This is the ioplane/docker-machine fork of the Nutanix Rancher node driver.
See the development plan and
compatibility contract before selecting a fork build.
Upstream installation URLs below describe the upstream release, not this fork.
This repository contains the Rancher Node Driver for Nutanix. Nutanix Node driver are used to provision hosts on Nutanix Enterprise Cloud, which Rancher uses to launch and manage Kubernetes clusters.
- Configure Prism Central and corresponding user to talk to Nutanix platform
- Define target cluster to deploy VM
- Ability to set a custom name for the newly created VM
- Ability to select VM's Main Memory in Megabytes
- Ability to select VM's vCPU count
- Ability to set the number of cores per vCPU
- Ability to specify the network(s) of the VM (Classic or VPC)
- Ability to specify the template disk in the VM by image name and modify his size (increase only)
- Ability to specify categories to applied to the VM ( flow, leap, ...)
- Ability to add one additional disk by specifying disk-size and storage-container
- Enable passthrough the host's CPU features to the newly created VM
- Define a Cloud-init user-data to send to the newly created VM
- Project support
- Serial Port support
- Boot type selection : Legacy or UEFI
- GPU support
- Prism Central Service Accounts support
If you want to use Nutanix Node Driver, you need add it in order to start using them to create node templates and eventually node pools for your Kubernetes cluster.
-
From the Home view, choose Cluster Management > Drivers in the navigation bar. From the Drivers page, select the Node Drivers tab.
-
Click Add Node Driver.
-
Complete the Add Node Driver form. Then click Create.
- Download URL:
https://github.com/nutanix/docker-machine/releases/download/v3.9.0/docker-machine-driver-nutanix - Custom UI URL:
https://nutanix.github.io/rancher-ui-driver/v3.9.0/component.js - Checksum:
dc1b9753534236830cbb64b3889bce09163dcfe239cf52d8e608c9aba4f67aac - Whitelist Domains:
nutanix.github.io
whitelist is mandatory and need to be changed if you relocate the UI driver
- Download URL:
- Wait for the driver to become "Active"
- Go to RKE1 Configuration > Node Templates, your can create a Nutanix Template and custom UI should show up.
| Arg | Description | Required | Default |
|---|---|---|---|
nutanix-endpoint |
The hostname/ip-address of the Prism Central | yes | |
nutanix-port |
The port to connect to Prism Central | no | 9440 |
nutanix-username |
The username of the nutanix management account | yes | |
nutanix-password |
The password of the nutanix management account | yes | |
nutanix-insecure |
Set to true to force SSL insecure connection | no | false |
nutanix-cluster |
The name of the cluster where deploy the VM (case sensitive) | yes | |
nutanix-boot-type |
The boot type of the VM (legacy or uefi) | no | legacy |
nutanix-vm-mem |
The amount of RAM of the newly created VM (MB) | no | 2 GB |
nutanix-vm-cpus |
The number of cpus in the newly created VM (core) | no | 2 |
nutanix-vm-cores |
The number of cores per vCPU | no | 1 |
nutanix-vm-nic-queues |
Positive decimal queues per NIC, at most CPUs × cores; empty leaves num_queues unset |
no | empty |
nutanix-vm-network |
The network(s) to which the VM is attached to ( name or UUID ) | yes | |
nutanix-vm-management-network |
Management subnet UUID identifying exactly one attached NIC; does not reorder NICs | no | first incoming NIC |
nutanix-vm-image |
The name of the Disk Image template we use for the newly created VM (must support cloud-init) | yes | |
nutanix-vm-image-size |
The new size of the Image we use as a template (in GiB) | no | |
nutanix-vm-categories |
The name of the categories who will be applied to the newly created VM | no | |
nutanix-vm-gpu |
The list of GPU device names to attach to the newly created VM (can be specified multiple times) | no | |
nutanix-project |
The name of the project where deploy the VM (default if empty) | no | default |
nutanix-disk-size |
The size of the additional disk to add to the VM (in GiB) | no | |
nutanix-storage-container |
The storage container UUID of the additional disk to add to the VM | no | |
nutanix-cloud-init |
Cloud-init to provide to the VM (will be patched with rancher root user) | no | |
nutanix-vm-cpu-passthrough |
Enable passthrough the host's CPU features to the newly created VM | no | false |
nutanix-vm-serial-port |
Attach a serial port to the newly created VM | no | false |
nutanix-vm-description |
The description of the newly created VM | no | VM created by Nutanix Rancher Node Driver |
For multi-NIC VMs, set nutanix-vm-management-network explicitly. The selector is
case-insensitive and must match exactly one resolved network before identity
allocation or VM submission. Omission retains the first incoming NIC, which may
differ from the original configuration order when a caller sorts repeated flags.
