. |
| IP Family | Supports `ipv4`. |
| Limitations | You can choose only a minor version, not a patch version. The AKS installer chooses the latest patch for that minor version. For additional limitations that apply to all distributions, see Limitations. |
| Common Use Cases | Customer release tests |
CMX supports creating Oracle Container Engine for Kubernetes (OKE) clusters.
| Type | Description |
|---|---|
| Supported Kubernetes Versions | {/* START_oke_VERSIONS */}1.32.10, 1.33.1, 1.34.2, 1.35.0{/* END_oke_VERSIONS */} |
| Supported Instance Types | VM.Standard2.1, VM.Standard2.2, VM.Standard2.4, VM.Standard2.8, VM.Standard2.16, VM.Standard3.Flex.1, VM.Standard3.Flex.2, VM.Standard3.Flex.4, VM.Standard3.Flex.8, VM.Standard3.Flex.16, VM.Standard.A1.Flex.1 (arm), VM.Standard.A1.Flex.2 (arm), VM.Standard.A1.Flex.4 (arm), VM.Standard.A1.Flex.8 (arm), VM.Standard.A1.Flex.16 (arm) |
| Node Groups | Yes |
| Node Auto Scaling | No. |
| Nodes | Supports multiple nodes. |
| IP Family | Supports `ipv4`. |
| Limitations | Provising an OKE cluster does take between 8 to 10 minutes. If needed, some timeouts in your CI pipelines might have to be adjusted. For additional limitations that apply to all distributions, see Limitations. |
| Common Use Cases | Customer release tests |
The following VM types are supported:
| Distribution | Versions | Instance Types |
|---|---|---|
| ubuntu | 24.04, 22.04 | r1.small, r1.medium, r1.large, r1.xlarge, r1.2xlarge. See Replicated Instance Types. |
| almalinux | 8, 9, 10 | r1.small, r1.medium, r1.large, r1.xlarge, r1.2xlarge. See Replicated Instance Types. |
When creating a VM-based cluster with CMX, you must specify a Replicated instance type.
| Type | Memory (GiB) | VCPU Count |
|---|---|---|
| r1.small | 8 GB | 2 VCPUs |
| r1.medium | 16 GB | 4 VCPUs |
| r1.large | 32 GB | 8 VCPUs |
| r1.xlarge | 64 GB | 16 VCPUs |
| r1.2xlarge | 128 GB | 32 VCPUs |
We do not maintain forks or patches of the supported distributions. When a Kubernetes version in CMX is out of support (EOL), Replicated will attempt to continue to support this version for six months for compatibility testing to support customers who are running out-of-date versions of Kubernetes. In the event that a critical security issue or bug is found and unresolved, we might discontinue support for EOL versions of Kubernetes prior to 6 months post EOL.