Describe the bug:
If there are more than 30 error tasks that could not be handled immediately by the safety worker, then crash happens.
- Spawn 30+ safe tasks, return fail for all of them
- Prevent immediate handling of the failure e.g. by waiting for all of them in
join_all
- Crash will happen when 31st task returns error.
Steps to reproduce the behavior:
git checkout igorostrowskiq_safety_worker_testscd component_integration_tests/rust_test_scenarioscargo run -- --name runtime.worker.safety_worker.queue_safety_worker_tasks --input '{"runtime": {"task_queue_size": 256, "workers": 1, "safety_worker": {}}, "test": {"task_count": 31}}'
Observed behavior:
Segmentation fault happens
Expected behavior
No segmentation fault. If there is hard limit for such tasks, there should be a proper documentation + handling of such case (panic with detailed info?)
Occurrence:
Always
Attachments / Logs:
crash_log.txt
Original report: https://github.com/qorix-group/inc_orchestrator_internal/issues/395
Describe the bug:
If there are more than 30 error tasks that could not be handled immediately by the safety worker, then crash happens.
join_allSteps to reproduce the behavior:
git checkout igorostrowskiq_safety_worker_testscd component_integration_tests/rust_test_scenarioscargo run -- --name runtime.worker.safety_worker.queue_safety_worker_tasks --input '{"runtime": {"task_queue_size": 256, "workers": 1, "safety_worker": {}}, "test": {"task_count": 31}}'Observed behavior:
Segmentation fault happens
Expected behavior
No segmentation fault. If there is hard limit for such tasks, there should be a proper documentation + handling of such case (panic with detailed info?)
Occurrence:
Always
Attachments / Logs:
crash_log.txt