Create a Node on a Private Cluster
Use this guide to launch a node on a private cluster assigned to your account. Private clusters are dedicated physical hosts provisioned specifically for your account. When you create a node on a private cluster, you define the CPU, RAM, and storage the node should use, and choose which host or hosts in the cluster are eligible for placement.
Private cluster access is not available by default. The option appears in the create-node flow only when your account has been provisioned with at least one active private cluster. If the option is not visible, your account uses the standard E2E cluster path. Contact support to inquire about private cluster access.
For the full node creation reference, see Create a node.
Before You Begin
Before creating a node on a private cluster:
| Requirement | Where to check |
|---|---|
| Private cluster access | Go to the Private Cluster section in MyAccount. At least one active cluster must be listed. |
| Available capacity | In the create-node flow, the host table shows available CPU and RAM for each host. Confirm that at least one host has enough capacity for your requested configuration. |
| OS selection | You must select a CPU or Linux-compatible OS before the private cluster option appears. Private cluster is not available for GPU nodes, Marketplace nodes, or Windows nodes. |
| Billing access | The account must have sufficient credits or billing approval to launch the node. |
Nodes on private clusters are billable resources. Review the Summary and Estimated Cost before launch.
Step 1: Open the Create-Node Flow and Select an OS
- Log in to MyAccount.
- Go to Compute > Nodes.
- Open the node creation flow.
- On the first screen, choose the CPU tab.
- Select a Linux OS from the list. The private cluster option does not appear for Windows, GPU, or Marketplace selections.
- Proceed to the plan screen.
Step 2: Choose Private Cluster
On the plan screen, a toggle appears at the top if your account has private cluster access:
| Option | What it means |
|---|---|
| E2E Cluster | Your node is placed on the shared E2E infrastructure. Standard plans are available. |
| Private Cluster | Your node is placed on dedicated physical hosts assigned to your account. You define the resource configuration instead of selecting a standard plan. |
Select Private Cluster. The standard instance type and plan selector is replaced by the private cluster configuration form.
If the toggle does not appear, your account is not enabled for private cluster deployments, or the selected OS is Windows, GPU, or Marketplace.
Step 3: Select a Cluster and Configure Resources
After choosing the Private Cluster option:
Choose a cluster
Select your private cluster from the dropdown. The portal lists all active clusters assigned to your account for the current location.
Choose a scheduling policy
Select how the portal distributes the node placement across eligible hosts in the cluster. Round Robin is the available policy, which distributes placement evenly across the selected hosts.
Configure CPU, RAM, and Storage
Set the resources the node should use:
| Resource | Minimum | Maximum | Notes |
|---|---|---|---|
| CPU | 1 core | Highest available cores on any eligible host | The slider adjusts to the available capacity after you select a cluster. |
| RAM | 4 GB | Highest available RAM on any eligible host | The slider adjusts to the available capacity after you select a cluster. |
| Storage | 100 GB | 2000 GB | Adjustable in 20 GB increments. |
The portal also offers quick configuration presets:
| Preset | CPU | RAM | Storage |
|---|---|---|---|
| Dev & Test | 2 cores | 4 GB | 100 GB |
| General Purpose | 4 cores | 8 GB | 100 GB |
| Mission Critical | 8 cores | 16 GB | 200 GB |
You can adjust values manually with the stepper controls or sliders after applying a preset.
Step 4: Select Hosts
After selecting a cluster and configuring resources, the portal shows an Available Hosts table. This table lists the physical hosts in the selected cluster and their current utilization.
| Column | What it shows |
|---|---|
| Host Name | The name of the physical host. |
| Nodes | Number of nodes currently running on this host. |
| CPU Usage | Used cores vs. total cores on the host. |
| RAM Usage | Used RAM vs. total RAM on the host, in GB. |
Hosts that do not have enough available CPU or RAM to satisfy your requested configuration are shown as disabled and cannot be selected.
Use the checkboxes to select one or more hosts that the scheduler may use for placement. By default, all eligible hosts are selected. The scheduling policy then places the node on one of the selected hosts when you launch.
Selecting multiple hosts means any of the selected hosts can receive the workload. The scheduling policy, not the selection, determines which host is used at launch time. Selecting fewer hosts constrains where the node can be placed.
If no hosts are available, either reduce the requested CPU or RAM until at least one host becomes selectable, or contact support to review cluster capacity.
Step 5: Configure Backup, Network, Security, and Access
After the host selection, the same configuration options available for standard nodes appear. These work the same way as for any CPU node.
Backup: Enable CDP Backup if the node needs recovery points. You can enable it later from node settings if needed.
VPC and Public IP: Use VPC for private network isolation. If the node must be reachable from the internet, confirm the public IP assignment in the Summary.
Public IPv4: Attach a specific primary public IP if the node must launch with a stable public IP. A specific public IP cannot be attached when launching multiple nodes.
Security group: Your account has a default security group that is pre-selected. Review and update it before launch to allow only the ports your workload requires.
SSH keys: Optional. Select one or more existing keys for key-based login. SSH access works the same as on standard Linux nodes.
