18:00:27 <rtheis> #startmeeting networking_ovn 18:00:29 <openstack> Meeting started Thu Sep 22 18:00:27 2016 UTC and is due to finish in 60 minutes. The chair is rtheis. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:30 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 18:00:33 <openstack> The meeting name has been set to 'networking_ovn' 18:00:40 * regXboi wanders in 18:00:56 <rtheis> o/ 18:01:02 <rtheis> #topic Announcements 18:01:18 <rtheis> ovs 2.6.0 should be released soon 18:01:35 <rtheis> Once released and Newton is closed, we should include in Ocata 18:01:38 <rtheis> #link https://bugs.launchpad.net/networking-ovn/+bug/1622647 18:01:40 <openstack> Launchpad bug 1622647 in networking-ovn "Move OpenStack requirements to ovs 2.6.0" [Undecided,Confirmed] - Assigned to Richard Theis (rtheis) 18:02:07 <rtheis> That's all I have for announcements today 18:02:15 <rtheis> anyone else? 18:02:45 <rtheis> #topic: Newton RC2 18:03:16 <rtheis> We'll need rc2 after several DHCP issues were encountered over the last week 18:03:33 <rtheis> Here is the WIP for RC2 release 18:03:36 <rtheis> #link https://review.openstack.org/#/c/374820/ 18:04:05 <rtheis> Should be able to cut RC2 once the following merge on stable/newton 18:04:08 <rtheis> #link https://review.openstack.org/#/c/374852/ 18:04:17 <rtheis> #link https://review.openstack.org/#/c/374454/ 18:05:22 <rtheis> neutron DHCP force_metadata end enable_isolated_metadata features seem to be broken in the gate 18:06:13 <rtheis> There was one more fix on the rc2 list 18:06:15 <rtheis> #link https://review.openstack.org/#/c/371191/ 18:06:26 <chandrav> rtheis: Can we get the NAT stuff in RC2. 18:06:31 <rtheis> Ramu are you able to continue with this ^^^ 18:06:36 <rtheis> chandrav: no 18:06:41 <russellb_> hi, i'm technically out today, sorry i'm late (and missed ovn meeting) 18:07:01 <rtheis> russellb_ and I discussed a couple days ago 18:07:03 <Sam-I-Am> oops, late. 18:07:11 <regXboi> russellb_ - go away and vacate, man! 18:07:14 <rtheis> I don't think stable maintainers will allow it 18:07:17 <russellb_> i'd really like to get the NAT stuff in 18:07:22 <chandrav> got it 18:07:31 <rtheis> russellb_: rc2? 18:07:32 <russellb_> but it's probably against stable policy 18:07:36 <russellb_> yes 18:08:09 <Sam-I-Am> feature backports are no 18:08:09 <rtheis> I have not asked, but could 18:08:17 * mestery lurks 18:08:23 <ramur> @rtheis: I am updating: https://review.openstack.org/#/c/371191/ based on comments + unit test 18:08:27 <Sam-I-Am> bug fixes and sec patches go back 1 rel, sec patches go back 2 iirc 18:08:36 <russellb_> yep. 18:08:58 <Sam-I-Am> you could argue that lack of nat is a bug. i consider it a feature :) 18:09:10 <Sam-I-Am> that lack of, that is :) 18:09:29 <rtheis> are there any other items that we should target for rc2? 18:10:34 <russellb_> we could always propose it and see what happens ... 18:10:44 <russellb_> since this is first release, no regression risk 18:10:51 <Sam-I-Am> this is also true 18:11:13 <rtheis> sure, I can ask 18:11:30 <russellb_> ok, well +1 from me fwiw 18:12:09 <rtheis> #topic Reviews 18:12:32 <rtheis> We have several big features in review queue 18:13:04 <rtheis> (1) NAT support, (2) native DHCP v6 and (3) native metadata 18:13:16 <rtheis> please take a look if you have time 18:14:04 <rtheis> Also starting to setup a multinode job: https://review.openstack.org/#/c/373528/ 18:14:13 <rtheis> experimental to start 18:14:19 <russellb_> is the multi-node job using native services? 