16:02:32 <yamahata> #startmeeting neutron_northbound 16:02:32 <odl_meetbot> Meeting started Fri Jan 15 16:02:32 2016 UTC. The chair is yamahata. Information about MeetBot at http://ci.openstack.org/meetbot.html. 16:02:32 <odl_meetbot> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:02:32 <odl_meetbot> The meeting name has been set to 'neutron_northbound' 16:02:39 <yamahata> #topic agenda bashing and roll call 16:02:43 <yamahata> #info yamahata 16:02:59 <yamahata> #link https://wiki.opendaylight.org/view/NeutronNorthbound:Meetings#Agenda_for_Next_Meeting_.281.2F15.2C_2016.29 16:03:20 <yamahata> any topic for agenda? 16:03:32 <yamahata> in addition to the above link 16:04:15 <flaviof> #info flaviof 16:04:20 <yamahata> flaviof: hi 16:04:37 <flaviof> yamahata hello! 16:04:51 <yamahata> #chair flaviof 16:04:51 <odl_meetbot> Current chairs: flaviof yamahata 16:05:03 <yamahata> #topic Announcements 16:05:25 <yamahata> the details for opendaylight design forum has been announced 16:05:30 * yamahata looking for the link 16:06:12 <yamahata> #link https://lists.opendaylight.org/pipermail/tsc/2016-January/004543.html design forum announce ment 16:06:35 <yamahata> It's Feb 29 and Mar 1 at San Mateo CA 16:07:17 <yamahata> Finally the first patch for rewriting of odl mechanism driver was merged! 16:07:29 <yamahata> Kudos to asomya rcurran 16:07:43 <flaviof> yamahata +1 asomya rcurran ! 16:07:53 <yamahata> I really thank you for your patiance. It was a long process. 16:08:03 <rcurran> thanks ... on to L3 + SG ... then lots of testing :-) 16:08:23 <asomya> yamahata flaviof: thanks for the reviews 16:08:24 <yamahata> Yeah many things to do are on our road ahead 16:09:14 <yamahata> call for speaker for ONS 2016 is today 16:09:29 <yamahata> and call for speaker for openstack summit 2016 austin is Feb 1 16:09:35 <yamahata> announce again. 16:09:50 <flaviof> asomya my pleasure. fyi there are some smart folks joining in on that effort too I'll send email to ml to introduce you guys to each other 16:10:07 <yamahata> I suppose we would like to apply for it jointly to promote our effort. 16:10:21 <yamahata> I uploaded the draft as a base. 16:10:21 <flaviof> #action flaviof to send email to ml to introdude Oded, Joshua to nn 16:10:30 <yamahata> #link https://lists.opendaylight.org/pipermail/tsc/2016-January/004543.html 16:11:16 <yamahata> the setting is everyone-can-review. If you send a request, I'll add editable-right. 16:11:44 <yamahata> flaviof: very cool! 16:12:03 <yamahata> anything else to announce? 16:13:18 <yamahata> moving on 16:13:21 <yamahata> #topic action items from last meeting 16:13:42 <yamahata> rcurran: posted patches for ODL plugin 16:14:04 <yamahata> yamahata: posted a patch related to CI 16:14:30 <yamahata> #topic patches reviews 16:14:47 <yamahata> Do we have patches that needs special care or urgent? 16:15:19 <asomya> yamahata: The patch i posted for the SG fix to the V2 driver is kinda urgent 16:15:26 <asomya> #link https://review.openstack.org/#/c/267120/ 16:15:49 <yamahata> #link https://git.opendaylight.org/gerrit/#/c/32109/ 16:15:53 <asomya> a very short fix, should be a quick review 16:16:00 <yamahata> asomya: sure 16:16:09 <yamahata> #action everyone to review https://review.openstack.org/#/c/267120/ 16:16:24 <asomya> yamahata: thanks 16:16:25 <yamahata> https://git.opendaylight.org/gerrit/#/c/32109/ is a patch for documentation of neutron northbound 16:16:44 <yamahata> It needs to be merged before Beryllium release. Please review. 16:17:04 <yamahata> It's urgent because Beryllium release is counting down. 16:17:12 <yamahata> thanks vithar for review it. 16:17:24 <yamahata> any other patches? or bug? 16:18:20 <yamahata> #topic bugs 16:18:37 <yamahata> Right now I'm starting to look at CI timeout. 16:18:59 <yamahata> Actually it has starting before Dec. 16:19:46 <yamahata> most of scenario tests fails due to ssh connection failure. timeout ~= 5min *N => jenkins abort 16:20:08 <flaviof> yamahata do you have a bug or any other info on this issue? I was not aware of it, other than the transaction retry changes 16:20:09 <yamahata> That's what I've found for now. 16:21:09 <yamahata> #action yamahata file a bug for openstack CI jenkins abort 16:21:22 <flaviof> cool, thanks. 