Vouch request: [Manoj-engineer]

[Manoj-engineer]

What do you want to work on?

I want to fix issue #211 — improving the error message when a Helm chart has malformed YAML during gateway startup

Why this change?

when I was first trying to get a local gateway running. I had a template issue in a custom chart and spent 20 minutes staring at "namespace not ready" before I figured out I needed to manually run kubectl logs -n kube-system inside the container.

The real error is buried in a job log the user doesn't know to look at. I want to fix wait_for_namespace in openshell-bootstrap so that on timeout it checks whether any helm-install-* Job in kube-system has a non-zero status.failed and if so, pulls the job's status conditions and last 30 lines of pod logs and includes them directly in the error message alongside a kubectl logs hint.

That's a painful first-run experience, especially for people who aren't already comfortable with k8s internals. The fix is small, contained entirely in [lib.rs], touches no public API.

Checklist

• I wrote this request myself (not AI-generated)
• I have read CONTRIBUTING.md

[drew]

Thanks for starting a discussion on this, could you expand a bit on "I had a template issue in a custom chart"? What do you mean by custom chart?

/vouch

[Manoj-engineer]

Sure — it was the built-in OpenShell chart, not something I brought in. I was editing serviceaccount.yaml locally and accidentally merged apiVersion: v1 with the line below it, so the YAML was invalid. gateway start timed out with the generic "namespace not ready" — the real error was buried in the Helm job log inside the container, which I only found by manually running kubectl logs inside the gateway.

I've implemented and live-tested the fix. On timeout, wait_for_namespace now checks helm-install-* jobs, finds the failure, and prints the actual Helm error in the terminal alongside a kubectl logs hint. Contained entirely in lib.rs, no public API changes, all 78 tests pass. Also caught a secondary bug during testing — the job failure check was silently skipping jobs still retrying because status.failed stays until retries are exhausted. Fixed by checking != "0" instead.

Happy to open the PR whenever vouched.

[drew]

You should be all set to open a PR. Thanks in advance for your contribution.