Workers SDK Issue Reports

← Back to Dashboard

#11031 vitest-pool-runner Concurrent DO requests fail storage cleanup

Recommendation:KEEP OPEN
Difficulty:medium
Reasoning:

Root cause identified: SQLite temp files not cleaned up; reproduction available

Suggested Action:

Fix loopback.ts to handle SQLite temp files

View on GitHub →

Analysis Report

Issue Review: cloudflare/workers-sdk#11031

Summary

Concurrent DO requests that throw exceptions cause vitest-pool-workers to fail popping isolated storage stack frame due to SQLite temp files.

Findings

  • Created: 2025-10-20
  • Updated: 2026-01-26
  • Version: @cloudflare/vitest-pool-workers 0.9.13, wrangler 4.43.0
  • Component: vitest-pool-workers
  • Labels: bug, vitest
  • Comments: 4

Key Evidence

  • Reproduction repo: https://github.com/matthewdavidrodgers/pop-do-storage-bug-repro
  • Root cause identified: SQLite shared memory files (.sqlite-shm) remain after concurrent DO operations
  • Code location: packages/vitest-pool-workers/src/pool/loopback.ts:280 doesn't handle SQLite temp files
  • Also affects D1 (confirmed by @erkkah) and DO alarms (confirmed by @dmaretskyi)
  • Recent comment (Jan 2026) shows users still affected

Recommendation

Status: KEEP OPEN

Reasoning: This is a valid bug with clear root cause identified. The vitest-pool-workers isolated storage cleanup doesn't account for SQLite temp files (.sqlite-shm, .sqlite-wal). Multiple users confirm the issue with different scenarios.

Action: Fix loopback.ts to handle SQLite temporary files during storage stack cleanup.

Suggested Comment

N/A - Root cause identified, awaiting fix.

Notes & Feedback (0)

No notes yet.

Add Note