Creating an Azure function using CLI

Follow these steps to deploy a DataFlow Function using the Azure CLI.

  • Your NiFi flow definition is stored in the DataFlow Catalog and you have the CRN corresponding to the flow version you want to execute as a function

  • You have created a Machine User with the proper role and you have its Access Key and Private Key credentials

  • You have installed and configured the Azure CLI on your local machine

These steps provide an example for deploying a Function App that listens for HTTPS requests with a publicly exposed endpoint. Cloudera recommends going through the documented steps in the Azure Console UI to generate the appropriate templates and use those templates for CLI-based deployments.
  1. Create a Resource Group that you will use for the Function App. You can skip this step if you want to reuse an existing Resource Group.
    az group create -l francecentral -n MyDFFResourceGroup
    
  2. Download the template.json file available here.
    az deployment group create --resource-group MyDFFResourceGroup \
    	--template-file <path to template.json> \
    	--parameters \
    		subscriptionId=<existing subscription ID to use>  \
    		name=<name of your FunctionApp> \
    		location=<region where to deploy your Function App> \
    		hostingPlanName=<name of the plan to be created> \
    		resourceGroup=<name of the resource group previously created> \
    		storageAccountName=<name of the storage account to be created> \
    		FLOW_CRN=... \
    		DF_ACCESS_KEY=... \
    		DF_PRIVATE_KEY=...

    If you have additional parameters, you can add those in the appSettings section of the template.json file.

    Example:
    az deployment group create --resource-group MyDFFResourceGroup \
    	--template-file /tmp/template.json \
    	--parameters \
    	    subscriptionId=00000000-0000-0000-000000000000 \
    	    name=DFFResizeImage \
    	    location=francecentral \
    	    hostingPlanName=MyHostingPlanDFF \
    	    resourceGroup=MyDFFResourceGroup \
    	    storageAccountName=mydffstorageaccountname \
    	    FLOW_CRN=crn:cdp:df:us-west-1:00000000-0000-0000-0000-000000000000:flow:my-flow/v.1 \
    	    DF_ACCESS_KEY=00000000-0000-0000-0000-000000000000 \
    	    DF_PRIVATE_KEY=00000000000000000000000000000000000
    
  3. Retrieve the File Share name that has been created during the previous operation by executing the below command:
    az storage share list --account-name mydffstorageaccountname

    You can automate the retrieval with jq using the below command:

    FILESHARE=`az storage share list --account-name mydffstorageaccountname | jq -r '.[0].name'`
  4. Create the directories in the File Share:
    az storage directory create --account-name mydffstorageaccountname --share-name dffresizeimageadcb --name data
    az storage directory create --account-name mydffstorageaccountname --share-name dffresizeimageadcb --name data/SitePackages
    
  5. Upload the binary to the File Share and create the package file:
    az storage file upload -s dffresizeimageadcb --source /tmp/NaafAzureFunctions.zip --path data/SitePackages/ --account-name mydffstorageaccountname
    
    echo "NaafAzureFunctions.zip" > /tmp/packagename.txt && az storage file upload -s dffresizeimageadcb --source /tmp/packagename.txt --path data/SitePackages/ --account-name mydffstorageaccountname && rm /tmp/packagename.txt
    

Your function is now ready to be used with HTTPS triggers. See the documentation for more details. If another type of trigger is required, you need to create an Event Grid subscription as explained in Azure Function triggers.