Tests fail in GoogleCloudPlatform / magic-modules with sdk release 2.24.0

[zli82016]

SDK version

2.24.0

Relevant provider source code

https://github.com/GoogleCloudPlatform/magic-modules

Terraform Configuration Files

...

Debug Output

Expected Behavior

The following tests should pass.

TestAccContainerNodePool_basicWithClusterId|TestAccContainerNodePool_autoscaling|TestAccContainerNodePool_totalSize|TestAccContainerNodePool_regionalAutoscaling|TestAccContainerNodePool_basic|TestAccContainerCluster_withResourceUsageExportConfig|TestAccComputeVpnTunnel_vpnTunnelBetaExample|TestAccComputeAutoscaler_autoscalerSingleInstanceExample|TestAccProviderBasePath_setInvalidBasePath|TestAccProviderBasePath_setBasePath|TestAccSecurityScannerScanConfig_scanConfigBasicExample

Actual Behavior

The above tests fail.

Steps to Reproduce

References

• Upgrade terraform-plugin-sdk to 2.24.0 GoogleCloudPlatform/magic-modules#6818

[zli82016]

It looks like the commit in sdk (ef65fde#diff-3894fabafd3f4b1861dd167c6194c6d79e5824d2bec30d37b257d80a824255c2R19 ) causes the tests fail in GoogleCloudPlatform / magic-modules.

[bflad]

Hi @zli82016 👋 Thank you for raising this and sorry your provider is running into testing issues.

Two questions for you:

• Can you please post the failed test output and relevant test code? The maintainers here cannot access the GCP acceptance testing links for the failed tests nor run the tests ourselves to triage this issue.
• Does upgrading to 2.24.1 fix the issue(s)? If it relates to test configuration handling, this was a bug fix from that release:

helper/resource: Prevented provider configuration already given error when TestStep type Config field already contained provider configuration block (#1092)

Thank you so much.

[zli82016]

Test code: https://gist.github.com/zli82016/80a171c9403f6a8a317d9a918e7c31fe
Test output: https://gist.github.com/zli82016/54bd8d0adbbb1790995357fe7cb00d46

[zli82016]

I checked the fix in 2.24.1. I do find the place to set the flag skipProviderBlock for our test cases.

[zli82016]

V2.24.1 does not fix the tests.

[zli82016]

Thanks @bflad for the fix in V2.24.1. It is weird that it does not fix our tests. If you need more information, feel free to let me know. Thanks.

[bflad]

If you have trace logging enabled, it should output the full Terraform configuration being applied with a message Setting Terraform configuration.

Looking at the test code itself, it is using the deprecated TestCase type Providers field (Providers: testAccProvidersOiCS,). There is some legacy logic in the testing code that implicitly adds provider configuration blocks, but only when using that field. We can try to determine if that is safe to skip in this case. Are you able to try switching the TestCase to using the ProviderFactories or ProtoV5ProviderFactories field instead to see if that resolves this issue for you?

[zli82016]

Thanks, @bflad. I can see the Terraform configuration with enabled trace logging. It has two provider configuration block.

provider "google-beta" {}

provider "google-beta" {
region = "us-central1"
zone = "us-central1-a"
}

I switched to ProviderFactories and the test runs successfully.

It needs some work to switch to ProviderFactories for our system and I am not sure if we have the plan to do that.

Can you try to check if it is safe to skip the provider configuration block in this case? Thanks.

[zli82016]

Hello, @bflad , sorry to check again. Do you think is it safe to skip the provider configuration block in this case? Thanks.

[bflad]

It looks like this was resolved last year, but if not, please feel free to submit a bug report with a full reproduction case.

[github-actions]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.
If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.