terraform-azurerm-openai

Azure OpenAI Terraform Module and Samples

Requirements

Name	Version
terraform	>= 1.3.0
azurerm	~> 3.80
modtm	>= 0.1.8, < 1.0
random	>= 3.0

Providers

Name	Version
azurerm	~> 3.80
modtm	>= 0.1.8, < 1.0
random	>= 3.0

Modules

No modules.

Resources

Name	Type
azurerm_cognitive_account.this	resource
azurerm_cognitive_deployment.this	resource
azurerm_monitor_diagnostic_setting.setting	resource
azurerm_private_dns_zone.dns_zone	resource
azurerm_private_dns_zone_virtual_network_link.dns_zone_link	resource
azurerm_private_endpoint.this	resource
modtm_telemetry.this	resource
random_integer.this	resource
azurerm_private_dns_zone.dns_zone	data source
azurerm_resource_group.pe_vnet_rg	data source
azurerm_resource_group.this	data source
azurerm_subnet.pe_subnet	data source
azurerm_virtual_network.vnet	data source

Inputs

Name	Description	Type	Default	Required
account_name	Specifies the name of the Cognitive Service Account. Changing this forces a new resource to be created. Leave this variable as default would use a default name with random suffix.	string	""	no
application_name	Name of the application. A corresponding tag would be created on the created resources if var.default_tags_enabled is true.	string	""	no
custom_subdomain_name	The subdomain name used for token-based authentication. Changing this forces a new resource to be created. Leave this variable as default would use a default name with random suffix.	string	""	no
customer_managed_key	type = object({ key_vault_key_id = (Required) The ID of the Key Vault Key which should be used to Encrypt the data in this OpenAI Account. identity_client_id = (Optional) The Client ID of the User Assigned Identity that has access to the key. This property only needs to be specified when there're multiple identities attached to the OpenAI Account. })	object({ key_vault_key_id = string identity_client_id = optional(string) })	null	no
default_tags_enabled	Determines whether or not default tags are applied to resources. If set to true, tags will be applied. If set to false, tags will not be applied.	bool	false	no
deployment	type = map(object({ name = (Required) The name of the Cognitive Services Account Deployment. Changing this forces a new resource to be created. cognitive_account_id = (Required) The ID of the Cognitive Services Account. Changing this forces a new resource to be created. model = { model_format = (Required) The format of the Cognitive Services Account Deployment model. Changing this forces a new resource to be created. Possible value is OpenAI. model_name = (Required) The name of the Cognitive Services Account Deployment model. Changing this forces a new resource to be created. model_version = (Required) The version of Cognitive Services Account Deployment model. } scale = { scale_type = (Required) Deployment scale type. Possible value is Standard. Changing this forces a new resource to be created. } rai_policy_name = (Optional) The name of RAI policy. Changing this forces a new resource to be created. capacity = (Optional) Tokens-per-Minute (TPM). The unit of measure for this field is in the thousands of Tokens-per-Minute. Defaults to 1 which means that the limitation is 1000 tokens per minute. version_upgrade_option = (Optional) Deployment model version upgrade option. Possible values are OnceNewDefaultVersionAvailable, OnceCurrentVersionExpired, and NoAutoUpgrade. Defaults to OnceNewDefaultVersionAvailable. Changing this forces a new resource to be created. }))	map(object({ name = string model_format = string model_name = string model_version = string scale_type = string rai_policy_name = optional(string) capacity = optional(number) version_upgrade_option = optional(string) }))	{}	no
diagnostic_setting	A map of objects that represent the configuration for a diagnostic setting." type = map(object({ name = (Required) Specifies the name of the diagnostic setting. Changing this forces a new resource to be created. log_analytics_workspace_id = (Optional) (Optional) Specifies the resource id of an Azure Log Analytics workspace where diagnostics data should be sent. log_analytics_destination_type = (Optional) Possible values are AzureDiagnostics and Dedicated. When set to Dedicated, logs sent to a Log Analytics workspace will go into resource specific tables, instead of the legacy AzureDiagnostics table. eventhub_name = (Optional) Specifies the name of the Event Hub where diagnostics data should be sent. eventhub_authorization_rule_id = (Optional) Specifies the resource id of an Event Hub Namespace Authorization Rule used to send diagnostics data. storage_account_id = (Optional) Specifies the resource id of an Azure storage account where diagnostics data should be sent. partner_solution_id = (Optional) The resource id of the market partner solution where diagnostics data should be sent. For potential partner integrations, click to learn more about partner integration. audit_log_retention_policy = (Optional) Specifies the retention policy for the audit log. This is a block with the following properties: enabled = (Optional) Specifies whether the retention policy is enabled. If enabled, days must be a positive number. days = (Optional) Specifies the number of days to retain trace logs. If enabled is set to true, this value must be set to a positive number. request_response_log_retention_policy = (Optional) Specifies the retention policy for the request response log. This is a block with the following properties: enabled = (Optional) Specifies whether the retention policy is enabled. If enabled, days must be a positive number. days = (Optional) Specifies the number of days to retain trace logs. If enabled is set to true, this value must be set to a positive number. trace_log_retention_policy = (Optional) Specifies the retention policy for the trace log. This is a block with the following properties: enabled = (Optional) Specifies whether the retention policy is enabled. If enabled, days must be a positive number. days = (Optional) Specifies the number of days to retain trace logs. If enabled is set to true, this value must be set to a positive number. metric_retention_policy = (Optional) Specifies the retention policy for the metric. This is a block with the following properties: enabled = (Optional) Specifies