working version of a managed app

infra/README.md

@@ -1,17 +1,41 @@
 # Managed App Instructions
 
-This guide is a temporary document that walks through the progress made so far to convert the graphrag solution accelerator to a managed app.
+This guide is a temporary document that walks through the process to convert the graphrag solution accelerator to a managed app.
 
-1. Auto format the bicep code
+### 1. Auto format the bicep code
+
+As a precaution, start by auto-formating and linting the bicep code to detect any mistakes early-on.
 
 ```bash
+cd <repo_root_directory>/infra
 find . -type f -name "*.bicep" -exec az bicep format --file {} \;
+find . -type f -name "*.bicep" -exec az bicep lint --file {} \;
 ```
 
-2. Convert bicep -> ARM
+### 2. Convert bicep -> ARM
 ```bash
 az bicep build --file main.bicep --outfile managed-app/mainTemplate.json
 ```
 
-3. Test the Portal Interface
+### 3. Create & test the Azure portal interface
-Use the [Azure Portal Sandbox](https://portal.azure.com/#blade/Microsoft_Azure_CreateUIDef/SandboxBlade) to test and make any UI changes in `managed-app/createUiDefinition.json`. To make additional changes to the Azure portal experience, start by reading some [documentation](https://learn.microsoft.com/en-us/azure/azure-resource-manager/managed-applications/create-uidefinition-overview) and copying the contents of `createUiDefinition.json` into the sandbox environment.
+
+Use the [Azure Portal Sandbox](https://portal.azure.com/#blade/Microsoft_Azure_CreateUIDef/SandboxBlade) to test and make any UI changes that are defined in [createUiDefinition.json](createUiDefinition.json). To make additional changes to the Azure portal experience, start by reading some [documentation](https://learn.microsoft.com/en-us/azure/azure-resource-manager/managed-applications/create-uidefinition-overview) and copying the contents of `createUiDefinition.json` into the sandbox environment.
+
+### 4. Package up the managed app code
+
+The name of the final two files (`mainTemplate.json` and `createUiDefinition.json`) cannot be changed. The file names are also case-sensitive and cannot be changed at this time. Managed apps require these files to be packaged up into a zip file (where the json files must be at the root directory).
+
+```bash
+cd <repo_root_directory>/infra/managed-app
+zip -rj managed-app.zip .
+```
+
+This zip file can then be uploaded to an Azure Storage location when setting up a [Service Catalog Managed Application Definition](https://ms.portal.azure.com/#view/Microsoft_Azure_Marketplace/GalleryItemDetailsBladeNopdl/id/Microsoft.ApplianceDefinition/selectionMode~/false/resourceGroupId//resourceGroupLocation//dontDiscardJourney~/false/selectedMenuId/home/launchingContext~/%7B%22galleryItemId%22%3A%22Microsoft.ApplianceDefinition%22%2C%22source%22%3A%5B%22GalleryFeaturedMenuItemPart%22%2C%22VirtualizedTileDetails%22%5D%2C%22menuItemId%22%3A%22home%22%2C%22subMenuItemId%22%3A%22Search%20results%22%2C%22telemetryId%22%3A%2220409084-39a1-4800-bbce-d0b26a6f46a4%22%7D/searchTelemetryId/d7d20e05-ca16-47f7-bed5-9c7b8d2fa641).
+
+### 5. Create the Service Catalog Managed App Definition
+
+In the Azure Portal, go to Marketplace and create a `Service Catalog Managed App Definition`. You must provide a uri link to the uploaded `managed-app.zip` file as part of the creation process.
+
+### 6. Deploy the managed app
+
+In the Azure Portal, find and click on the managed app definition resource that was created in the previous step. A button option to `Deploy from definition` will be available. Click on it and proceed through the setup steps (defined by the `createUiDefinitions.json` file) that a consumer would experience when installing the managed app.

