ci: clean TKE checkin env on failure

[aptend]

Summary

• stop recreating the TKE checkin namespace in cleanup
• run environment cleanup whenever image build succeeded, even if tests failed
• keep trace collection/upload before deleting the namespace

Validation

• inspected workflow diff and cleanup step ordering
• local YAML parser/actionlint unavailable in this environment

[Copilot]

Pull request overview

This PR updates the TKE “checkin regression” cleanup behavior in the merge-trigger-tke workflow so the environment is cleaned up whenever the Docker image build succeeded, even if downstream tests fail, while keeping trace collection/upload before namespace deletion and avoiding namespace recreation during cleanup.

Changes:

• Remove namespace recreation from the cleanup job, replacing it with a non-failing namespace check.
• Run the cleanup step under always() (gated by successful image build) rather than requiring all test jobs to succeed.
• Add a guard to skip cleanup work when the namespace does not exist.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[aptend]

Addressed the namespace check review: cleanup now uses kubectl get namespace <name> directly and only skips when stderr indicates NotFound. Other kubectl errors are surfaced and fail the cleanup step instead of silently skipping.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 1 comment.

[aptend]

Addressed the latest review as well: the namespace check no longer writes to a fixed /tmp path. It captures kubectl stderr in a local shell variable and only skips cleanup when the captured error contains NotFound; other errors still fail the step.

[aptend]

✅ Review Complete

The changes look solid and address the original issue well:

Strengths:

1. Better failure handling: Running cleanup with always() gated by successful image build ensures environment cleanup even when tests fail - prevents resource waste
2. No namespace recreation: Removes the redundant namespace recreation, replacing it with a simple check. This is cleaner and avoids potential race conditions
3. Proper error handling: The new namespace check properly handles NotFound vs other errors, allowing graceful skip when namespace doesn't exist
4. Trace collection preserved: Keeps Check MO Status and Collect Trace before cleanup to ensure diagnostic data is captured

Implementation Details:

• The namespace existence check using kubectl get namespace with proper error capture is correct
• The condition if: ${{ always() && needs.docker_image_build.result == 'success' }} ensures cleanup runs when image build succeeds, regardless of test results
• Guard logic properly distinguishes between "namespace doesn't exist" (exit 0) and actual errors (exit 1)

Minor note:

• The logic is sound. The YAML structure and variable substitution follow GitHub Actions best practices.

Great job on improving the TKE cleanup robustness!