hashicorp / terraform-provider-google

Terraform Provider for Google Cloud Platform
https://registry.terraform.io/providers/hashicorp/google/latest/docs
Mozilla Public License 2.0
2.36k stars 1.75k forks source link

Removing `schedule_info` from `google_data_pipeline_pipeline` does not delete Cloud Scheduler job #17489

Open kpocius opened 8 months ago

kpocius commented 8 months ago

Community Note

Terraform Version

1.7.4

Affected Resource(s)

google_data_pipeline_pipeline

Terraform Configuration

No response

Debug Output

No response

Expected Behavior

Removing schedule_info should remove the associated Cloud Scheduler job

Actual Behavior

Cloud Scheduler job remains in place and the pipeline keeps getting executed on the schedule which is now not defined anywhere in the code.

Steps to reproduce

  1. Create a pipeline with schedule_info initially set
  2. Remove schedule_info block

Important Factoids

No response

References

No response

b/328769966

ggtisc commented 8 months ago

Issue confirmed! After creates a pipeline and then remove the block of the scheduler job it remains (everything through terraform) Followed steps:

  1. Create a 'google_data_pipeline_pipeline' resource in terraform
  2. Execute 'terraform apply'
  3. Edit the 'tf' file removing the 'schedule_info' block
  4. Update resources in GCP with a 2nd 'terraform apply'
  5. Watch the resource on GCP... The schedule remains
jonathan-lemos commented 8 months ago

This happens because the underlying API doesn't support removing a schedule from a Data Pipeline. This is also why the GCP Web UI lacks this option. Can I ask what the use case of a pipeline without scheduling is? I ask because Data Pipelines is not meant to support pipelines without scheduling, and next steps depend on the use case.

kpocius commented 8 months ago

We are creating our own schedules, so we can control additional parameters such a retries.

For now a workaround is not to set the schedule on google_data_pipeline_pipeline in the first place, and just clean up old schedules manually.

Thanks for looking into this.

jonathan-lemos commented 8 months ago

Makes sense. Unfortunately, our headcount is short these days and it's unlikely that we'll have time to fix this service-side any time soon. For the time being, the workaround will have to do. Sorry.

kpocius commented 8 months ago

No worries, thanks for looking into it.