Re: [openstack-dev] [ironic][neutron] ML2 plugin for Ironic needs
On Mon, Oct 10, 2016 at 4:32 AM, Vasyl Saienkowrote: > Hello Community, > > > Ironic and Neutron projects have become integrated even closer with > multitenancy implementation in Ironic. > > There are 2 bugs that require separate ML2 driver specifically for Ironic > needs: > > > Booting ironic instance, Neutron port remains in down state [0] > > Ironic needs to synchronize port status change events with Neutron [1] > > > I was told (commented) that keeping code in Neutron tree is not right > approach [2]. Of course I agree that Neutron has powerful and flexible > support of out-of-tree ML2 drivers and such functionality must be a separate > ML2 plugin. > > > So the question to consult with the whole community: > > Do we need a new networking-ironic-* ML2 driver? > > > To fix [0] and [1] we can use existing networking-generic-switch [3] ML2 > driver [4,5]. It was designed specially for Ironic case. It already helps us > to test Ironic multitenancy on the gates. > > > My team would prefer supporting one driver without multiplying entities. So I think I said before I'd prefer this to be a separate repo. Thinking about it more, if we'll be doing [1] in the ML2 driver, this reinforces it for me. Other ML2 drivers (like Arista's) may want to also make callbacks to ironic, so I think it's best for the community if the callback bits are a separate library. Which means to me that it isn't much work to maintain a separate driver to fix [0], since the code for [1] will be a library anyway. That said, I'm still open to the fix for [0] being in networking-generic-switch, if that's what the community prefers. As long as the callback mechanism is a library so other vendors can use it. :) I think I'd be fine with putting some/all of these under ironic governance, so that the community can help maintain it, but we can have a separate conversation about that. // jim > > > [0] https://bugs.launchpad.net/neutron/+bug/1599836 > > [1] https://bugs.launchpad.net/ironic/+bug/1304673 > > [2] https://bugs.launchpad.net/neutron/+bug/1610898 > > [3] https://github.com/openstack/networking-generic-switch > > [4] https://review.openstack.org/357779 > > [5] https://review.openstack.org/357780 > > > > Sincerely, > Vasyl Saienko > > __ > OpenStack Development Mailing List (not for usage questions) > Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev > __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [ironic][neutron] ML2 plugin for Ironic needs
Vasyl Saienkowrote: Hello Community, Ironic and Neutron projects have become integrated even closer with multitenancy implementation in Ironic. There are 2 bugs that require separate ML2 driver specifically for Ironic needs: • Booting ironic instance, Neutron port remains in down state [0] • Ironic needs to synchronize port status change events with Neutron [1] I was told (commented) that keeping code in Neutron tree is not right approach [2]. Of course I agree that Neutron has powerful and flexible support of out-of-tree ML2 drivers and such functionality must be a separate ML2 plugin. So the question to consult with the whole community: Do we need a new networking-ironic-* ML2 driver? To fix [0] and [1] we can use existing networking-generic-switch [3] ML2 driver [4,5]. It was designed specially for Ironic case. It already helps us to test Ironic multitenancy on the gates. My team would prefer supporting one driver without multiplying entities. If I understand the matter correctly, Ironic may want to adopt the existing repository as part of its official deliverables, so probably governance discussion should happen between ironic and networking-generic-switch core teams. (At the moment, the driver is not even part of Big Tent). Alternatively, Ironic may just declare dependency on that other project for its networking multi-tenancy. What I see in the code, the ml2 driver just executes switch specific commands via ssh, so the device driver is just a declaration of those commands, with some substitution involved. Meaning, the switch specific code is really tiny, and should not be a huge burden to maintain for Ironic team if they choose so. To answer your question, I don’t think we need another driver if an existing one may cover your needs, but governance should be considered. That said, seeing you in the core team of the existing driver repo, I believe it’s solvable. Ihar __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
[openstack-dev] [ironic][neutron] ML2 plugin for Ironic needs
Hello Community, Ironic and Neutron projects have become integrated even closer with multitenancy implementation in Ironic. There are 2 bugs that require separate ML2 driver specifically for Ironic needs: - Booting ironic instance, Neutron port remains in down state [0] - Ironic needs to synchronize port status change events with Neutron [1] I was told (commented) that keeping code in Neutron tree is not right approach [2]. Of course I agree that Neutron has powerful and flexible support of out-of-tree ML2 drivers and such functionality must be a separate ML2 plugin. So the question to consult with the whole community: Do we need a new networking-ironic-* ML2 driver? To fix [0] and [1] we can use existing networking-generic-switch [3] ML2 driver [4,5]. It was designed specially for Ironic case. It already helps us to test Ironic multitenancy on the gates. My team would prefer supporting one driver without multiplying entities. [0] https://bugs.launchpad.net/neutron/+bug/1599836 [1] https://bugs.launchpad.net/ironic/+bug/1304673 [2] https://bugs.launchpad.net/neutron/+bug/1610898 [3] https://github.com/openstack/networking-generic-switch [4] https://review.openstack.org/357779 [5] https://review.openstack.org/357780 Sincerely, Vasyl Saienko __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev