[openstack-dev] Problem with debugging program having eventlet.monkey_patch() code
mordred at inaugust.com
Thu Sep 27 17:31:09 UTC 2012
On 09/27/2012 10:23 AM, Monty Taylor wrote:
> On 09/27/2012 10:05 AM, Adam Young wrote:
>>> As a quick workaround you can change monkey patching at nova-scheduler
>> In Keystone, we have been moving toward a command line switch
>> --no_monkeypatch that as of now is in the test runner, and I would like
>> to submit for the keystone-all code as well. I origianlly thought I
>> would be removing all of the monkeypatching, until I realized that it
>> was just the thread monkeypatch that was causing the debugging problem.
> When you say "in the test runner" - where do you mean? Monkeypatching
> should be occuring the test base classes at this point, no?
> Could I make a feature request that the conditional that controls this
> also responds to an env-var so that we can have the same functionality
> when using standard test runners?
Hrm. My bad - I didn't realize we still had a run_tests.py in keystone.
We removed it from nova and glance last cycle and moved the various
monkeypatching and log setup code into test base classes so that we
could just use the standard nose test runner ... which in turn is a step
close to being able to use any test runner which will let us get to
advanced things like parallel test running.
But somehow I missed keystone- for which I apologize.
In any case, the env-var approach request still stands, and for the two
different areas run_test.sh can give a command line option to
run_tests.py in keystone, and can just set an env var before launching
nose in nova/glance.
>> Can we standardize a name for this, and use it across all projects?
>> Something like:
>> OpenStack-dev mailing list
>> OpenStack-dev at lists.openstack.org
> OpenStack-dev mailing list
> OpenStack-dev at lists.openstack.org
More information about the OpenStack-dev