Bug #10535

test suite hangs (in arvbox) in sdk/python test_callback

Added by Joshua Randall almost 3 years ago. Updated almost 2 years ago.

Status:
New
Priority:
Normal
Assigned To:
Category:
SDKs
Target version:
Start date:
11/15/2016
Due date:
% Done:

100%

Estimated time:
Story points:
-

Description

                 ********** Running sdk/python tests **********

running test
running egg_info
writing requirements to arvados_python_client.egg-info/requires.txt
writing arvados_python_client.egg-info/PKG-INFO
writing top-level names to arvados_python_client.egg-info/top_level.txt
writing dependency_links to arvados_python_client.egg-info/dependency_links.txt
writing pbr to arvados_python_client.egg-info/pbr.json
reading manifest file 'arvados_python_client.egg-info/SOURCES.txt'
reading manifest template 'MANIFEST.in'
writing manifest file 'arvados_python_client.egg-info/SOURCES.txt'
running build_ext
test_callback (tests.test_events.PollClientTestCase) …

This test gets stuck for an apparently infinite amount of time on my system, also with slowly increasing memory usage. Seems to be a race condition involving PollClient. Adding a time.sleep(0.1) before `self.logs.add(test_log.copy())` (https://github.com/curoverse/arvados/blob/0b5d04beb288175a285c36a38f255399dfe7d0d7/sdk/python/tests/test_events.py#L338-L339) avoids the issue on my system.

Associated revisions

Revision a983bbee (diff)
Added by Tom Clegg almost 3 years ago

10535: Fix race-to-deadlock bugs in tests.test_events: use threading.Event, not threading.Condition. refs #10535

History

#1 Updated by Joshua Randall almost 3 years ago

  • Assigned To set to Joshua Randall
  • % Done changed from 0 to 100

#2 Updated by Joshua Randall almost 3 years ago

I suspect you may be able to reproduce this bug if you run the tests on a system with only one CPU. I had been running arvbox on a VirtualBox VM configured to have only one CPU.

#3 Updated by Joshua Randall almost 3 years ago

I had a chance to test the unpatched version of this test over the weekend and it appears that it hangs on test_callback for me even with 4 CPUs, so that doesn't explain the difference.

#4 Updated by Tom Morris almost 2 years ago

  • Target version set to Arvados Future Sprints

Also available in: Atom PDF