This is a continuation of the "next steps" topic [0]. I created a blueprint [1] and related etherpad [2] to describe failure modes (timeouts, retrurn codes and so on) at least for the Oslo messaging RPC API calls, which seem the most critical point for app and the library devs. As the end goal, an example [3] should describe a full list of possible situations (considering the given failure modes). Instead of a brief "Invoke a method and wait for a reply." Makes sense? Thank you, Konstantin Kalin for clearly defined basic use cases. If you think there are more, please describe them at the etherpad. [0] http://lists.openstack.org/pipermail/openstack-dev/2015-June/067471.html [1] https://goo.gl/0U5P9D [2] https://etherpad.openstack.org/p/oslo.messaging.api.failure.modes [3] http://goo.gl/G0uQPk -- Best regards, Bogdan Dobrelya, Irc #bogdando