Recently I was developing a simple Terraform pipeline in Azure DevOps. Everything seemed intuitive and easy - install, init, plan, apply. Clear steps and UI forms for every setting.
Everything was configured properly. Code was working fine from my local machine, but in Azure DevOps, pipeline run successfully only for the first time. Every next attempt resulted in the following error:
Error: The process 'C:\hostedtoolcache\windows\terraform\1.0.8\x64\terraform.exe' failed with exit code 1
And details from the log:
Error: [0m[0m[1mA resource with the ID "/subscriptions/45197771-2d3e-4685-aaef-e96f62dd80b4/resourceGroups/tfexample-sample-rg" already exists - to be managed via Terraform this resource needs to be imported into the State. Please see the resource documentation for "azurerm_resource_group" for more information.[0m
The problem was in my code
provider "azurerm" {
subscription_id = "45197771-2d3e-4685-aaef-e96f62dd80b4"
features {}
}
It was missing the backend part! Code worked from my local machine, because terraform saved the state in local terraform.tfstate file (default terraform behavior), but in Azure DevOps the settings provided in the UI was ignored with no clear error message. terraform apply couldn't find the state file, so was creating every resource from scratch each time.
The solution
I fixed the code by adding empty properties. That was enough to have the pipeline working correctly. Note that no values are required...
provider "azurerm" {
subscription_id = "45197771-2d3e-4685-aaef-e96f62dd80b4"
features {}
}
terraform {
backend "azurerm" {
resource_group_name = ""
storage_account_name = ""
container_name = ""
key = ""
}
}
Please leave some comment if this helped.
Top comments (5)
You saved me hours of struggle. Thank you so much. I implemented your suggestion and it started working :)
hello, is this the solution ?
because , i have a similare one
when i create a resource using a pipeline azure and when i want to update the code with adding some resources, it give me the some error
hello, sorry for late response
generally - yes. With this error, it's an issue with state - probably it's not stored anywhere
If we leave backend settings blank, then where would our state file have saved?
in Azure Dev Ops