Tilt helm extension fails to cleanup resources and fails to restart with kyverno

[bb03]

Setup: Kubernetes cluster: kind v0.20.0 go1.20.5 linux/amd64 Python version: Python 3.9.16

With this file as kyverno-test.tilt

load("ext://helm_resource", "helm_repo", "helm_resource")
load("ext://namespace", "namespace_create")

namespace_create("kyverno")

k8s_resource(
    new_name = "kyverno-ns",
    labels = ["kyverno"],
    objects = ["kyverno:namespace"],
)

helm_repo(
    name = "kyverno",
    url = "https://kyverno.github.io/kyverno/",
    labels = ["kyverno"],
)

helm_resource(
    name = "kyverno-deployment",
    chart = "kyverno/kyverno",
    labels = ["kyverno"],
    resource_deps = [
        "kyverno",
        "kyverno-ns",
    ],
    namespace = "kyverno",
)

Repro steps I ran the following commands in this order

1. kind delete cluster && kind create cluster
2. tilt up -f kyverno-test.tilt
3. Ctrl-c to exit tilt from the legacy terminal
4. tilt down -f kyverno-test.tilt
5. tilt up -f kyverno-test.tilt

On step 4 I'm seeing the following issue

Loading Tiltfile at: <path>/tilt/kyverno/kyverno-test.tilt
Successfully loaded Tiltfile (358.015303ms)
Not deleting namespaces: kyverno
Run with --delete-namespaces to delete namespaces as well.
Running cmd: python3  <path>/.local/share/tilt-dev/tilt_modules/github.com/tilt-dev/tilt-extensions/helm_resource/helm-delete-helper.py
Running cmd: helm uninstall --namespace kyverno kyverno-deployment
Error: Deleting k8s entities for cmd: python3 <path>/.local/share/tilt-dev/tilt_modules/github.com/tilt-dev/tilt-extensions/helm_resource/helm-delete-helper.py: exit status 137

And then on step 5 I see this

kyverno-depl… │ 
kyverno-depl… │ Initial Build
kyverno-depl… │ STEP 1/1 — Deploying
kyverno-depl… │      Running cmd: python3 <path>/.local/share/tilt-dev/tilt_modules/github.com/tilt-dev/tilt-extensions/helm_resource/helm-apply-helper.py
kyverno-depl… │      Running cmd: ['helm', 'upgrade', '--install', '--namespace', 'kyverno', 'kyverno-deployment', 'kyverno/kyverno']
kyverno-depl… │      Error: UPGRADE FAILED: "kyverno-deployment" has no deployed releases
kyverno-depl… │      Traceback (most recent call last):
kyverno-depl… │        File "<path>/.local/share/tilt-dev/tilt_modules/github.com/tilt-dev/tilt-extensions/helm_resource/helm-apply-helper.py", line 71, in <module>
kyverno-depl… │          subprocess.check_call(install_cmd, stdout=sys.stderr)
kyverno-depl… │        File "/opt/pyenv/versions/3.9.16/lib/python3.9/subprocess.py", line 373, in check_call
kyverno-depl… │          raise CalledProcessError(retcode, cmd)
kyverno-depl… │      subprocess.CalledProcessError: Command '['helm', 'upgrade', '--install', '--namespace', 'kyverno', 'kyverno-deployment', 'kyverno/kyverno']' returned non-zero exit status 1.
kyverno-depl… │ 
kyverno-depl… │ ERROR: Build Failed: apply command exited with status 1

Adding --delete-namespaces to the delete command - step 4, resulted in

 % tilt down -f kyverno-test.tilt --delete-namespaces
Loading Tiltfile at: <path>/kyverno-test.tilt
Successfully loaded Tiltfile (357.493223ms)
Deleting kubernetes objects:
→ Namespace/kyverno
Running cmd: python3 <path>/.local/share/tilt-dev/tilt_modules/github.com/tilt-dev/tilt-extensions/helm_resource/helm-delete-helper.py
Running cmd: helm uninstall --namespace kyverno kyverno-deployment
Error: warning: Hook pre-delete kyverno/templates/hooks/pre-delete.yaml failed: 1 error occurred:
	* jobs.batch "kyverno-deployment-hook-pre-delete" is forbidden: unable to create new content in namespace kyverno because it is being terminated

Expected outcomes I would like to understand why the helm-delete-helper failed to delete the kyverno resourcess, and if possible, can tilt handle these types of errors so that all resources created by a tilt up get cleaned up. Thank you!

[nicks]

Hmmm... What does helm version show?

[bb03]

version.BuildInfo{Version:"v3.13.3", GitCommit:"c8b948945e52abba22ff885446a1486cb5fd3474", GitTreeState:"clean", GoVersion:"go1.21.5"}

[nicks]

(deleted a previous comment, i misunderstood something about the bug report)

hmmm...i can reproduce this in https://app.circleci.com/pipelines/github/tilt-dev/tilt-extensions/1504/workflows/1e98f416-7181-4bcc-b789-cf8d0b18f32c/jobs/3578, though not sure what's going on.

• exit status 137 usually means some sort of OOM killer
• my theory is that something is killing helm mid uninstall and leaving it in a bad state

[nicks]

ya, i think this is just tilt assuming the process got stuck and killing the process. we should have a more graceful shutdown, but for now, you can do something like:

update_settings(k8s_upsert_timeout_secs=120)

to give it a longer timeout.

[bb03]

Thanks Nick! Your workaround did the trick. Would you like me to keep this ticket open regarding having a more graceful shutdown? Or a more user-friendly timeout message?