-
Notifications
You must be signed in to change notification settings - Fork 11
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
helper/resource: Refresh during plan rather than separately and remove state shimming when possible #223
Conversation
…e state shimming when possible Reference: hashicorp/terraform-plugin-sdk#515 Reference: #194 Reference: #222 Reference: https://developer.hashicorp.com/terraform/cli/commands/refresh This change replaces the usage of explicit refresh commands by instead calling the next command with the relevant refresh flags. Separating the refresh from plan was previously done with the idea that the testing logic may introduce further assertion hooks based on that explicit refresh. Over time though, the Terraform recommendations for refreshing and planning have solidified around always recommending to use a plan with behavior changing flags as necessary. Even the Terraform CLI implementation of the refresh command itself has been shifted to call a plan with special flags. This should reduce testing times decently as it means that the testing logic will skip spinning up and tearing down any relevant providers two times per `TestStep` and whatever time would be associated with Terraform CLI loading for those commands. Comparing the testing performed across this Go module: Prior: `TF_ACC=1 go test -count=1 ./... 111.07s user 59.37s system 233% cpu 1:12.99 total` Now: `TF_ACC=1 go test -count=1 ./... 90.81s user 48.25s system 229% cpu 1:00.54 total` The separated refresh also prevented `PreApply` plan checks from catching certain situations because the refreshed state could cause the plan to be missing changes.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM 🚀
If they have time, I'm soliciting a |
If any bugs are caused by this change, please open a bug report and we will take a look. |
Reference: #223 Reference: #237 Reference: #256 Post-release update to the v1.6.0 CHANGELOG to capture now-known situations that would affect existing provider testing. This update ensures developers upgrading to v1.6.0 will receive the latest guidance and the guidance will remain for this version after future releases. If approved and merged, the v1.6.0 GitHub Release description will also be updated with the same additional content.
Reference: #223 Reference: #237 Reference: #256 Post-release update to the v1.6.0 CHANGELOG to capture now-known situations that would affect existing provider testing. This update ensures developers upgrading to v1.6.0 will receive the latest guidance and the guidance will remain for this version after future releases. If approved and merged, the v1.6.0 GitHub Release description will also be updated with the same additional content.
I'm going to lock this pull request because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active contributions. |
Reference: hashicorp/terraform-plugin-sdk#515
Reference: #194
Reference: #222
Reference: https://developer.hashicorp.com/terraform/cli/commands/refresh
This change replaces the usage of explicit refresh commands by instead calling the next command with the relevant refresh flags. Separating the refresh from plan was previously done with the idea that the testing logic may introduce further assertion hooks based on that explicit refresh. Over time though, the Terraform recommendations for refreshing and planning have solidified around always recommending to use a plan with behavior changing flags as necessary. Even the Terraform CLI implementation of the refresh command itself has been shifted to call a plan with special flags.
This should reduce testing times decently as it means that the testing logic will skip spinning up and tearing down any relevant providers two times per
TestStep
and whatever time would be associated with Terraform CLI loading for those commands. Comparing the testing performed across this Go module:Prior:
TF_ACC=1 go test -count=1 ./... 111.07s user 59.37s system 233% cpu 1:12.99 total
Now:
TF_ACC=1 go test -count=1 ./... 90.81s user 48.25s system 229% cpu 1:00.54 total
The separated refresh also prevented
PreApply
plan checks from catching certain situations because the refreshed state could cause the plan to be missing changes.