Setting scheduler timeout=0 results in Timeout being raised outside the task

[Steve Baker]

When timeout=0 the Timeout is raised, but can't be caught by the task code.

Setting timeout=0 is desirable in some unit tests.

[OpenStack Infra (hudson-openstack)]

Related fix proposed to branch: master
Review: https://review.openstack.org/109885

[Steve Baker (steve-stevebaker)]

With the above gerrit change, test_timeout_zero passes when it should fail, and the following is logged:

Traceback (most recent call last):
  File "/usr/lib64/python2.7/runpy.py", line 162, in _run_module_as_main
    "__main__", fname, loader, pkg_name)
  File "/usr/lib64/python2.7/runpy.py", line 72, in _run_code
    exec code in run_globals
  File "/usr/lib/python2.7/site-packages/subunit/run.py", line 131, in <module>
    main()
  File "/usr/lib/python2.7/site-packages/subunit/run.py", line 127, in main
    stdout=sys.stdout)
  File "/usr/lib/python2.7/site-packages/testtools/run.py", line 220, in __init__
    self.runTests()
  File "/usr/lib/python2.7/site-packages/testtools/run.py", line 371, in runTests
    self.result = testRunner.run(self.test)
  File "/usr/lib/python2.7/site-packages/subunit/run.py", line 62, in run
    test(result)
  File "/usr/lib64/python2.7/unittest/suite.py", line 70, in __call__
    return self.run(*args, **kwds)
  File "/usr/lib64/python2.7/unittest/suite.py", line 108, in run
    test(result)
  File "/usr/lib64/python2.7/unittest/case.py", line 433, in __call__
    return self.run(*args, **kwds)
  File "/usr/lib/python2.7/site-packages/testscenarios/testcase.py", line 65, in run
    return super(WithScenarios, self).run(result)
  File "/usr/lib/python2.7/site-packages/testtools/testcase.py", line 568, in run
    return self.__RunTest(self, self.exception_handlers).run(result)
  File "/usr/lib/python2.7/site-packages/testtools/runtest.py", line 74, in run
    return self._run_one(actual_result)
  File "/usr/lib/python2.7/site-packages/testtools/runtest.py", line 88, in _run_one
    return self._run_prepared_result(ExtendedToOriginalDecorator(result))
  File "/usr/lib/python2.7/site-packages/testtools/runtest.py", line 100, in _run_prepared_result
    self._run_core()
  File "/usr/lib/python2.7/site-packages/testtools/runtest.py", line 125, in _run_core
    self.case._run_test_method, self.result):
  File "/usr/lib/python2.7/site-packages/testtools/runtest.py", line 176, in _run_user
    return self._got_user_exception(sys.exc_info())
  File "/usr/lib/python2.7/site-packages/testtools/runtest.py", line 203, in _got_user_exception
    raise e
heat.engine.scheduler.Timeout

[Zane Bitter (zaneb)]

When we throw() the Timeout, it gets raised at the point where the co-routine is stopped - which may be at a yield, or in this case at the very beginning (where it is impossible to catch). The alternative would be to advance the co-routine to the first yield using next() before calling throw()... however this is never what you want in the real world. If you _know_ that the operation is going to time out, it's much better to abort it before it starts than to get it into a half-executed state.

An easier approach for the unit tests might be to just use a very short timeout rather than 0, or to plan on catching the Timeout outside of the co-routine.

[OpenStack Infra (hudson-openstack)]

Change abandoned by Steve Baker (<email address hidden>) on branch: master
Review: https://review.openstack.org/109885

[Steve Baker (steve-stevebaker)]

timeout=0.01 works fine for https://review.openstack.org/#/c/102397/

This bug should probably remain, and the fix could be to raise an exception when timeout <= 0

[Shashwat Srivastava (shashwat-srivastava)]

Change has been successfully merged into the git repository. Changing the status.

[Zane Bitter (zaneb)]

Actually, no change was ever merged, and nor should it be.