whether the retention policy is enabled. If enabled, days must be a positive number. days = (Optional) Specifies the number of days to retain trace logs. If enabled is set to true, this value must be set to a positive number. }))	map(object({ name = string log_analytics_workspace_id = optional(string) log_analytics_destination_type = optional(string) eventhub_name = optional(string) eventhub_authorization_rule_id = optional(string) storage_account_id = optional(string) partner_solution_id = optional(string) audit_log_retention_policy = optional(object({ enabled = optional(bool, true) days = optional(number, 7) })) request_response_log_retention_policy = optional(object({ enabled = optional(bool, true) days = optional(number, 7) })) trace_log_retention_policy = optional(object({ enabled = optional(bool, true) days = optional(number, 7) })) metric_retention_policy = optional(object({ enabled = optional(bool, true) days = optional(number, 7) })) }))	{}	no
dynamic_throttling_enabled	Determines whether or not dynamic throttling is enabled. If set to true, dynamic throttling will be enabled. If set to false, dynamic throttling will not be enabled.	bool	null	no
environment	Environment of the application. A corresponding tag would be created on the created resources if var.default_tags_enabled is true.	string	""	no
fqdns	List of FQDNs allowed for the Cognitive Account.	list(string)	null	no
identity	type = object({ type = (Required) The type of the Identity. Possible values are SystemAssigned, UserAssigned, SystemAssigned, UserAssigned. identity_ids = (Optional) Specifies a list of User Assigned Managed Identity IDs to be assigned to this OpenAI Account. })	object({ type = string identity_ids = optional(list(string)) })	null	no
local_auth_enabled	Whether local authentication methods is enabled for the Cognitive Account. Defaults to true.	bool	true	no
location	Azure OpenAI deployment region. Set this variable to null would use resource group's location.	string	n/a	yes
network_acls	type = set(object({ default_action = (Required) The Default Action to use when no rules match from ip_rules / virtual_network_rules. Possible values are Allow and Deny. ip_rules = (Optional) One or more IP Addresses, or CIDR Blocks which should be able to access the Cognitive Account. virtual_network_rules = optional(set(object({ subnet_id = (Required) The ID of a Subnet which should be able to access the OpenAI Account. ignore_missing_vnet_service_endpoint = (Optional) Whether ignore missing vnet service endpoint or not. Default to false. }))) }))	set(object({ default_action = string ip_rules = optional(set(string)) virtual_network_rules = optional(set(object({ subnet_id = string ignore_missing_vnet_service_endpoint = optional(bool, false) }))) }))	null	no
outbound_network_access_restricted	Whether outbound network access is restricted for the Cognitive Account. Defaults to false.	bool	false	no
pe_subresource	A list of subresource names which the Private Endpoint is able to connect to. subresource_names corresponds to group_id. Possible values are detailed in the product documentation in the Subresources column. Changing this forces a new resource to be created.	list(string)	[ "account" ]	no
private_dns_zone	A map of object that represents the existing Private DNS Zone you'd like to use. Leave this variable as default would create a new Private DNS Zone. type = object({ name = "(Required) The name of the Private DNS Zone." resource_group_name = "(Optional) The Name of the Resource Group where the Private DNS Zone exists. If the Name of the Resource Group is not provided, the first Private DNS Zone from the list of Private DNS Zones in your subscription that matches name will be returned." }	object({ name = string resource_group_name = optional(string) })	null	no
private_endpoint	A map of objects that represent the configuration for a private endpoint." type = map(object({ name = (Required) Specifies the Name of the Private Endpoint. Changing this forces a new resource to be created. vnet_rg_name = (Required) Specifies the name of the Resource Group where the Private Endpoint's Virtual Network Subnet exists. Changing this forces a new resource to be created. vnet_name = (Required) Specifies the name of the Virtual Network where the Private Endpoint's Subnet exists. Changing this forces a new resource to be created. subnet_name = (Required) Specifies the name of the Subnet which Private IP Addresses will be allocated for this Private Endpoint. Changing this forces a new resource to be created. dns_zone_virtual_network_link_name = (Optional) The name of the Private DNS Zone Virtual Network Link. Changing this forces a new resource to be created. Default to dns_zone_link. private_dns_entry_enabled = (Optional) Whether or not to create a private_dns_zone_group block for the Private Endpoint. Default to false. private_service_connection_name = (Optional) Specifies the Name of the Private Service Connection. Changing this forces a new resource to be created. Default to privateserviceconnection. is_manual_connection = (Optional) Does the Private Endpoint require Manual Approval from the remote resource owner? Changing this forces a new resource to be created. Default to false. }))	map(object({ name = string vnet_rg_name = string vnet_name = string subnet_name = string location = optional(string, null) dns_zone_virtual_network_link_name = optional(string, "dns_zone_link") private_dns_entry_enabled = optional(bool, false) private_service_connection_name = optional(string, "privateserviceconnection") is_manual_connection = optional(bool, false) }))	{}	no
public_network_access_enabled	Whether public network access is allowed for the Cognitive Account. Defaults to false.	bool	false	no
resource_group_name	Name of the azure resource group to use. The resource group must exist.	string	n/a	yes
sku_name	Specifies the SKU Name for this Cognitive Service Account. Possible values are F0, F1, S0, S, S1, S2, S3, S4, S5, S6, P0, P1, P2, E0 and DC0. Default to S0.	string	"S0"	no
tags	(Optional) A mapping of tags to assign to the resource.	map(string)	{}	no
tracing_tags_enabled	Whether enable tracing tags that generated by BridgeCrew Yor.	bool	false	no
tracing_tags_prefix	Default prefix for generated tracing tags	string	"avm_"	no

