Proxmox VE Clusters - Ankra Platform

Ankra supports provisioning fully managed Kubernetes clusters on Proxmox VE. You bring your own Proxmox nodes, storage, and network bridge - Ankra creates the VMs, installs Kubernetes, and manages the full cluster lifecycle: node groups, scaling, upgrades, and deprovisioning.

Closed beta. Proxmox VE cluster provisioning is in closed beta. The workflow is stable but the surface may still change, and it is enabled per organisation on request. Contact support to have it turned on for your organisation.

Prerequisites

Before creating a Proxmox VE cluster, you need two credentials:

Proxmox VE API Credential

The Proxmox API URL and an API token with VM management privileges. See Proxmox VE Credentials.

SSH Key Credential

An SSH public key for VM access. You can provide your own or let Ankra generate one. See SSH Key Credentials.

Your Proxmox environment must also provide:

A cloud-init VM template (e.g., Ubuntu 24.04) with the QEMU guest agent installed, registered on the node you deploy to. Ankra clones this template for every cluster VM and reads each VM’s IP address through the guest agent. You can pin a specific template with the template option; otherwise the first template on the node is used.
An existing network bridge with DHCP. Ankra does not create or modify bridges - you select one at creation, and all cluster VMs get their addresses from DHCP on that bridge.
Storage with free space on the Proxmox node you deploy to, for the VM disks.
API reachability. The Proxmox API must be reachable from Ankra either directly or through an SSH jumphost.

Hybrid Connectivity (SSH Jumphost)

Many Proxmox environments are not directly reachable from the internet. If the Proxmox API and the VM network cannot be reached from Ankra, attach an SSH jumphost (host, port, username, and private key) to your Proxmox VE credential. Ankra then tunnels both the Proxmox API calls and the SSH connections to your cluster nodes through the jumphost. If your Proxmox API uses a self-signed certificate, enable the TLS insecure toggle (tls_insecure) on the credential. Both options are configured on the credential - see Proxmox VE Credentials.

Creating a Proxmox VE Cluster

Via the Platform UI

Navigate to Clusters

Go to Clusters in the Ankra dashboard and click Create Cluster.

Select Proxmox VE

Choose Proxmox VE as the provider.

Select Credentials

Pick your Proxmox VE API credential and SSH key credential from the dropdowns. You can also create new credentials directly from the wizard.

Choose Placement

Select the Proxmox node to host the cluster VMs, the storage for VM disks, and the network bridge the VMs attach to. The bridge must provide DHCP.

Configure Nodes

Set your cluster topology:

Bastion - Instance size for the SSH bastion VM (e.g., px-small)
Control Plane - Count (1, 3, or 5) and size (e.g., px-medium)
Workers - Count and size (e.g., 2× px-medium)

The wizard shows vCPUs, memory, and disk for each size. Proxmox has no list pricing, so no cost estimate is shown.

Choose Distribution

Pick the Kubernetes distribution: k3s Lightweight Kubernetes with a user-selectable CNI (Flannel default, Calico, or Cilium). kubeadm Vanilla upstream Kubernetes bootstrapped with kubeadm and containerd. kubeadm clusters always use Cilium CNI and optionally support an external etcd topology with dedicated etcd VMs.See Kubernetes Distribution for details.

Create & Track Progress

Click Create to start provisioning. A live progress view tracks credential setup, SSH key deployment, bastion provisioning, VM creation, Kubernetes installation (k3s or kubeadm), and Ankra Agent setup. The cluster appears with an offline state until provisioning completes, then transitions to online.

Via the API

Proxmox VE clusters are managed from the portal or API during the closed beta - the ankra CLI does not yet include Proxmox commands. Create the Proxmox VE credential and SSH key credential from the portal first.

curl -X POST https://platform.ankra.app/api/v1/clusters/proxmox \
  -H "Authorization: Bearer $ANKRA_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "my-cluster",
    "credential_id": "<proxmox-credential-id>",
    "ssh_key_credential_id": "<ssh-key-credential-id>",
    "node": "pve01",
    "storage": "local-lvm",
    "bridge": "vmbr0",
    "control_plane_count": 1,
    "control_plane_instance_type": "px-medium",
    "node_groups": [
      {"name": "default", "instance_type": "px-medium", "count": 2}
    ],
    "distribution": "k3s"
  }'

