Comment 12 for bug 1320617

I'd like to see us use the diagnostics API in Tempest to get information off the instance when we hit a timeout, just haven't had the time yet.

http://docs.openstack.org/api/openstack-compute/2/content/GET_os-diagnostics-v2_getDiagnostics__v2__tenant_id__servers__server_id__diagnostics_ext-diagnostics.html

Note that not all virt drivers implement that so we'd have to handle that appropriately in Tempest. It's also inconsistent data in the V2 API, but that's OK in my opinion for just dumping it to the tempest/console log for debug.

As for the glance/cinder changes to devstack for running as many API workers as there are CPUs on the host, I've moved that work to the actual projects rather than devstack, tracked via bug 1333325 and bug 1333370.