18:14:24 <rtheis> Yes 18:14:32 <russellb_> k :) 18:14:33 <Sam-I-Am> rtheis: cool 18:14:45 <russellb_> i'd like to remove out l3 agent job once we merge NAT support 18:14:48 <russellb_> anyone opposed to that? 18:14:57 <russellb_> and related, remote support for that entirely 18:15:04 <russellb_> remove* 18:15:27 <Sam-I-Am> support for the conventional l3 agent? 18:15:45 <russellb_> correct 18:15:46 <Sam-I-Am> if there's no realistic reason anyone would run it, go for it. 18:15:50 <rtheis> I'd like to see the same for dhcp and metadata 18:16:05 <Sam-I-Am> we had them there to work around missing native functionality 18:16:06 <russellb_> Sam-I-Am: that's my thinking 18:16:19 <russellb_> dhcp agent still provides some minor functionality that native dhcp does not 18:16:26 <regXboi> dns? 18:16:28 <russellb_> metadata support is 1 (though we'll have that resolved soon) 18:16:30 <russellb_> dns, yes 18:16:40 <Sam-I-Am> what about dns? resolution forwarding? 18:16:47 <russellb_> yes, internal dns support 18:16:56 <russellb_> resolving internal names -- nova instance name, usually 18:17:11 <Sam-I-Am> but you can hand it a dns ip address, right? 18:17:19 <russellb_> yes 18:17:21 <rtheis> russellb_ the only concern with removing the L3 agent job is that it also cover dhcp 18:17:28 <russellb_> rtheis: hm, fair point 18:17:36 <Sam-I-Am> most installs i know about always specify dns as something besides the dnsmasq ip address 18:17:39 <russellb_> we could just change it to be a variant of native services that uses dhcp agent instead 18:17:44 <rtheis> ack 18:17:50 <rtheis> makes sense 18:18:17 <russellb_> another CI related thing that i think babu will work on when it's time is updating our grenade migration job to use native services 18:18:20 <Sam-I-Am> using dnsmasq is sort of there as a last resort for resolution. however, using dns integration might be a different story. 18:18:41 <russellb_> it uses the l3 agent because the tests expect floating ips 18:19:46 <russellb_> rtheis: sorry, topic was reviews, i took us on a tangent.. 18:19:52 <rtheis> np 18:20:08 <rtheis> so just one grenade job for now 18:20:19 <russellb_> yes, but maybe we need 2 18:20:26 <rtheis> or do we also now need newton to ocata 18:20:31 <russellb_> the current one does a migration from ml2+ovs to ovn as a part of the upgrade 18:20:46 <russellb_> we probably need a new one that does newton->master with ovn before and after 18:20:49 <Sam-I-Am> oh, i didnt know we were doing that 18:20:55 <russellb_> Sam-I-Am: yeah man 18:21:06 <russellb_> it's got some hacks, but making it work was step 1 :) 18:21:09 <russellb_> a huge step 1 18:21:16 <rtheis> yes 18:21:18 <Sam-I-Am> guess we need to document it 18:21:44 <rtheis> I think we may have a bug opened for the newton->master grenade job 18:21:44 <russellb_> probably! 18:21:50 <rtheis> I'll check 18:21:57 <russellb_> hm sounds familiar 18:22:07 <russellb_> i remember filing a bug for "we should add a grenade job someday when it makes sense" 18:22:08 <rtheis> I'll also open a bug to track deprecating L3 agent 18:22:18 <rtheis> yeah 18:22:20 <rtheis> that one 18:23:33 <rtheis> anything else on this or other reviews to discuss? 