infra/core/aoai/aoai.bicep

@@ -4,17 +4,23 @@ param openAiName string = 'openai${uniqueString(resourceGroup().id)}'
 @description('Location for the Azure OpenAI instance')
 param location string = resourceGroup().location
 
-@description('LLM model deployment name')
+@description('LLM model name')
-param llmModelDeploymentName string = 'gpt-4o'
+param llmModelName string = 'gpt-4o'
 
-@description('Embedding model deployment name')
+@description('LLM Model API version')
-param embeddingModelDeploymentName string = 'text-embedding-ada-002'
+param llmModelVersion string
 
-@description('TPM quota for GPT-4o deployment')
+@description('Embedding model name')
-param gpt4oTpm int = 10
+param embeddingModelName string = 'text-embedding-ada-002'
 
-@description('TPM quota for text-embedding-ada-002 deployment')
+@description('Embedding Model API version')
-param textEmbeddingAdaTpm int = 10
+param embeddingModelVersion string
+
+@description('TPM quota for llm model deployment (x1000)')
+param llmTpmQuota int = 10
+
+@description('TPM quota for embedding model deployment (x1000)')
+param embeddingTpmQuota int = 10
 
 @description('Array of objects with fields principalId, roleDefinitionId')
 param roleAssignments array = []
@@ -32,39 +38,39 @@ resource aoai 'Microsoft.CognitiveServices/accounts@2024-10-01' = {
   }
 }
 
-resource gpt4oDeployment 'Microsoft.CognitiveServices/accounts/deployments@2024-10-01' = {
+resource llmDeployment 'Microsoft.CognitiveServices/accounts/deployments@2024-10-01' = {
   parent: aoai
-  name: llmModelDeploymentName
+  name: llmModelName
   sku: {
     name: 'GlobalStandard'
-    capacity: gpt4oTpm
+    capacity: llmTpmQuota
   }
   properties: {
     model: {
       format: 'OpenAI'
-      name: 'gpt-4o'
+      name: llmModelName
-      version: '2024-05-13'
+      version: llmModelVersion
     }
-    currentCapacity: gpt4oTpm
+    currentCapacity: llmTpmQuota
   }
 }
 
-resource textEmbeddingAdaDeployment 'Microsoft.CognitiveServices/accounts/deployments@2024-10-01' = {
+resource embeddingDeployment 'Microsoft.CognitiveServices/accounts/deployments@2024-10-01' = {
   parent: aoai
-  name: embeddingModelDeploymentName
+  name: embeddingModelName
   // NOTE: simultaneous model deployments are not supported at this time. As a workaround, use dependsOn to force the models to be deployed in a sequential manner.
-  dependsOn: [gpt4oDeployment]
+  dependsOn: [llmDeployment]
   sku: {
     name: 'Standard'
-    capacity: textEmbeddingAdaTpm
+    capacity: embeddingTpmQuota
   }
   properties: {
     model: {
       format: 'OpenAI'
-      name: 'text-embedding-ada-002'
+      name: embeddingModelName
-      version: '2'
+      version: embeddingModelVersion
     }
-    currentCapacity: textEmbeddingAdaTpm
+    currentCapacity: embeddingTpmQuota
   }
 }
 
@@ -77,9 +83,9 @@ resource roleAssignment 'Microsoft.Authorization/roleAssignments@2022-04-01' = [
 ]
 
 output openAiEndpoint string = aoai.properties.endpoint
-output llmModel string = gpt4oDeployment.properties.model.name
+output llmModel string = llmDeployment.properties.model.name
-output llmModelDeploymentName string = gpt4oDeployment.name
+output llmModelDeploymentName string = llmDeployment.name
-output llmModelApiVersion string = gpt4oDeployment.apiVersion
+output llmModelApiVersion string = llmDeployment.apiVersion
-output textEmbeddingModel string = textEmbeddingAdaDeployment.properties.model.name
+output textEmbeddingModel string = embeddingDeployment.properties.model.name
-output textEmbeddingModelDeploymentName string = textEmbeddingAdaDeployment.name
+output textEmbeddingModelDeploymentName string = embeddingDeployment.name
-output textEmbeddingModelApiVersion string = textEmbeddingAdaDeployment.apiVersion
+output textEmbeddingModelApiVersion string = embeddingDeployment.apiVersion