Cluster Configuration Options

Parameter	Default	Description
`name`	required	Unique cluster name
`credential_id`	required	Proxmox VE API credential ID
`ssh_key_credential_id`	required	SSH key credential ID
`node`	required	Proxmox node that hosts the cluster VMs
`bridge`	required	Existing network bridge the VMs attach to (must provide DHCP)
`storage`	`local-lvm`	Proxmox storage for VM disks
`template`	auto-picked	Cloud-init template to clone, by VMID or name. Defaults to the first template on the node
`bastion_instance_type`	`px-small`	Instance size for the bastion VM
`control_plane_count`	`1`	Number of control plane nodes (1, 3, or 5)
`control_plane_instance_type`	`px-medium`	Instance size for control planes
`worker_count`	`1`	Number of worker nodes (legacy, use `node_groups` instead)
`worker_instance_type`	`px-medium`	Instance size for workers (legacy, use `node_groups` instead)
`node_groups`		Array of node group definitions (see Node Groups)
`distribution`	`k3s`	Kubernetes distribution (`k3s` or `kubeadm`)
`kubernetes_version`	latest stable	Kubernetes version (optional)
`cni`	`flannel` (k3s)	CNI plugin (`flannel`, `calico`, or `cilium`). kubeadm clusters always use `cilium`
`etcd_topology`	`stacked`	kubeadm only. `stacked` (etcd on control planes) or `external` (dedicated etcd VMs)
`etcd_node_count`	`3`	kubeadm `external` topology only. Number of dedicated etcd VMs (3 or 5)
`etcd_instance_type`	`px-medium`	kubeadm `external` topology only. Instance size for dedicated etcd nodes

Instance Sizes

Proxmox VE clusters use fixed instance size presets:

Size	vCPUs	RAM	Disk	Typical use
`px-small`	2	4 GB	40 GB	Bastion
`px-medium`	4	8 GB	80 GB	Control plane, general workers
`px-large`	8	16 GB	160 GB	Larger workloads
`px-xlarge`	16	32 GB	320 GB	Heavy workloads

Proxmox VE has no list pricing, so Ankra does not show cost estimates for Proxmox clusters - neither in the creation wizard nor in Cloud Cost.

Networking

Ankra does not create networks or bridges on Proxmox. You select an existing bridge at creation, and all cluster VMs receive their IP addresses via DHCP on that bridge. Make sure the bridge has a DHCP server available before creating the cluster.

Kubernetes Distribution

Proxmox VE clusters can be provisioned with either k3s (default) or kubeadm.

	k3s	kubeadm
Kubernetes	Lightweight, single-binary distribution	Vanilla upstream Kubernetes
CNI	User-selectable (Flannel default, Calico, or Cilium)	Cilium (fixed, cannot be changed after creation)
etcd	Embedded	`stacked` (on control planes) or `external` (dedicated VMs)
Version format	`v1.35.1+k3s1`	`v1.31.0` (plain upstream tag)

kubeadm clusters always use Cilium CNI (eBPF-based networking, L7 policies, Hubble observability). The CNI cannot be changed after creation.

External etcd topology (kubeadm)

By default kubeadm runs etcd stacked on the control plane nodes. For larger clusters you can run etcd on dedicated VMs by setting etcd_topology to external:

curl -X POST https://platform.ankra.app/api/v1/clusters/proxmox \
  -H "Authorization: Bearer $ANKRA_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "my-cluster",
    "credential_id": "<proxmox-credential-id>",
    "ssh_key_credential_id": "<ssh-key-credential-id>",
    "node": "pve01",
    "storage": "local-lvm",
    "bridge": "vmbr0",
    "control_plane_count": 3,
    "control_plane_instance_type": "px-medium",
    "node_groups": [
      {"name": "default", "instance_type": "px-medium", "count": 2}
    ],
    "distribution": "kubeadm",
    "etcd_topology": "external",
    "etcd_node_count": 3,
    "etcd_instance_type": "px-medium"
  }'

Node Groups

Node groups let you organize worker nodes into logical groups with independent instance sizes, counts, labels, and taints. Manage node groups from Settings > Nodes in the dashboard, or via the API.

Node Group API Reference

Endpoint	Method	Description
`/api/v1/clusters/proxmox/{id}/node-groups`	GET	List all node groups
`/api/v1/clusters/proxmox/{id}/node-groups`	POST	Add a node group
`/api/v1/clusters/proxmox/{id}/node-groups/{name}/scale`	PUT	Scale a node group (0-100)
`/api/v1/clusters/proxmox/{id}/node-groups/{name}/instance-type`	PUT	Change instance size
`/api/v1/clusters/proxmox/{id}/node-groups/{name}/labels`	PUT	Update labels
`/api/v1/clusters/proxmox/{id}/node-groups/{name}/taints`	PUT	Update taints
`/api/v1/clusters/proxmox/{id}/node-groups/{name}`	DELETE	Delete a node group

For detailed usage examples, see Hetzner Node Groups - the API is identical across all providers.

Legacy Worker Scaling

The legacy scale-workers and worker-count endpoints operate on all workers as a single pool.

curl https://platform.ankra.app/api/v1/clusters/proxmox/<cluster_id>/worker-count \
  -H "Authorization: Bearer $ANKRA_API_TOKEN"

curl -X POST https://platform.ankra.app/api/v1/clusters/proxmox/<cluster_id>/scale-workers \
  -H "Authorization: Bearer $ANKRA_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"worker_count": 4}'

Prefer using Node Groups for more granular control.

Upgrading Kubernetes Version

You can upgrade the Kubernetes version on all nodes in a Proxmox VE cluster. Upgrades are applied to control plane nodes first, then workers. Both k3s and kubeadm clusters are supported.

