ensure servers are deleted between tests
Possible fix for bug #1079687
When running tests we create a lot of servers, and do delete
calls at the end of tests. However delete is an async action,
so the previous servers will often be hanging around after our
tests. In memory constrained environments like the OpenStack CI
system if deletes happen too slowly we hit the memory limits,
nova scheduler can no longer schedule VMs, and the tests get flakey.
This *may* fix the periodic fails we've seen on tempest in nova.
Pair programmed with Matt Treinish <treinish@linux.vnet.ibm.com>
Change-Id: I1f93ac26064c68bf529e2efa1bf0c35e1f4e7d2c
1 file changed