18:24:11 <rtheis> #topic Conferences 18:24:28 <rtheis> OpenStack summit in a month 18:24:34 <rtheis> etherpad started 18:24:36 <rtheis> #link https://etherpad.openstack.org/p/networking-ovn-ocata-summit 18:24:40 <rtheis> feel free to add topics 18:24:57 <rtheis> my room request was denied for now 18:25:12 <rtheis> hopefully something will open or we just do an ad-hoc meetup 18:25:35 <russellb_> i've been wondering if it'd make sense to propose ovn as a neutron room topic 18:25:44 <russellb_> not sure if it makes sense to discuss it with the broader neutron dev team 18:26:13 <russellb_> review status, talk about future 18:26:26 <russellb_> was thinking devs of ovs support might be interested in talking about it 18:26:33 <russellb_> but i dunno 18:27:00 <rtheis> who to ask on that? 18:27:17 <russellb_> armax 18:27:47 <rtheis> sure, I can ask him 18:28:39 <russellb_> i don't feel strongly about, just happy to do it if there's enough people interested 18:29:24 <rtheis> sounds good 18:30:03 <russellb_> speaking of conferences, OVSCon is a couple weeks after the barcelona summit 18:30:15 <russellb_> call for presentation proposals is open until end of this week i believe 18:30:18 <rtheis> just going to mention that ;) 18:30:30 <russellb_> i'll be there 18:30:40 <russellb_> haven't proposed any talks though 18:31:36 <Sam-I-Am> i wont be there 18:31:44 <rtheis> me either 18:32:19 <rtheis> anything else on the conferences? 18:32:32 <rtheis> #topic Open Discussion 18:32:56 <rtheis> We already covered my one topic on deprecating L3 agent 18:33:34 <rtheis> I do want to note with Ocata opened, we should be adding release notes for new features, deprecations, upgrade notes, etc. 18:33:51 <russellb_> rtheis: ah, good point 18:34:16 <rtheis> It's nice to have them with the code review but they are easy enough to add later 18:34:28 <russellb_> we should probably do out best to require them with the code ... 18:34:34 <rtheis> agreed 18:34:37 <rtheis> #link http://docs.openstack.org/project-team-guide/release-management.html#managing-release-notes 18:35:09 <rtheis> and as alway docs would be nics 18:35:40 <rtheis> Sam-I-Am: should we be doing additional items for docs? 18:36:04 <rtheis> I feel we've been falling behind on docs 18:36:39 <Sam-I-Am> rtheis: yeah 18:36:45 <Sam-I-Am> docs have more or less stalled out 18:37:14 <Sam-I-Am> if you're writing a feature into networking-ovn, you must include docs with it 18:37:26 <Sam-I-Am> its a little difficult because of the disconnect between ovs/ovn and networking-ovn 18:37:38 <rtheis> docs can be within the project or ...? 18:37:46 <Sam-I-Am> yeah, they're all within networking-ovn 18:37:52 <rtheis> good 18:38:16 <Sam-I-Am> granted native implementations get a little weird 18:38:45 <Sam-I-Am> someone enabling the feature in networking-ovn for openstack should not be reponsible for writing all the docs for the underlying ovn implementation 18:39:05 <Sam-I-Am> primarily write about how to configure it and how it pertains to openstack 18:39:20 <Sam-I-Am> hopefully people are writing docs in ovn for the operation of things 18:39:45 <russellb_> i've got to run 18:39:55 <rtheis> thanks russellb_ 18:40:04 <russellb_> talk to you all later 18:40:07 <rtheis> I don't have anything else for today 18:40:10 <russellb_> back monday 18:40:26 <rtheis> other topics? 18:40:56 <Sam-I-Am> nothing here 18:41:01 <Sam-I-Am> but reminding people of docs is a thing :) 18:41:15 <rtheis> definitely 18:41:19 <Sam-I-Am> me and ryan cant be the only ones writing them 18:41:35 <Sam-I-Am> and occasionally you, i think 18:42:24 <rtheis> okay, let's wrap up for today 18:42:30 <rtheis> thanks 18:42:31 <rtheis> #endmeeting