[Bug]: ate-setup delete demo returns before the demo namespace finishes terminating

オープン
  • 何か問題がありますか？
#3,525 コメント 1 件 リアクション 0 件 担当者 2 名 GitHub で見る

メンテナーはふだん 1 日以内に返信

@HavenXia がすでに取り組んでいます。

2026年10月31日 から。

評価

この issue はまだ評価されていません。

説明

area/cli area/demos kind/bug
What happened?

ate-setup delete demo <name> (Substrate.Delete in cmd/ate-setup/internal/demos/substrate.go) calls e.Kube.DeleteBytes(ctx, manifest) on the demo's WorkerPoolManifest and returns while the demo Namespace is still Terminating.

Running ate-setup deploy demo <name> right after ate-setup delete demo <name> (for example, redeploying counter with --with-external-volume) fails because the Kubernetes NamespaceLifecycle admission controller rejects new objects in a terminating namespace:

Error: while applying workerpool/counter -n ate-demo-counter: workerpools.ate.dev "counter" is forbidden: unable to create new content in namespace ate-demo-counter because it is being terminated
Expected Behavior

ate-setup delete demo <name> and ate-setup deploy demo <name> can be run back to back, or the async behavior is an explicit CLI contract.

Steps to Reproduce
  1. Deploy ate-system and the counter demo on a cluster: 
    go run ./cmd/ate-setup deploy demo counter
  2. Delete and immediately redeploy the demo: 
    go run ./cmd/ate-setup delete demo counter
go run ./cmd/ate-setup deploy demo counter
  3. deploy demo counter fails applying workerpool/counter while namespace/ate-demo-counter is still terminating.
Sandbox Runtime

Both / Runtime Agnostic

Agent Substrate Version / Commit SHA

8a2c01563

Kubernetes Version & Environment

kind v0.31.0 / Kubernetes v1.35

Host OS & Architecture

Linux (x86_64)

Relevant Logs and Diagnostic Output
[step]: demo-counter_deploy
[step]: ensure_crds
[step]: images: built from source with ko
  (ko resolve -f - took 7.003s)
  (apply of 2 objects took 0.017s)
Error: while applying workerpool/counter -n ate-demo-counter: workerpools.ate.dev "counter" is forbidden: unable to create new content in namespace ate-demo-counter because it is being terminated
Additional Context

Client.Delete (cmd/ate-setup/internal/kube/apply.go) documents itself as the kubectl delete --ignore-not-found -f equivalent. kubectl delete defaults to --wait=true, whereas Client.DeleteOne calls ri.Delete via client-go and returns as soon as deletion is requested.

Meanwhile, Substrate.Deploy blocks synchronously on workload rollouts and golden snapshot creation, and DeleteAteSystem (cmd/ate-setup/internal/steps/delete.go) calls e.Kube.WaitDeleted on ate-system and podcertificate-controller-system.

For discussion: should ate-setup delete demo <name> wait on demo Namespace deletion by default (matching kubectl delete and DeleteAteSystem), or keep deletion asynchronous and leave kubectl wait --for=delete namespace/<ns> to callers that immediately redeploy?

Confirmation
  • I have searched existing issues and verified that this is not a duplicate.
  • I have verified that this issue occurs on the latest commit on main.
主要言語
Go
スター
5.9k
フォーク
637
平均マージ
2日 1時間
マージ済み PR（30日）
523

環境構築

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

agent-substrate/substrate のほかの issue

agent-substrate/substrate の issue をすべて見る

似ている issue

Go の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。