The Azure SDN connector uses two types of queries to interact with the Azure management API.
The first query is targeted to a special IP address to get a token. This token is used to authenticate the subsequent queries. The second type of query is used to retrieve information about the Azure resources, such as virtual machines, network interfaces, network security groups, and public IP addresses. Some queries are made to manage public IP addresses, such as assigning or releasing them from the FortiGate VM.

To get Terraform ready using Microsoft Azure Cloud Shell, you need to perform the following steps:
* Set up a storage account in Azure. This is required to store the Terraform state file in a blob container, which enables collaboration and persistence of the infrastructure configuration1.
* Use the wget (terraform_version) command to upload Terraform. This command downloads the latest version of Terraform from the official website and saves it as a zip file in the current directory2.
* Move the Terraform file to the bin directory. This step extracts the Terraform executable from the zip file and moves it to the bin directory, which is part of the PATH environment variable. This allows you to run Terraform commands from any directory in Cloud Shell2.
The other options are incorrect because:
* You do not need to use the -O command to download Terraform. This command is used to specify a different output file name for the downloaded file, but it is not necessary for this task3.
* You do not need to subscribe to Terraform in Azure. Terraform is an open-source tool that can be used with any cloud provider, and there is no subscription or registration required to use it with Azure4. References:
* Updating the route table and adding an IAM policy
* Configure Terraform in Azure Cloud Shell with Bash
* wget(1) - Linux man page
* Terraform by HashiCorp

A is correct because in this deployment, the passive FortiGate issues API calls to Azure to update the routing table and the public IP address of the active FortiGate123. This way, the traffic is redirected to the new active FortiGate after a failover.
B is incorrect because the vdom-exception command is used to exclude specific VDOMs from being synchronized in an HA cluster.This command is not related to this deployment scenario.
C is incorrect because Microsoft Azure does provide an SLA for API calls. According to the Azure Service Level Agreements, the API Management service has a monthly uptime percentage of at least
99.9% for the standard tier and higher.
D is correct because by default, the configuration is not synchronized between the primary and secondary devices in this deployment. The administrator needs to manually enable configuration synchronization on both devices123. Alternatively, the administrator can use FortiManager to manage and synchronize the configuration of both devices4.

In this scenario, the issue is caused by the Azure service principle account not having a contributor role. This is required for the FortiGate HA floating IP to work properly. Without this role, the new primary device will not have the previous primary device floating IP address after failover.