Comment 7 for bug 1254238

I investigated this problem based on http://logs.openstack.org/15/90015/10/check/check-tempest-dsvm-full/84aef9b/logs/ .

The scenario is the following:

04:35:42.111 "create a server01" API returned HTTP202
04:35:42.358 (n-cpu) Started to create server01
04:35:42.629 "create a server02" API returned HTTP202
04:35:42.945 "create a server03" API returned HTTP202
04:35:43.458 "delete a server03" API returned HTTP202
[..]
04:35:43.737 the problem happened because "list servers" API with change-since parameter could not find server02.
04:35:44.728 (n-cpu) Started to create server02
04:35:44.752 (n-cpu) Started to create server03
04:35:44.847 "delete a server02" API returned HTTP202

The test could not find server02 because the DB record of server02 was just created and not updated.
The change-since parameter of "list servers" API finds the updated server, and it does not catch the non-updated servers.
As the above log, the way to create server02 was started after the test, and I guess this situation would be due to high stress environment.
So I feel we need to avoid this problem with Tempest test case change.