Support for detection and reduction of nondeterminism in terraform behaviour

[marcindulak]

Is there an existing issue for this?

• I have searched the existing issues

Community Note

• Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
• Please do not leave "+1" or "me too" comments, they generate extra noise for issue followers and do not help prioritize the request
• If you are interested in working on this issue or have submitted a pull request, please leave a comment and review the contribution guide to help.

Description

For some time, order or months, I have a subjective impression about an increase of nondeterministic behaviour of terraform.
By nondeterministic I mean that problems happen only during some "terraform apply" without the code or provider version change.

Below are some examples of nondeterministic problems. Other users may provide more of them.

#20441
#23713
#15048
#26183
#24804

The impact of these problems is that they create costs for the operation teams: the "terraform apply" fails, and a human needs to be involved to identify the reason, wait or intervene manually in Azure portal or by other means, and rerun "terraform apply". Since there are no solutions to these nondeterministic problems, they are coming back, and the value of terraform as an infrastructure as code tool is reduced.

This is meta issue, similar to #25394, created for the purpose of giving a heads up to the team about something going wrong with the quality of the provider, Terraform Cloud backend, or terraform itself.

A feedback from the core contributors, about what are the plans to detect such flaky features, and how to improve this situation is expected.

New or Affected Resource(s)/Data Source(s)

azurerm

Potential Terraform Configuration

No response

References

No response