yahoo-eng-team team mailing list archive

[Thierry Carrez <thierry.carrez+lp@xxxxxxxxx>]

** Changed in: nova
       Status: Fix Committed => Fix Released

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Compute (nova).
https://bugs.launchpad.net/bugs/1218133

Title:
  MultiprocessWSGITest.test_terminate_sigterm fails because the wait for
  workers to die is unreliable.

Status in OpenStack Compute (Nova):
  Fix Released

Bug description:
  The test run logged at http://logs.openstack.org/63/43863/1/gate/gate-
  nova-python26/a49f89c/ appears to have failed because the waits at
  http://git.openstack.org/cgit/openstack/nova/tree/nova/tests/integrated/test_multiprocess_api.py#n140
  and
  http://git.openstack.org/cgit/openstack/nova/tree/nova/tests/integrated/test_multiprocess_api.py#n163
  are unreliable. Instead these tests should probably use something like
  a waitpid with WNOHANG option poll loop where the worker processes are
  polled rather than waited for. This will allow the tests to use a much
  longer timeout as they will short circuit when all workers have
  exited.

  Additionally, the try except at
  http://git.openstack.org/cgit/openstack/nova/tree/nova/tests/integrated/test_multiprocess_api.py#n106
  may not work in all cases. If a timeout exception happens anywere else
  in the test this code will not be run. A fix for this problem would be
  to put the os.kill at
  http://git.openstack.org/cgit/openstack/nova/tree/nova/tests/integrated/test_multiprocess_api.py#n112
  in a test cleanup rather than a test teardown so that it runs all the
  time and is forceful with the worker process killing. A bit bruteforce
  but hopefully that results in more consistent tests.

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1218133/+subscriptions