I read these logs slightly differently. I don't think the test ever got to the attach call, based on looking at the code that was present on June 17 and the fact that addFIxedIp only shows up once: in the JSON version of this test. In the old code, we call _console_dump when a server unexpectedly goes to ERROR. Further, in that June 17 code, there is a bug in the XML get_console code where it does not handle length of None. In the current code we no longer dump the console in this case, and the XML console bug was fixed.
doesn't that mean that the create failed, we got an error trying to dump console in the wait error handler, and then a DELETE happened?
I am not denying that there is a race around the call to attach but I don't see this test run as triggering it and don't quite understand exactly where you see a race in the code.
I read these logs slightly differently. I don't think the test ever got to the attach call, based on looking at the code that was present on June 17 and the fact that addFIxedIp only shows up once: in the JSON version of this test. In the old code, we call _console_dump when a server unexpectedly goes to ERROR. Further, in that June 17 code, there is a bug in the XML get_console code where it does not handle length of None. In the current code we no longer dump the console in this case, and the XML console bug was fixed.
Given this in the console
2014-06-17 19:15:15.665 10882 INFO tempest. common. waiters [-] State transition "BUILD/scheduling" ==> "ERROR/None" after 1 second wait common. rest_client [req-5d3b75e5- c6c9-4b55- b9f8-050824bcaf f7 None] Request (AttachInterfac esTestXML: test_add_ remove_ fixed_ip) : 400 POST http:// 127.0.0. 1:8774/ v2/97b84e38f795 456d8c63a64526f 8e8b5/servers/ 94bb0663- 1b9e-4098- bedf-9c7d4fc6f8 c9/action 0.116s common. rest_client [req-fd7aab03- cb8a-44a6- b8e7-aed5ae2f36 18 None] Request (AttachInterfac esTestXML: test_add_ remove_ fixed_ip) : 204 DELETE http:// 127.0.0. 1:8774/ v2/97b84e38f795 456d8c63a64526f 8e8b5/servers/ 94bb0663- 1b9e-4098- bedf-9c7d4fc6f8 c9 0.320s
2014-06-17 19:15:15.782 10882 INFO tempest.
2014-06-17 19:15:16.103 10882 INFO tempest.
doesn't that mean that the create failed, we got an error trying to dump console in the wait error handler, and then a DELETE happened?
I am not denying that there is a race around the call to attach but I don't see this test run as triggering it and don't quite understand exactly where you see a race in the code.