Inconclusive result on Kubernetes Job metric

[object88]

Summary

The current implementation of the Job metric does not allow for an "inconclusive" result. This puts it at a technical deficit against other metrics.

Use Cases

I can think of 2 use cases:

1. The job is unable to start, as with a CLB or Image Pull Error, as mentioned in Issue 1329
2. The job successfully starts, but it itself is unable to determine if the results was certainly a success or a failure.

My personal use-case is the second: we have a Job that starts a GHA workflow. In some cases, the workflow simply does not run or complete in time (say, because Github is down, or we have a terrible backlog). I would like to be able to go to an Argo console or CLI and indicate whether the promotion should proceed. For example, maybe I know that my GHA workflow (some particular time) is going to fail due to $REASONS, but I want to ignore that and proceed anyway.

The current implementation only allows me to work in absolutes.

Notes

The framework for an inconclusive result is established for other metrics.

I would propose that if a Job returns JSON, the existing evaluation could be applied.

I would also propose that the Job be augmented with a new property that describes the anticipated reaction if the Job does not start or has a non-zero exit code. The default would be consistent with current behavior (success) so that existing Jobs are not impacted. Non-default values would indicate that the Job was a fail or inconclusive.

This should maybe be two separate proposals?

---

Message from the maintainers:

Impacted by this bug? Give it a 👍. We prioritize the issues with the most 👍.

[github-actions[bot]]

This issue is stale because it has been open 60 days with no activity.

[github-actions[bot]]

This issue is stale because it has been open 60 days with no activity.

[Abhishek-Kaudare]

Is there any updates on this enhancement?

[github-actions[bot]]

This issue is stale because it has been open 60 days with no activity.