Copy pasting this from my email:
I think I found a serious bug in our usage of eventlet thread local storage. Please check out this snippet [1].
This is how we use eventlet TLS in Nova and common Oslo code [2]. This could explain how [3] actually breaks TripleO devtest story and our gates.
Am I right? Or I am missing something and should get some sleep? :)
[1] http://paste.openstack.org/show/53686/ [2] https://github.com/openstack/nova/blob/master/nova/openstack/common/local.py#L48 [3] https://github.com/openstack/nova/commit/85332012dede96fa6729026c2a90594ea0502ac5
Copy pasting this from my email:
I think I found a serious bug in our usage of eventlet thread local
storage. Please check out this snippet [1].
This is how we use eventlet TLS in Nova and common Oslo code [2]. This
could explain how [3] actually breaks TripleO devtest story and our
gates.
Am I right? Or I am missing something and should get some sleep? :)
[1] http:// paste.openstack .org/show/ 53686/ /github. com/openstack/ nova/blob/ master/ nova/openstack/ common/ local.py# L48 /github. com/openstack/ nova/commit/ 85332012dede96f a6729026c2a9059 4ea0502ac5
[2] https:/
[3] https:/