Volume: Block storage volumes can be attached to private cluster nodes. Select an available volume if extra storage is needed beyond the configured root storage.
Advanced settings: BitNinja, encryption, IPv6, and start scripts work the same as for standard nodes.
Step 6: Launch the Node
Before clicking Launch, review the Summary panel:
- OS and version
- Cluster name and selected hosts
- CPU, RAM, and storage configuration
- Scheduling policy
- Backup status
- VPC and public IP
- Security group
- Estimated cost
Click Launch. The portal redirects to Compute > Nodes. The node provisions and reaches Running after placement completes. A secure password link is sent to the registered email address.
The password link expires after 168 hours. Save the password securely when you retrieve it.
Connect to the Node After Launch
Connect to a private cluster node the same way you connect to any Linux node.
- Go to Compute > Nodes.
- Find the node.
- Copy the primary public IP, Add-on IP, or reachable private IP.
- Connect with SSH:
ssh <username>@<node-ip>
For the full SSH connection guide including troubleshooting, see Connect to a Linux node.
View and Manage Your Private Clusters
You can view the clusters and hosts assigned to your account from the Private Cluster section in MyAccount.
Cluster listing
Go to the Private Cluster section in the left navigation. The listing shows:
- Cluster name
- Creation date
- Number of hosts in the cluster
Click a cluster row to open its detail view.
Cluster detail — Hosts tab
The Hosts tab shows each physical host in the cluster with its current CPU and RAM utilization, and the number of nodes running on it. Click a host to see the nodes placed on that host.
Cluster detail — Audit Logs tab
The Audit Logs tab shows a paginated log of operations performed on the cluster, such as CPU overcommit updates. Each log entry includes the host affected, the event type, the timestamp, and the IP of the operator.
CPU overcommit
Each physical host has a baseline CPU core count recorded at first use. You can increase the effective CPU capacity of a host by applying a CPU overcommit factor. The available factors are:
| Factor | Effect |
|---|---|
| 0.5 | Increases effective CPU by 50% above the physical core count. |
| 1.0 | Doubles the effective CPU (100% overcommit above the physical core count). |
The overcommit factor can only be increased, not decreased. Applying overcommit allows more nodes to be placed on the host, but nodes share the physical CPU threads. Use overcommit for workloads where CPU is not the sustained bottleneck.
Increasing CPU overcommit on a host affects all nodes already running on that host. Review the current workload before applying overcommit.
Important Notes
No Standard Plans
Private cluster nodes do not use standard instance plans. You configure CPU, RAM, and storage directly. The billing for a private cluster node is based on the host's contracted pricing, not the per-plan rate shown for standard nodes.
Limitations
| Limitation | Detail |
|---|---|
| OS | Linux only. Windows is not supported for private cluster node creation. |
| Node type | CPU only. GPU and Marketplace selections do not show the private cluster option. |
| Multiple nodes | Creating multiple identical nodes at once follows the same restriction as standard nodes for specific public IP and volume attachment. |
| Delete cluster | Cluster deletion is not available from the portal. Contact support if you need to decommission a cluster. |
Common Problems
| Problem | Likely cause | What to do |
|---|---|---|
| Private Cluster toggle does not appear | Your account is not enabled for private clusters, or the selected OS is Windows, GPU, or Marketplace. | Check the Private Cluster section. If no clusters are listed, contact support. |
| All hosts are disabled in the host table | No host has enough available CPU or RAM to meet the requested configuration. | Reduce the requested CPU or RAM, or contact support to review cluster capacity. |
| No private clusters listed when the toggle is selected | No active cluster exists for your account in the current location. | Check the Private Cluster section. If no clusters appear, contact support. |
| Node is launched but shows on a different host than expected | The scheduler places the node on one of the selected hosts based on the scheduling policy. Host selection constrains eligibility, not final placement. | Review the Hosts tab in the Private Cluster section to see which host the node was placed on. |
| SSH connection fails after launch | TCP 22 is blocked, the node has no reachable IP, or the credentials are wrong. | Check the public IP, security group inbound rules, and the password or SSH key. |
Automate This
Use the approved MyAccount API docs for node operations:
- MyAccount API Reference
- Node creation endpoint:
POST /nodes/
For private cluster nodes, the creation request includes a custom_sku object with the cpu, memory, disk, os_type, os_version, category, and cluster_id fields, along with host_ids for the selected hosts. Treat unpublished or portal-only fields as implementation details unless they are present in the approved public API documentation.
Related Resources
| Resource | Use it for |
|---|---|
| Create a node | Full node creation reference for standard CPU, GPU, and Marketplace nodes. |
| Choose a node family | Compare private cluster plans against standard node families. |
| Connect to a Linux node | Full SSH connection guide after launch. |
| Security Groups | Control inbound and outbound traffic. |
| VPC | Use private networking with nodes. |
| Static Public IPs | Manage primary public IPs or Add-on IPs when required. |
| Block Storage | Create and attach additional volumes. |
| CDP Backups | Back up and restore nodes. |
Next Step
After the node reaches Running, connect to it using SSH and configure the application. Review the Private Cluster section in MyAccount to monitor host utilization and confirm correct placement.