I maybe missed something in the nova code base but I see it on the nova.conductor.rpcapi.ComputeTaskAPI:
master: https://github.com/openstack/nova/blob/78dfc7f1f5e1037892c20e867686540fccd632c2/nova/conductor/rpcapi.py#L253
pike: https://github.com/openstack/nova/blob/stable/pike/nova/conductor/rpcapi.py#L253
ocata: https://github.com/openstack/nova/blob/stable/ocata/nova/conductor/rpcapi.py#L252
Regarding the proposition to add it directly in oslo.messaging, I totally agree as this will catch all rpcs. But I don't think this will fix this particular issue (I'm still convince this is due to the wrapper evaluated at import time [1])
[1]: https://github.com/openstack/nova/blob/master/nova/profiler.py#L68
I maybe missed something in the nova code base but I see it on the nova.conductor. rpcapi. ComputeTaskAPI:
master: https:/ /github. com/openstack/ nova/blob/ 78dfc7f1f5e1037 892c20e86768654 0fccd632c2/ nova/conductor/ rpcapi. py#L253
pike: https:/ /github. com/openstack/ nova/blob/ stable/ pike/nova/ conductor/ rpcapi. py#L253
ocata: https:/ /github. com/openstack/ nova/blob/ stable/ ocata/nova/ conductor/ rpcapi. py#L252
Regarding the proposition to add it directly in oslo.messaging, I totally agree as this will catch all rpcs. But I don't think this will fix this particular issue (I'm still convince this is due to the wrapper evaluated at import time [1])
[1]: https:/ /github. com/openstack/ nova/blob/ master/ nova/profiler. py#L68