Category: Azure

I ran into some issues when actually attempting to apply my first Terraform template, specifically errors related to the location I had chosen:

* azurerm_virtual_machine.test: compute.VirtualMachinesClient#CreateOrUpdate:
Failure sending request: StatusCode=409 -- 
Original Error: failed request: autorest/azure: Service returned an error. 
Status=<nil> Code="SkuNotAvailable" 
Message="The requested size for resource '/subscriptions/f745d13d/resourceGroups/HelloWorld/providers/Microsoft.Compute/virtualMachines/helloworld' is currently not available in location 'WestUS' zones '' for subscription 'f745d13d'. 
Please try another size or deploy to a different location or zones. 
See https://aka.ms/azureskunotavailable for details."

This didn’t make much sense to me, because I was using a very normal VM size (like Standard_A2 or something) and it was definitely available in WestUS.

The key to solving was when I use the Azure Shell (powershell) to run this:

Get-AzureRmComputeResourceSku | where {$_.Locations.Contains("WestUS")}

The output of this command:

Turns out for my Subscription (Visual Studio Enterprise – MPN), WestUS is restricted to very few VM sizes (note the “NotAvailableForSubscription” items). When I target WestUS2 or EastUS, then there’s quite a bit more choice.

Having access to an Azure credit through my workplace, I’ve begun training and testing various things within Azure, and recently I was looking at ARM templates for VM deployment. This research led me to stumble across Terraform as an infrastructure-as-code deployment tool.

I started messing around with Terraform to deploy a basic set of Azure resources using these two guides:

• Getting started with Terraform on Windows and Azure (Peter Groenewegen)
• https://docs.microsoft.com/en-us/azure/virtual-machines/linux/terraform-create-complete-vm

Using Peter’s example was great, although I had to make a few tweaks since the module syntax has changed. In the near future I’ll post a full example of what I used recently for my first template. Terraform documentation has been excellent as I’ve worked through syntax and various examples, which is a great surprise.

I’ve found that I appreciate building a template in Terraform much more than an ARM template. Part of this is the easier syntax and readability, and the logical manner that I’ve seen others use Terraform .tf files for segregating resource deployment.

I think that the other primary factor is because declarative Terraform is what I’ve seen referred to as “idempotent” – a new word for me I’ve only ever seen used in this context. The answer on this StackOverflow question gives a great description of this context, and its clearly descriptive of how Terraform operates.

If I build an ARM template and apply it, it will deploy the resource as I’ve described it. And then if I apply that template again, it will attempt to build another of that resource. And if I modify the template and apply it again, it will attempt to create that slightly modified resource, leaving me with 3 different resources.

Terraform instead gives the ability to declare “what I want it to look like”, and ensures that the actual state matches the declared state. There are genuinely good reasons to use both approaches but right now this idempotent-style of deployment is new and attractive to me.

Soon on the horizon in my sandbox is to utilize remote state in an Azure blob storage to facilitate team work within one environment.