The selector persists in driver state and binds the Create journal; changing it
after reservation requires reconciliation and cannot rebind the existing operation.
Explicit nutanix-vm-nic-queues requires the existing COMPLETE VM readback to
contain the configured value on every NIC, with unique subnet identities, the
expected NIC count and the management subnet. UUID inputs are also matched
directly; name inputs retain their resolved management identity and count without
an extra inventory request. Missing or conflicting readback fails Create readiness
while retaining the allocated VM and journal. Recovery observes that VM without
resubmitting or deleting it. Omitted queues retain the previous readback behavior.
This verifies the Prism response; guest channel configuration requires separate validation.
Starting v3.3.0 the Rancher Node driver implements Nutanix Project support. The prerequisite needed to be able to use the Rancher Node Driver is the following:
- Target cluster and network available in the Project
- Role with the following recommended permission:
- VM Full Access
- Cluster View Access
- Image View Only
Starting v3.9.0 the Rancher Node Driver support Prism Central Service Accounts.
To use a Service Account, you need to provide X-ntnx-api-key as the user name and the corresponding API Key as the password.
The Rancher Node Driver supports attaching GPU devices to VMs. To use GPUs:
- Specify GPU devices by their name using the
nutanix-vm-gpuparameter - Multiple GPUs can be attached by specifying the parameter multiple times
- Only UNUSED GPUs from the target Prism Element cluster will be selected
- GPU names must match exactly with the GPU names available in the cluster
- The driver will search for available GPUs across all hosts in the specified cluster
Development tools run in the project's pinned Podman containers:
make dev-build lint-build
make checkmake build writes the static Linux binary to build/docker-machine-driver-nutanix.
make check runs tests, race detection, vet, golangci-lint v2.15.0, gofumpt checks,
govulncheck and the build. It requires no Prism or Rancher credentials.
The compatibility tests use an isolated subprocess and local HTTP fixtures.
make integration-build integration additionally exercises the real SDK through
one disposable Prism TLS container with Testcontainers v0.44.0. It requires an
already running rootless Podman service and PODMAN_SOCKET pointing at its
Unix socket. The test runner mounts this unprivileged socket only; rootful host
sockets are rejected. On SELinux hosts the test runner may require a per-container
label policy; host container configuration is not changed by these targets.
make integration-mod-check integration-quality verifies the separate integration
module, including all configured linters, formatting and test-aware govulncheck.
Ryuk is disabled for the isolated suite; CleanupContainer owns normal cleanup.
A killed test process can leave its fixture behind; use a disposable CI runner.
The pre-merge audit records remaining lifecycle risks. A passing mock suite is not a production acceptance result.
The queue flag applies to every NIC created by this driver. It does not change existing VM hardware or guest channel settings. Rancher2 provider/schema/module support and an approved isolated canary remain separate integration work.
- v1 is the original Nutanix docker machine driver that connect to Prism Element
- v2.x add Rancher 2.0 support
- v3.x is a rewrite of the driver that connect to Prism Central
This code is developed in the open with input from the community through issues and PRs. A Nutanix engineering team serves as the maintainer. Documentation is available in the project repository. Issues and enhancement requests can be submitted in the Issues tab of this repository. Please search for and review the existing open issues before submitting a new issue.
Copyright 2022-2023 Nutanix, Inc.
The project is released under Mozilla Public License Version 2.0.
Contribution and release rules are in CONTRIBUTING.md. The modernization plan defines the sequence and acceptance criteria; task state is maintained in the local um-docs Beads workspace.