Both k3s and kubeadm clusters are supported for version upgrades, including kubeadm clusters with an external etcd topology: the dedicated etcd members are upgraded first, one at a time, each saving a pre-upgrade snapshot before its static pod rolls to the etcd image matching the target Kubernetes version.
Use the matching version format for the target: v1.35.1+k3s1 for k3s, or a plain v1.31.0 upstream tag for kubeadm.
Downgrades are not supported - downgrades require an etcd snapshot restore.
You can only upgrade one minor version at a time (e.g., v1.33.x to v1.34.x, not v1.33.x to v1.35.x).
The cluster must be online with no active operations.

Check Current Version

curl https://platform.ankra.app/api/v1/clusters/proxmox/<cluster_id>/k8s-version \
  -H "Authorization: Bearer $ANKRA_API_TOKEN"

Upgrade Version

curl -X POST https://platform.ankra.app/api/v1/clusters/proxmox/<cluster_id>/upgrade-k8s-version \
  -H "Authorization: Bearer $ANKRA_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"target_version": "v1.35.1+k3s1"}'

For a kubeadm cluster, use a plain upstream tag instead (for example "target_version": "v1.31.0").

Deprovisioning

Deprovisioning deletes all VMs Ankra created on your Proxmox nodes (bastion, control planes, workers, and dedicated etcd VMs if any) and removes the cluster from Ankra. Your Proxmox nodes, storage, and bridges are left untouched.

This action is irreversible. All data on the cluster will be permanently deleted.

curl -X DELETE https://platform.ankra.app/api/v1/clusters/proxmox/<cluster_id> \
  -H "Authorization: Bearer $ANKRA_API_TOKEN"

Architecture

A Proxmox VE cluster provisions the following infrastructure:

Component	Description
Bastion VM	Jump server for secure SSH access to cluster nodes
Control Plane(s)	Kubernetes control plane VMs (1, 3, or 5)
Worker(s)	Kubernetes worker VMs organized in node groups
etcd Node(s)	Dedicated etcd VMs, only for kubeadm clusters with an `external` etcd topology
SSH Keys	Deployed to all VMs for access

All VMs are QEMU virtual machines on the Proxmox node you selected, attached to your existing bridge with DHCP addressing. The bastion VM provides the only SSH access point Ankra uses to reach the cluster nodes. When an SSH jumphost is configured, Ankra reaches both the Proxmox API and the bastion through the jumphost.

Proxmox VE clusters do not include a cloud controller manager or load balancer integration - external_cloud_provider is not supported. Kubernetes LoadBalancer services are not provisioned automatically; expose workloads with NodePort services, an ingress controller, or a load balancer solution you deploy yourself.

Troubleshooting

Common Issues

Issue	Solution
Proxmox API unreachable from Ankra	Attach an SSH jumphost to the credential - Ankra tunnels API calls and node SSH through it
TLS certificate errors (self-signed)	Enable the TLS insecure toggle (`tls_insecure`) on the Proxmox VE credential
Nodes never get an IP address	The selected bridge must provide DHCP - Ankra does not manage IP allocation. Verify a DHCP server answers on that bridge
Provisioning fails waiting for a VM’s IP address	The cloned template must have the QEMU guest agent installed and enabled - Ankra reads VM addresses through the agent. Rebuild the template with `qemu-guest-agent` and retry
No VM templates found on the node	Create a cloud-init template on the selected Proxmox node first - Ankra clones it for every cluster VM
Cluster stuck in provisioning	Check the API token privileges (see Proxmox VE Credentials) and free capacity on the selected node and storage
VM creation fails with a permission error	The API token is missing privileges (e.g., `VM.Allocate`, `Datastore.AllocateSpace`) on the selected node, storage, or bridge
LoadBalancer service stuck in `Pending`	Proxmox clusters have no cloud controller manager - use `NodePort`, ingress, or your own load balancer
No cost estimates shown	Expected - Proxmox has no list pricing, so cost estimation is not supported

​Prerequisites

Proxmox VE API Credential

SSH Key Credential

​Hybrid Connectivity (SSH Jumphost)

​Creating a Proxmox VE Cluster

​Via the Platform UI

​Via the API

​Cluster Configuration Options

​Instance Sizes

​Networking

​Kubernetes Distribution

​External etcd topology (kubeadm)

​Node Groups

​Node Group API Reference

​Legacy Worker Scaling

​Upgrading Kubernetes Version

​Check Current Version

​Upgrade Version

​Deprovisioning

​Architecture

​Troubleshooting

​Common Issues

Prerequisites

Hybrid Connectivity (SSH Jumphost)

Creating a Proxmox VE Cluster

Via the Platform UI

Via the API

Cluster Configuration Options

Instance Sizes

Networking

Kubernetes Distribution

External etcd topology (kubeadm)

Node Groups

Node Group API Reference

Legacy Worker Scaling

Upgrading Kubernetes Version

Check Current Version

Upgrade Version

Deprovisioning

Architecture

Troubleshooting

Common Issues