So the regression definitely seems to have been introduced by https://review.openstack.org/#/c/156923/
I see three different ways we could address it, but I'm not sure which is best. Maybe some combination?
1. Make copy more picky about when it sets orig_req_method, as in http://paste.openstack.org/show/601652/
2. Make SLO consciously opt out of any orig_req_method that may be set, as in http://paste.openstack.org/show/601653/
3. Make proxy-logging only respect orig_req_method when swift.source is unset (i.e., for client requests), as in http://paste.openstack.org/show/601654/
So the regression definitely seems to have been introduced by https:/ /review. openstack. org/#/c/ 156923/
I see three different ways we could address it, but I'm not sure which is best. Maybe some combination?
1. Make copy more picky about when it sets orig_req_method, as in http:// paste.openstack .org/show/ 601652/
2. Make SLO consciously opt out of any orig_req_method that may be set, as in http:// paste.openstack .org/show/ 601653/
3. Make proxy-logging only respect orig_req_method when swift.source is unset (i.e., for client requests), as in http:// paste.openstack .org/show/ 601654/