Outputs

Name	Description
openai_endpoint	The endpoint used to connect to the Cognitive Service Account.
openai_id	The ID of the Cognitive Service Account.
openai_primary_key	The primary access key for the Cognitive Service Account.
openai_secondary_key	The secondary access key for the Cognitive Service Account.
openai_subdomain	The subdomain used to connect to the Cognitive Service Account.
private_ip_addresses	A map dictionary of the private IP addresses for each private endpoint.

Contributing

Before submitting a pull request, please make sure the following is done:

We provide a docker image to run the pre-commit checks and tests for you: mcr.microsoft.com/azterraform:latest

To run the pre-commit task, we can run the following command:

docker run --rm -v $(pwd):/src -w /src mcr.microsoft.com/azterraform:latest make pre-commit

On Windows Powershell:

docker run --rm -v ${pwd}:/src -w /src mcr.microsoft.com/azterraform:latest make pre-commit

In pre-commit task, we will:

1. Run terraform fmt -recursive command for your Terraform code.
2. Run terrafmt fmt -f command for markdown files and go code files to ensure that the Terraform code embedded in these files are well formatted.
3. Run go mod tidy and go mod vendor for test folder to ensure that all the dependencies have been synced.
4. Run gofmt for all go code files.
5. Run gofumpt for all go code files.
6. Run terraform-docs on README.md file, then run markdown-table-formatter to format markdown tables in README.md.

