Completing the examples from Terraform Up & Running in aws would be too much of a copy operation. I decided to use Azure to create a more interesting learning experience. Each branch represents the order of the chapter work in the book, with main being the working tree at any given time.
This configuration is a tab bit more complex in Azure, but is more or less as described in the book.
The Azure Quickstart load balancer documents were a great way to validate the Terraform resources used, but they are always missing enough that hunting down further details in needed. For instance: Using scale_sets versus VMs requires the use of depends_on to make sure creation happens in the right order. As the book points out sometimes you need to provide "extra hints" to the instantiation process.
- az vmss get-instance-view --resource-group TFResourceGroupG --name example-machine
- az vmss extension show --resource-group TFResourceGroupG --vmss-name example-machine --name busyboxhttpd
- az vmss list-instances --resource-group TFResourceGroupG --name example-machine --query "[].{instanceId:instanceId, extension:resources[].id, extProvisioningState:resources[].provisioningState}"
The chapter material using an AWS S3 bucket for backend storage is simple, and the directions using the Azure CLI credentials (that you are probably already using) is just about the same. Issue #6 will create a better solution.
Per the book, all the backend details except individual keys were put into one .hcl file and called with in each init: terraform init -backend-config=../backend.hcl
This chapter raises an interesting point about the use of Azure ResourceGroups. When you manage parts of your infrastructure separately with remote backends and data sources, should the RGs be separate or can they be the same name?