Regression in Job deployment caused by #2596

[justinmchase]

Terraform Version, Provider Version and Kubernetes Version

Terraform version:
Kubernetes provider version: v2.34.0
Kubernetes version: AKS 1.31.1

Affected Resource(s)

• kubernetes_job

Terraform Configuration Files

What do you mean by "configuration"? Certainly you don't mean the entire terraform module.

Debug Output

Working on this, will attach soon.

Panic Output

N/A

Steps to Reproduce

1. Use kuberenetes provivder v2.34.0 or greater.
2. Create a kubernetes_job resource which does not have ttl_seconds_after_finished set.

Expected Behavior

1. The job should be recreated on every deployment
2. The deployment should succeed

Actual Behavior

1. The job is created and the job pod runs successuflly
2. Terraform returns a non-zero exit code and reports the Job as failing.

Important Factoids

This code has been running fine for a long time and the issue may be caused by this PR that was just merged #2596

Also, and I don't fully understand the PR creators scenario but this assumption seems bad to me:

This can cause Terraform to plan the recreation of the Job in subsequent runs, which is sort of undesirable behavior.

In fact this is highly desirable and required behavior for us. We explicitly want the Job to be re-created on every single deployment.

So is this change a regression? Is there some setting I can set it to that will cause it to be recreated on every deployment?

Regardless, having it fail even though the pod is succeeding seems like a bug.

References

• Handle Jobs with ttl_seconds_after_finished = 0 correctly #2596

Community Note

• Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
• If you are interested in working on this issue or have submitted a pull request, please leave a comment

[justinmchase]

We were able to rollback to version 2.33.0 and it resolved our issue, so its definitely a recent issue probably this PR specifically.