AzureVM (version 2.2.2)

create_vm: Create a new virtual machine or scaleset of virtual machines

Description

Method for the AzureRMR::az_subscription and AzureRMR::az_resource_group classes.

Value

For create_vm, an object of class az_vm_template representing the created VM. For create_vm_scaleset, an object of class az_vmss_template representing the scaleset.

Usage

## R6 method for class 'az_resource_group'
create_vm(name, login_user, size = "Standard_DS3_v2", config = "ubuntu_dsvm",
          managed_identity = TRUE, datadisks = numeric(0), ...,
          template, parameters, mode = "Incremental", wait = TRUE)
## R6 method for class 'az_subscription'
create_vm(name, ..., resource_group = name, location)
## R6 method for class 'az_resource_group'
create_vm_scaleset(name, login_user, instances, size = "Standard_DS1_v2",
                   config = "ubuntu_dsvm_ss", ...,
                   template, parameters, mode = "Incremental", wait = TRUE)
## R6 method for class 'az_subscription'
create_vm_scaleset(name, ..., resource_group = name, location)

Arguments

• name: The name of the VM or scaleset.

• location: For the subscription methods, the location for the VM or scaleset. Use the list_locations() method of the AzureRMR::az_subscription class to see what locations are available.

• resource_group: For the subscription methods, the resource group in which to place the VM or scaleset. Defaults to a new resource group with the same name as the VM.

• login_user: The details for the admin login account. An object of class user_config, obtained by a call to the user_config function.

• size: The VM (instance) size. Use the list_vm_sizes method to see what sizes are available.

• config: The VM or scaleset configuration. See 'Details' below for how to specify this. The default is to use an Ubuntu Data Science Virtual Machine.

• managed_identity: For create_vm, whether the VM should have a managed identity attached.

• datadisks: Any data disks to attach to the VM or scaleset. See 'Details' below.

• instances: For create_vm_scaleset, the initial number of instances in the scaleset.

• ... For the subscription methods, any of the other arguments listed here, which will be passed to the resource group method. For the resource group method, additional arguments to pass to the VM/scaleset configuration functions vm_config and vmss_config. See the examples below.

• template,parameters: The template definition and parameters to deploy. By default, these are constructed from the values of the other arguments, but you can supply your own template and/or parameters as well.

• wait: Whether to wait until the deployment is complete.

• mode: The template deployment mode. If "Complete", any existing resources in the resource group will be deleted.

Details

These methods deploy a template to create a new virtual machine or scaleset.

The config argument can be specified in the following ways:

• As the name of a supplied VM or scaleset configuration, like "ubuntu_dsvm" or "ubuntu_dsvm_ss". AzureVM comes with a number of supplied configurations to deploy commonly used images, which can be seen at vm_config and vmss_config. Any arguments in ... will be passed to the configuration, allowing you to customise the deployment.

• As a call to the vm_config or vmss_config functions, to deploy a custom VM image.

• As an object of class vm_config or vmss_config.

The data disks for the VM can be specified as either a vector of numeric disk sizes in GB, or as a list of datadisk_config objects, created via calls to the datadisk_config function. Currently, AzureVM only supports creating data disks at deployment time for single VMs, not scalesets.

You can also supply your own template definition and parameters for deployment, via the template and parameters arguments. See AzureRMR::az_template for information how to create templates.

The AzureRMR::az_subscription methods will by default create the VM in exclusive mode, meaning a new resource group is created solely to hold the VM or scaleset. This simplifies managing the VM considerably; in particular deleting the resource group will also automatically delete all the deployed resources.

See Also

az_vm_template, az_vmss_template

vm_config, vmss_config, user_config, datadisk_config

AzureRMR::az_subscription, AzureRMR::az_resource_group, Data Science Virtual Machine

Examples

# NOT RUN {
sub <- AzureRMR::get_azure_login()$
    get_subscription("subscription_id")

# default Ubuntu 18.04 VM:
# SSH key login, Standard_DS3_v2, publicly accessible via SSH
sub$create_vm("myubuntuvm", user_config("myname", "~/.ssh/id_rsa.pub"),
              location="australiaeast")

# Windows Server 2019, with a 500GB datadisk attached, not publicly accessible
sub$create_vm("mywinvm", user_config("myname", password="Use-strong-passwords!"),
              size="Standard_DS4_v2", config="windows_2019", datadisks=500, ip=NULL,
              location="australiaeast")

# Ubuntu DSVM, GPU-enabled
sub$create_vm("mydsvm", user_config("myname", "~/.ssh/id_rsa.pub"), size="Standard_NC12",
              config="ubuntu_dsvm_ss",
              location="australiaeast")

## custom VM configuration: Windows 10 Pro 1903 with data disks
## this assumes you have a valid Win10 desktop license
user <- user_config("myname", password="Use-strong-passwords!")
image <- image_config(
     publisher="MicrosoftWindowsDesktop",
     offer="Windows-10",
     sku="19h1-pro"
)
datadisks <- list(
    datadisk_config(250, type="Premium_LRS"),
    datadisk_config(1000, type="Standard_LRS")
)
nsg <- nsg_config(
    list(nsg_rule_allow_rdp)
)
config <- vm_config(
    image=image,
    keylogin=FALSE,
    datadisks=datadisks,
    nsg=nsg,
    properties=list(licenseType="Windows_Client")
)
sub$create_vm("mywin10vm", user, size="Standard_DS2_v2", config=config,
              location="australiaeast")


# default Ubuntu scaleset:
# load balancer and autoscaler enabled, Standard_DS1_v2
sub$create_vm_scaleset("mydsvmss", user_config("myname", "~/.ssh/id_rsa.pub"),
                       instances=5,
                       location="australiaeast"))

# Ubuntu DSVM scaleset with public GPU-enabled instances, no load balancer or autoscaler
sub$create_vm_scaleset("mydsvmss", user_config("myname", "~/.ssh/id_rsa.pub"),
                       instances=5, size="Standard_NC12", config="ubuntu_dsvm_ss",
                       options=scaleset_options(public=TRUE),
                       load_balancer=NULL, autoscaler=NULL,
                       location="australiaeast")

# RHEL scaleset, allow http/https access
sub$create_vm_scaleset("myrhelss", user_config("myname", "~/.ssh/id_rsa.pub"),
                        instances=5, config="rhel_8_ss",
                        nsg=nsg_config(list(nsg_rule_allow_http, nsg_rule_allow_https)),
                        location="australiaeast")

# Large Debian scaleset, using low-priority (spot) VMs
# need to set the instance size to something that supports low-pri
sub$create_vm_scaleset("mydebss", user_config("myname", "~/.ssh/id_rsa.pub"),
                       instances=50, size="Standard_DS3_v2", config="debian_9_backports_ss",
                       options=scaleset_options(priority="spot", large_scaleset=TRUE),
                       location="australiaeast")


## VM and scaleset in the same resource group and virtual network
# first, create the resgroup
rg <- sub$create_resource_group("rgname", "australiaeast")

# create the master
rg$create_vm("mastervm", user_config("myname", "~/.ssh/id_rsa.pub"))

# get the vnet resource
vnet <- rg$get_resource(type="Microsoft.Network/virtualNetworks", name="mastervm-vnet")

# create the scaleset
rg$create_vm_scaleset("slavess", user_config("myname", "~/.ssh/id_rsa.pub"),
                      instances=5, vnet=vnet, nsg=NULL, load_balancer=NULL, autoscaler=NULL)

# }