Skip to content

Fix: Prevent potential test environment pollution in shared scale clusters#626

Open
weizhoublue wants to merge 1 commit into
ai-dynamo:mainfrom
weizhoublue:fix-scale-test-cleanup
Open

Fix: Prevent potential test environment pollution in shared scale clusters#626
weizhoublue wants to merge 1 commit into
ai-dynamo:mainfrom
weizhoublue:fix-scale-test-cleanup

Conversation

@weizhoublue

@weizhoublue weizhoublue commented May 20, 2026

Copy link
Copy Markdown
Contributor

…ster pollution

What type of PR is this?

/kind bug

What this PR does / why we need it:

Issue Summary

Running scale tests (e.g., make run-scale-test) against a persistent or reused Kubernetes cluster.

In PR #621, WithSkipCleanupWait() was intentionally introduced in operator/e2e/tests/scale/scale_test.go to measure only the PodCliqueSet deletion latency without waiting for the slow Kubernetes API server and GC to delete all sub-resources (Pods) .
However, an oversight occurred: no thorough final cleanup was enforced at the suite level. Consequently, the test suite exits without ever waiting for the cascading deletion to complete.
A large number of Terminating resources may remain in the cluster for seconds after the scale tests finish.

When the same cluster is reused for subsequent test runs, these leftover resources pollute the environment, leading to scheduling conflicts, test misjudgments, and resource exhaustion.

Which issue(s) this PR fixes:

Fixes #

Special notes for your reviewer:

Does this PR introduce a API change?


Additional documentation e.g., enhancement proposals, usage docs, etc.:


…ster pollution

Signed-off-by: User <user@example.com>
@copy-pr-bot

copy-pr-bot Bot commented May 20, 2026

Copy link
Copy Markdown

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@weizhoublue weizhoublue changed the title Fix: Prevent test environment pollution and cross-test flakiness in shared scale clusters Fix: Prevent potential test environment pollution in shared scale clusters May 20, 2026

@renormalize renormalize left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the PR @weizhoublue!

Each contribution in grove requires a GitHub issue describing the bug/feature, and the corresponding issue is linked with an implementation PR.

So could you please specify the bug as you've described in this PR's description in a GitHub issue, and then link to that issue in this PR? We have an action that verifies this, and that is currently failing.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants