<div dir="ltr"><div class="gmail_quote"><div dir="ltr">Hi folks,<div><br></div><div>Now that it looks like the metadata proposal is more refined [0], I'd like</div><div>to get some feedback from you on the driver implementation.</div><div><br></div><div>The ovn-metadata-agent in networking-ovn will be responsible for</div><div>creating the namespaces, spawning haproxies and so on. But also,</div><div>it must implement most of the "old" neutron-metadata-agent functionality</div><div>which listens on a UNIX socket and receives requests from haproxy,</div><div>adds some headers and forwards them to Nova. This means that we can</div><div>import/reuse big part of neutron code.</div><div><br></div><div>I wonder what you guys think about depending on neutron tree for the</div><div>agent implementation despite we can benefit from a lot of code reuse.</div><div>On the other hand, if we want to get rid of this dependency, we could</div><div>probably write the agent "from scratch" in C (what about having C</div><div>code in the networking-ovn repo?) and, at the same time, it should</div><div>buy us a performance boost (probably not very noticeable since it'll</div><div>respond to requests from local VMs involving a few lookups and</div><div>processing simple HTTP requests; talking to nova would take most</div><div>of the time and this only happens at boot time). </div><div><br></div><div>I would probably aim for a Python implementation reusing/importing</div><div>code from neutron tree but I'm not sure how we want to deal with</div><div>changes in neutron codebase (we're actually importing code now).</div><div>Looking forward to reading your thoughts :)</div><div><br></div><div>Thanks,</div><div>Daniel</div><div><br></div><div>[0] <a href="https://review.openstack.org/#/c/452811/" target="_blank">https://review.openstack.<wbr>org/#/c/452811/</a></div></div>
</div><br></div>