test(drive): retry transient async cleanup contention

[yxzhaao]

Summary

Stabilize Drive-backed E2E cleanup when concurrent deletes under the same folder make an asynchronous delete task report server_error: drive task failed.

Changes

• Centralize the targeted retry in drive.DeleteDriveResourceAndVerify so Slides, Docs, Sheets, Base, Wiki, and Drive cleanups share one policy.
• Verify the resource's real terminal state after each failed delete; retry only when it still exists.
• Retry only the exact api/server_error/drive task failed envelope, with five attempts and 2s/4s/8s/8s bounded backoff.
• Keep unrelated permission, validation, authentication, and protocol failures fatal.
• Simplify the Slides screenshot cleanup back to a direct shared-helper call.

Impact

CLI behavior is unchanged. Live E2E teardown becomes more resilient to the backend's same-folder delete lock while preserving bounded failure behavior.

Root Cause

Live E2E jobs and Go packages run concurrently against the same test bot. Drive serializes deletes within a folder, so overlapping cleanup tasks can collide and one async task may terminate with drive task failed.

Test Plan

• gofmt
• git diff --check
• go vet ./tests/cli_e2e/drive ./tests/cli_e2e/slides
• Live E2E retry path in CI (the repository has no Docker E2E runner for a fixed local test account)

Related Issues

• None

Summary by CodeRabbit

• Bug Fixes
  • Improved reliability when deleting drive resources by retrying temporary backend failures.
  • Added safeguards to stop retries on successful deletion, cleanup warnings, permanent failures, exhausted attempts, or canceled operations.
  • Improved presentation test cleanup for more consistent resource removal.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 7df3dad4-1ae6-4c36-b750-601473f9acc9

📥 Commits

Reviewing files that changed from the base of the PR and between e0867e6 and 8489a3b.

📒 Files selected for processing (3)

• tests/cli_e2e/drive/drive_delete_workflow_test.go
• tests/cli_e2e/drive/helpers.go
• tests/cli_e2e/slides/slides_screenshot_workflow_test.go

---

📝 Walkthrough

Walkthrough

The shared drive cleanup helper now retries recognized transient deletion-task failures with bounded exponential backoff. The slides screenshot workflow uses this helper for presentation cleanup and retains failure reporting.

Changes

Drive cleanup behavior

Layer / File(s)	Summary
Transient deletion retry handling tests/cli_e2e/drive/helpers.go	The helper adds retry and backoff limits, detects "drive task failed" JSON errors, and stops on success, warnings, cancellation, non-transient failures, or exhausted attempts.
Slides cleanup integration tests/cli_e2e/slides/slides_screenshot_workflow_test.go	The screenshot workflow uses DeleteDriveResourceAndVerify for presentation cleanup and reports returned cleanup failures.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Sequence Diagram(s)

sequenceDiagram
  participant SlidesScreenshotWorkflow
  participant DeleteDriveResourceAndVerify
  participant CLI
  participant API
  SlidesScreenshotWorkflow->>DeleteDriveResourceAndVerify: delete presentation drive resource
  DeleteDriveResourceAndVerify->>CLI: run drive deletion
  CLI->>API: execute deletion task
  API-->>CLI: return deletion result
  DeleteDriveResourceAndVerify->>DeleteDriveResourceAndVerify: retry recognized transient failures
  DeleteDriveResourceAndVerify-->>SlidesScreenshotWorkflow: return cleanup result and error

Loading

Suggested reviewers: wittam-01

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}