README.md

Description

This module creates a GPU accelerated virtual machine that can be accessed using Chrome Remote Desktop.

Note: This is an experimental module. This module has only been tested in limited capacity with the Cluster Toolkit. The module interface may have undergo breaking changes in the future.

Example

The following example will create a single GPU accelerated remote desktop.

  - id: remote-desktop
    source: community/modules/remote-desktop/chrome-remote-desktop
    use: [network1]
    settings:
      install_nvidia_driver: true

Setting up the Remote Desktop

1. Once the remote desktop has been deployed, navigate to https://remotedesktop.google.com/headless.
2. Click through Begin, Next, & Authorize.
3. Copy the code snippet for Debian Linux.
4. SSH into the remote desktop machine. It will be listed under VM Instances in the Google Cloud web console.
5. Run the copied command and follow instructions to set up a PIN.
6. You should now see your machine listed on the Chrome Remote Desktop page under Remote devices.
7. Click on your machine and enter PIN if prompted.

License

Copyright 2022 Google LLC

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

 http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.

Requirements

Name	Version
terraform	>= 0.12.31

Providers

No providers.

Modules

Name	Source	Version
client_startup_script	github.com/GoogleCloudPlatform/hpc-toolkit//modules/scripts/startup-script	v1.39.0&depth=1
instances	github.com/GoogleCloudPlatform/hpc-toolkit//modules/compute/vm-instance	09ae2725

Resources

No resources.

Inputs

Name	Description	Type	Default	Required
add_deployment_name_before_prefix	If true, the names of VMs and disks will always be prefixed with deployment_name to enable uniqueness across deployments. See name_prefix for further details on resource naming behavior.	bool	false	no
auto_delete_boot_disk	Controls if boot disk should be auto-deleted when instance is deleted.	bool	true	no
bandwidth_tier	Tier 1 bandwidth increases the maximum egress bandwidth for VMs. Using the tier_1_enabled setting will enable both gVNIC and TIER_1 higher bandwidth networking. Using the gvnic_enabled setting will only enable gVNIC and will not enable TIER_1. Note that TIER_1 only works with specific machine families & shapes and must be using an image th at supports gVNIC. See official docs for more details.	string	"not_enabled"	no
deployment_name	Cluster Toolkit deployment name. Cloud resource names will include this value.	string	n/a	yes
disk_size_gb	Size of disk for instances.	number	200	no
disk_type	Disk type for instances.	string	"pd-balanced"	no
enable_oslogin	Enable or Disable OS Login with "ENABLE" or "DISABLE". Set to "INHERIT" to inherit project OS Login setting.	string	"ENABLE"	no
enable_public_ips	If set to true, instances will have public IPs on the internet.	bool	true	no
guest_accelerator	List of the type and count of accelerator cards attached to the instance. Requires virtual workstation accelerator if Nvidia Grid Drivers are required	list(object({ type = string, count = number }))	[ { "count": 1, "type": "nvidia-tesla-t4-vws" } ]	no
install_nvidia_driver	Installs the nvidia driver (true/false). For details, see https://cloud.google.com/compute/docs/gpus/install-drivers-gpu	bool	n/a	yes
instance_count	Number of instances	number	1	no
instance_image	Image used to build chrome remote desktop node. The default image is name="debian-12-bookworm-v20240815" and project="debian-cloud". NOTE: uses fixed version of image to avoid NVIDIA driver compatibility issues. An alternative image is from name="ubuntu-2204-jammy-v20240126" and project="ubuntu-os-cloud". Expected Fields: name: The name of the image. Mutually exclusive with family. family: The image family to use. Mutually exclusive with name. project: The project where the image is hosted.	map(string)	{ "name": "debian-12-bookworm-v20240815", "project": "debian-cloud" }	no
labels	Labels to add to the instances. Key-value pairs.	map(string)	{}	no
machine_type	Machine type to use for the instance creation. Must be N1 family if GPU is used.	string	"n1-standard-8"	no
metadata	Metadata, provided as a map	map(string)	{}	no
name_prefix	An optional name for all VM and disk resources. If not supplied, deployment_name will be used. When name_prefix is supplied, and add_deployment_name_before_prefix is set, then resources are named by "<deployment_name>-<name_prefix>-<#>".	string	null	no
network_interfaces	A list of network interfaces. The options match that of the terraform network_interface block of google_compute_instance. For descriptions of the subfields or more information see the documentation: https://registry.terraform.io/providers/hashicorp/google/latest/docs/resources/compute_instance#nested_network_interface _NOTE:_ If network_interfaces are set, network_self_link and subnetwork_self_link will be ignored, even if they are provided through the use field. bandwidth_tier and enable_public_ips also do not apply to network interfaces defined in this variable. Subfields: network (string, required if subnetwork is not supplied) subnetwork (string, required if network is not supplied) subnetwork_project (string, optional) network_ip (string, optional) nic_type (string, optional, choose from ["GVNIC", "VIRTIO_NET"]) stack_type (string, optional, choose from ["IPV4_ONLY", "IPV4_IPV6"]) queue_count (number, optional) access_config (object, optional) ipv6_access_config (object, optional) alias_ip_range (list(object), optional)	list(object({ network = string, subnetwork = string, subnetwork_project = string, network_ip = string, nic_type = string, stack_type = string, queue_count = number, access_config = list(object({ nat_ip = string, public_ptr_domain_name = string, network_tier = string })), ipv6_access_config = list(object({ public_ptr_domain_name = string, network_tier = string })), alias_ip_range = list(object({ ip_cidr_range = string, subnetwork_range_name = string })) }))	[]	no
network_self_link	The self link of the network to attach the VM.	string	"default"	no
network_storage	An array of network attached storage mounts to be configured.	list(object({ server_ip = string, remote_mount = string, local_mount = string, fs_type = string, mount_options = string, client_install_runner = map(string) mount_runner = map(string) }))	[]	no
on_host_maintenance	Describes maintenance behavior for the instance. If left blank this will default to MIGRATE except for when placement_policy, spot provisioning, or GPUs require it to be TERMINATE	string	"TERMINATE"	no
project_id	Project in which Google Cloud resources will be created	string	n/a	yes
region	Default region for creating resources	string	n/a	yes
service_account	Service account to attach to the instance. See https://www.terraform.io/docs/providers/google/r/compute_instance_template.html#service_account.	object({ email = string, scopes = set(string) })	{ "email": null, "scopes": [ "https://www.googleapis.com/auth/cloud-platform" ] }	no
spot	Provision VMs using discounted Spot pricing, allowing for preemption	bool	false	no
startup_script	Startup script used on the instance	string	null	no
subnetwork_self_link	The self link of the subnetwork to attach the VM.	string	null	no
tags	Network tags, provided as a list	list(string)	[]	no
threads_per_core	Sets the number of threads per physical core	number	2	no
zone	Default zone for creating resources	string	n/a	yes

Outputs

Name	Description
instance_name	Name of the first instance created, if any.
startup_script	script to load and run all runners, as a string value.