infra/deploy.sh

@@ -16,7 +16,6 @@ GRAPHRAG_IMAGE=""
 PUBLISHER_EMAIL=""
 PUBLISHER_NAME=""
 RESOURCE_BASE_NAME=""
-CONTAINER_REGISTRY_NAME=""
 
 requiredParams=(
     LOCATION
@@ -316,7 +315,6 @@ deployAzureResources () {
         --parameters "apiPublisherName=$PUBLISHER_NAME" \
         --parameters "apiPublisherEmail=$PUBLISHER_EMAIL" \
         --parameters "enablePrivateEndpoints=$ENABLE_PRIVATE_ENDPOINTS" \
-        --parameters "acrName=$CONTAINER_REGISTRY_NAME" \
         --output json)
     # errors in deployment may not be caught by exitIfCommandFailed function so we also check the output for errors
     exitIfCommandFailed $? "Error deploying Azure resources..."

infra/main.bicep

@@ -56,6 +56,24 @@ param storageAccountName string = ''
 param cosmosDbName string = ''
 param aiSearchName string = ''
 
+// AOAI parameters
+@description('Name of the AOAI LLM model to use. Must match official model id. For more information: https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models')
+@allowed(['gpt-4o', 'gpt-4o-mini'])
+param llmModelName string = 'gpt-4o'
+@description('Version of the AOAI LLM model to use.')
+param llmModelVersion string = '2024-08-06'
+@description('Quota of the AOAI LLM model to use.')
+@minValue(1)
+param llmModelQuota int = 10
+
+@description('Name of the AOAI embedding model to use. Must match official model id. For more information: https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models')
+@allowed(['text-embedding-ada-002', 'text-embedding-3-large'])
+param embeddingModelName string = 'text-embedding-ada-002'
+param embeddingModelVersion string = '2'
+@description('Quota of the AOAI embedding model to use.')
+@minValue(1)
+param embeddingModelQuota int = 10
+
 var abbrs = loadJsonContent('abbreviations.json')
 var tags = { 'azd-env-name': resourceGroup }
 var workloadIdentityName = '${abbrs.managedIdentityUserAssignedIdentities}${resourceBaseNameFinal}'
@@ -175,10 +193,12 @@ module aoai 'core/aoai/aoai.bicep' = {
   params: {
     openAiName: '${abbrs.cognitiveServicesAccounts}${resourceBaseNameFinal}'
     location: location
-    llmModelDeploymentName: 'gpt-4o-${uniqueString(resourceBaseNameFinal)}'
+    llmModelName: llmModelName
-    gpt4oTpm: 10
+    llmModelVersion: llmModelVersion
-    embeddingModelDeploymentName: 'text-embedding-ada-002-${uniqueString(resourceBaseNameFinal)}'
+    llmTpmQuota: llmModelQuota
-    textEmbeddingAdaTpm: 10
+    embeddingModelName: embeddingModelName
+    embeddingModelVersion: embeddingModelVersion
+    embeddingTpmQuota: embeddingModelQuota
     roleAssignments: [
       {
         principalId: workloadIdentity.outputs.principalId
@@ -419,6 +439,11 @@ output azure_ai_search_name string = aiSearch.outputs.name
 output azure_acr_login_server string = acr.outputs.loginServer
 output azure_acr_name string = acr.outputs.name
 