Then we can run the pr-check task to check whether our code meets our pipeline's requirement (We strongly recommend you run the following command before you commit):

docker run --rm -v $(pwd):/src -w /src mcr.microsoft.com/azterraform:latest make pr-check

On Windows Powershell:

docker run --rm -v ${pwd}:/src -w /src mcr.microsoft.com/azterraform:latest make pr-check

To run the e2e-test, we can run the following command:

docker run --rm -v $(pwd):/src -w /src -e ARM_SUBSCRIPTION_ID -e ARM_TENANT_ID -e ARM_CLIENT_ID -e ARM_CLIENT_SECRET mcr.microsoft.com/azterraform:latest make e2e-test

On Windows Powershell:

docker run --rm -v ${pwd}:/src -w /src -e ARM_SUBSCRIPTION_ID -e ARM_TENANT_ID -e ARM_CLIENT_ID -e ARM_CLIENT_SECRET mcr.microsoft.com/azterraform:latest make e2e-test

Enable or disable tracing tags

We're using BridgeCrew Yor and yorbox to help manage tags consistently across infrastructure as code (IaC) frameworks. In this module you might see tags like:

resource "azurerm_resource_group" "rg" {
  location = "eastus"
  name     = random_pet.name
  tags = merge(var.tags, (/*<box>*/ (var.tracing_tags_enabled ? { for k, v in /*</box>*/ {
    avm_git_commit           = "3077cc6d0b70e29b6e106b3ab98cee6740c916f6"
    avm_git_file             = "main.tf"
    avm_git_last_modified_at = "2023-05-05 08:57:54"
    avm_git_org              = "lonegunmanb"
    avm_git_repo             = "terraform-yor-tag-test-module"
    avm_yor_trace            = "a0425718-c57d-401c-a7d5-f3d88b2551a4"
  } /*<box>*/ : replace(k, "avm_", var.tracing_tags_prefix) => v } : {}) /*</box>*/))
}

To enable tracing tags, set the variable to true:

module "example" {
  source               = "{module_source}"
  ...
  tracing_tags_enabled = true
}

The tracing_tags_enabled is default to false.

To customize the prefix for your tracing tags, set the tracing_tags_prefix variable value in your Terraform configuration:

module "example" {
  source              = "{module_source}"
  ...
  tracing_tags_prefix = "custom_prefix_"
}

The actual applied tags would be:

{
  custom_prefix_git_commit           = "3077cc6d0b70e29b6e106b3ab98cee6740c916f6"
  custom_prefix_git_file             = "main.tf"
  custom_prefix_git_last_modified_at = "2023-05-05 08:57:54"
  custom_prefix_git_org              = "lonegunmanb"
  custom_prefix_git_repo             = "terraform-yor-tag-test-module"
  custom_prefix_yor_trace            = "a0425718-c57d-401c-a7d5-f3d88b2551a4"
}

Telemetry Collection

This module uses terraform-provider-modtm to collect telemetry data. This provider is designed to assist with tracking the usage of Terraform modules. It creates a custom modtm_telemetry resource that gathers and sends telemetry data to a specified endpoint. The aim is to provide visibility into the lifecycle of your Terraform modules - whether they are being created, updated, or deleted. This data can be invaluable in understanding the usage patterns of your modules, identifying popular modules, and recognizing those that are no longer in use.

The ModTM provider is designed with respect for data privacy and control. The only data collected and transmitted are the tags you define in module's modtm_telemetry resource, an uuid which represents a module instance's identifier, and the operation the module's caller is executing (Create/Update/Delete/Read). No other data from your Terraform modules or your environment is collected or transmitted.

One of the primary design principles of the ModTM provider is its non-blocking nature. The provider is designed to work in a way that any network disconnectedness or errors during the telemetry data sending process will not cause a Terraform error or interrupt your Terraform operations. This makes the ModTM provider safe to use even in network-restricted or air-gaped environments.

If the telemetry data cannot be sent due to network issues, the failure will be logged, but it will not affect the Terraform operation in progress(it might delay your operations for no more than 5 seconds). This ensures that your Terraform operations always run smoothly and without interruptions, regardless of the network conditions.

You can turn off the telemetry collection by declaring the following provider block in your root module:

provider "modtm" {
  enabled = false
}