Hire engineers who solve
your problems.

HackerRank and CodeSignal show you who can solve puzzles. Canary shows you who can do the job.

Join waitlist Try demo →

Candidate experience

canary|CAR-523·Sandbox temp dirs not cleaned up on failure

37:42|▶ RunSubmit

Problem

🎫 CAR-523

⎇ PR #1098

CAR / Backlog / CAR-523

Sandbox temp directories not cleaned up on failure

● In Progress

INC-3201 — Disk usage on the execution host hit 94% on 2026-05-10, causing new runs to fail with No space left on device.

Orphaned temp dirs from timed-out and crashed runs — cleanup() never runs when run_tests() raises.

Workspace

solution.py

tests.py

README.md

from sandbox import create_sandbox, write_files

from sandbox import run_tests, cleanup

def execute_run(req):

sandbox = create_sandbox(req["run_id"])

write_files(sandbox, req["code"])

result = run_tests(sandbox, req["timeout"])

cleanup(sandbox) # only runs on success ↑

return result

Tests 2/5

✓successful run returns output

✓cleanup after success

✗cleanup after timeoutSandbox dir still exists

HackerRank tells you if they can code. Canary tells you if they can work.

Your bugs, not textbook puzzles

HackerRank and CodeSignal test whether candidates can reverse a linked list. Canary tests whether they can fix the thing that broke your system last week.

Same bar for every candidate

Solutions run against your hidden test suite in a sandboxed environment. No self-grading, no vibes — just a consistent, repeatable signal.

How they think, not just what they shipped

Algo platforms show you a score. Canary shows you the session — every edit, every AI prompt, every dead end they backed out of.

See a live problem →

Get early access.