+output azure_aks_name string = aks.outputs.name
+output azure_aks_controlplanefqdn string = aks.outputs.controlPlaneFqdn
+output azure_aks_managed_rg string = aks.outputs.managedResourceGroup
+output azure_aks_service_account_name string = aksServiceAccountName
+
 output azure_aoai_endpoint string = aoai.outputs.openAiEndpoint
 output azure_aoai_llm_model string = aoai.outputs.llmModel
 output azure_aoai_llm_model_deployment_name string = aoai.outputs.llmModelDeploymentName
@@ -427,32 +452,27 @@ output azure_aoai_embedding_model string = aoai.outputs.textEmbeddingModel
 output azure_aoai_embedding_model_deployment_name string = aoai.outputs.textEmbeddingModelDeploymentName
 output azure_aoai_embedding_model_api_version string = aoai.outputs.textEmbeddingModelApiVersion
 
-output azure_aks_name string = aks.outputs.name
+output azure_apim_name string = apim.outputs.name
-output azure_aks_controlplanefqdn string = aks.outputs.controlPlaneFqdn
+output azure_apim_gateway_url string = apim.outputs.apimGatewayUrl
-output azure_aks_managed_rg string = aks.outputs.managedResourceGroup
-output azure_aks_service_account_name string = aksServiceAccountName
 
-output azure_workload_identity_client_id string = workloadIdentity.outputs.clientId
+output azure_app_hostname string = appHostname
-output azure_workload_identity_principal_id string = workloadIdentity.outputs.principalId
+output azure_app_url string = appUrl
-output azure_workload_identity_name string = workloadIdentity.outputs.name
 
-output azure_storage_account string = storage.outputs.name
+output azure_app_insights_connection_string string = apim.outputs.appInsightsConnectionString
-output azure_storage_account_blob_url string = storage.outputs.primaryEndpoints.blob
 
 output azure_cosmosdb_endpoint string = cosmosdb.outputs.endpoint
 output azure_cosmosdb_name string = cosmosdb.outputs.name
 output azure_cosmosdb_id string = cosmosdb.outputs.id
 
-output azure_app_insights_connection_string string = apim.outputs.appInsightsConnectionString
-
-output azure_apim_name string = apim.outputs.name
-output azure_apim_gateway_url string = apim.outputs.apimGatewayUrl
-
 output azure_dns_zone_name string = privateDnsZone.outputs.name
 
-output azure_app_hostname string = appHostname
-output azure_app_url string = appUrl
-
 output azure_private_dns_zones array = enablePrivateEndpoints
   ? union(privatelinkPrivateDns.outputs.privateDnsZones, [privateDnsZone.outputs.name])
   : []
+
+output azure_storage_account string = storage.outputs.name
+output azure_storage_account_blob_url string = storage.outputs.primaryEndpoints.blob
+
+output azure_workload_identity_client_id string = workloadIdentity.outputs.clientId
+output azure_workload_identity_principal_id string = workloadIdentity.outputs.principalId
+output azure_workload_identity_name string = workloadIdentity.outputs.name

