Comment 13 for bug 1568191

Revision history for this message
Brian Rosmaita (brian-rosmaita) wrote :

I have a few questions around this issue.

The Compute os-extra_specs API allows you to put arbitrary key/value pairs on a flavor. How do operators discover what keys actually work with Nova, so they know to use hw_cpu_max_sockets instead of hw_cpu_maxsockets?

For interoperability, it makes sense to use the Glance metadefs catalog to define the VirtCPUTopology metadata keys and value datatypes. But operators who are doing that now are apparently putting non-effective metadata keys on their images. Would the non-effectiveness be obvious to operators, so that they'd correct the keys manually, or could this go unnoticed?

I'm asking because if you have a large number of images and server snapshot images around, do we expect an operator to migrate the keys on all these images? A better solution might be to change the nova code to accept either keyname. On the other hand, if the ineffectiveness is so obvious that the ops would have corrected their systems already, then maybe we don't have to worry about backward compatability so much. But as Flavio and Erno already commented on the patch, I'm worried about the effect this change will have on current operators.

It's an empirical question, we can go to the ML to ask. I'd like to know about the current situation with respect to discoverability, though.