infra/managed-app/createUiDefinition.json

@@ -0,0 +1,176 @@
+{
+  "$schema": "https://schema.management.azure.com/schemas/0.1.2-preview/CreateUIDefinition.MultiVm.json#",
+  "handler": "Microsoft.Azure.CreateUIDef",
+  "version": "0.1.2-preview",
+  "parameters": {
+    "basics": [
+      {}
+    ],
+    "steps": [
+      {
+        "name": "aoaiSettings",
+        "label": "AOAI Settings",
+        "subLabel": {
+          "preValidation": "Configure the AOAI settings",
+          "postValidation": "Completed"
+        },
+        "elements": [
+          {
+            "name": "llmModel",
+            "type": "Microsoft.Common.DropDown",
+            "label": "LLM Model",
+            "defaultValue": "gpt-4o",
+            "toolTip": "LLM model to use.",
+            "constraints": {
+              "allowedValues": [
+                {
+                  "label": "gpt-4o",
+                  "value": "gpt-4o"
+                },
+                {
+                  "label": "gpt-4o-mini",
+                  "value": "gpt-4o-mini"
+                }
+              ],
+              "required": true
+            },
+            "visible": true
+          },
+          {
+            "name": "llmModelVersion",
+            "type": "Microsoft.Common.DropDown",
+            "label": "LLM Model Version",
+            "defaultValue": "2024-08-06",
+            "toolTip": "LLM model version to use.",
+            "constraints": {
+              "allowedValues": [
+                {
+                  "label": "2024-08-06",
+                  "value": "2024-08-06"
+                },
+                {
+                  "label": "2024-07-18",
+                  "value": "2024-07-18"
+                }
+              ],
+              "required": true
+            },
+            "visible": true
+          },
+          {
+            "name": "llmModelQuota",
+            "type": "Microsoft.Common.TextBox",
+            "label": "LLM Model Quota (x1000)",
+            "placeholder": "85",
+            "defaultValue": "",
+            "toolTip": "Model quota to use.",
+            "constraints": {
+              "required": true,
+              "regex": "^[1-9][0-9]*$",
+              "validationMessage": "Valid LLM model quota."
+            },
+            "visible": true
+          },
+          {
+            "name": "embeddingModel",
+            "type": "Microsoft.Common.DropDown",
+            "label": "Embedding Model",
+            "defaultValue": "text-embedding-ada-002",
+            "toolTip": "Embedding model to use",
+            "constraints": {
+              "allowedValues": [
+                {
+                  "label": "text-embedding-ada-002",
+                  "value": "text-embedding-ada-002"
+                },
+                {
+                  "label": "text-embedding-3-large",
+                  "value": "text-embedding-3-large"
+                }
+              ],
+              "required": true
+            },
+            "visible": true
+          },
+          {
+            "name": "embeddingModelQuota",
+            "type": "Microsoft.Common.TextBox",
+            "label": "Embedding Model Quota (x1000)",
+            "placeholder": "100",
+            "defaultValue": "",
+            "toolTip": "Model quota to use.",
+            "constraints": {
+              "required": true,
+              "regex": "^[1-9][0-9]*$",
+              "validationMessage": "Valid embedding model quota."
+            },
+            "visible": true
+          },
+          {
+            "name": "embeddingModelVersion",
+            "type": "Microsoft.Common.DropDown",
+            "label": "Embedding Model Version",
+            "defaultValue": "2",
+            "toolTip": "Use a valid embedding model version.",
+            "constraints": {
+              "allowedValues": [
+                {
+                  "label": "2",
+                  "value": "2"
+                },
+                {
+                  "label": "1",
+                  "value": "1"
+                }
+              ],
+              "required": true
+            },
+            "visible": true
+          }
+        ]
+      },
+      {
+        "name": "graphragSettings",
+        "label": "GraphRAG Settings",
+        "subLabel": {
+          "preValidation": "Configure the graphrag settings",
+          "postValidation": "Completed"
+        },
+        "elements": [
+          {
+            "name": "apimTier",
+            "type": "Microsoft.Common.DropDown",
+            "label": "APIM Tier",
+            "defaultValue": "StandardV2",
+            "toolTip": "APIM tier to use",
+            "constraints": {
+              "allowedValues": [
+                {
+                  "label": "Developer",
+                  "value": "Developer"
+                },
+                {
+                  "label": "StandardV2",
+                  "value": "StandardV2"
+                }
+              ],
+              "required": true
+            },
+            "visible": true
+          }
+        ]
+      }
+    ],
+    "outputs": {
+      "resourceGroup": "[resourceGroup().name]",
+      "location": "[location()]",
+      "apimTier": "[steps('graphragSettings').apimTier]",
+      "llmModelName": "[steps('aoaiSettings').llmModel]",
+      "llmModelQuota": "[int(steps('aoaiSettings').llmModelQuota)]",
+      "embeddingModelName": "[steps('aoaiSettings').embeddingModel]",
+      "embeddingModelQuota": "[int(steps('aoaiSettings').embeddingModelQuota)]",
+      "llmModelVersion": "[steps('aoaiSettings').llmModelVersion]",
+      "embeddingModelVersion": "[steps('aoaiSettings').embeddingModelVersion]"
+    }
+  }
+}