16:21:34 <yamahata> For now I have only jenkins logs. I haven't dug into OF table details yet. 16:21:58 <yamahata> Oh there is a filed bug 16:22:10 <yamahata> #link https://bugs.opendaylight.org/show_bug.cgi?id=3913 Neutron Tempest Test Failure: tempest.scenario.test_network_basic_ops 16:23:13 <yamahata> any other bugs to discuss? 16:23:14 <flaviof> yamahata that is the timeout issue or a different one? 16:23:26 <yamahata> flaviof: timeout issue. 16:23:37 <flaviof> i wanted to mention another bug 16:24:07 <flaviof> #link https://bugs.opendaylight.org/show_bug.cgi?id=4924 fixed_ips 16:24:53 <flaviof> Oded is changing the filter to allow fixed_ips to be provided on port update. This was noticed when testing SGs 16:25:27 <flaviof> #link https://review.openstack.org/#/c/266991/ networking-odl side of change for bug 4924 16:25:56 <flaviof> just wanted to mention it here, in case anybody thinks that would be a problem 16:26:43 <flaviof> i think not... 16:27:42 <yamahata> any other bugs? 16:27:45 <flaviof> that is it from me 16:28:19 <yamahata> #topic ML2 ODL driver rewrite 16:28:41 <yamahata> Now we are moving forward to have new patches. 16:28:57 <yamahata> asomya: rcurran anything to share for those patches? 16:29:31 <yamahata> #link https://review.openstack.org/#/c/266583/ 16:29:31 <yamahata> Opendaylight L3 service plugin refactor to handle out of sync issues 16:29:47 <rcurran> L3 is in the review queue. Next up would be SG. Goal is to reach parity with existing L3/SG v1 (original) code 16:29:54 <yamahata> #link https://review.openstack.org/#/c/267120/ Instance creation fails with the new V2 driver 16:31:02 <yamahata> a patch for SG precommit event seems near to merge. So it can be utilized 16:31:14 <yamahata> #link https://review.openstack.org/#/c/252755/ Add precommit_XXX event for security group and rules 16:31:23 <yamahata> rcurran: cool! 16:31:57 <flaviof> rcurran +1 16:32:31 <rcurran> there will be some juggling of code as folks make updates to original driver that may need to be factored in w/ v2 driver 16:34:38 <yamahata> anything else or move on... 16:34:56 <yamahata> #topic OpenStack release support 16:35:20 <yamahata> So far he hasn't responded. So there is no progress. 16:35:50 <yamahata> Regarding to release support, is there any opinion? 16:36:10 <yamahata> If no, we'll just focus on Kilo/Mitaka vs Lithium/Beryllium 16:36:25 <yamahata> Oh Liberty/Mitaka, not Kilo 16:37:16 <flaviof> yamahata ack +1 on Liberty/Mitaka 16:38:18 <yamahata> I hope the patch for openstack project so that openstack CI will be run against both lithium and beryllium 16:38:35 <yamahata> #link https://review.openstack.org/#/c/265437/ networking-odl: enable test with ODL release of beryllium-snapshot 16:39:05 <yamahata> seems there is no specific requirement for now. 16:39:11 <yamahata> #topic open mike 16:39:26 <yamahata> anything else to discuss? 16:39:32 <flaviof> i got a couple... bear with me 16:39:37 <flaviof> :) 16:39:42 <vthapar> I had couple questions too. 16:39:46 <flaviof> Using MD-SAL, and removing iAware API 16:39:57 <flaviof> ovsdb is now 100% md-sal based in regards to neutron northbound. Any other projects still using iAware in Boron? 16:40:07 <flaviof> the push for this is that iAware api is not useable if ODL is used with clustering 16:40:13 <flaviof> #link https://git.opendaylight.org/gerrit/#/c/27291/ removing iAware req 16:40:23 <yamahata> At least Group Based Policy is still using IAware API 16:41:13 <flaviof> yamahata ack. i think that for Boron, we will have green light for removing iaware, but that is definitely something that needs to be coord. 16:41:14 <yamahata> and I don't see any patches for it in GBP patch queue 16:41:46 <yamahata> My plan is to raise it at design forum and/or talk with Keith 16:41:46 <flaviof> second "topic": Clustering — any of you ever deployed O/S HA with ODL clustering (Beryllium)? 16:42:01 <flaviof> yamahata +1 thanks! 16:42:32 <vthapar> flaviof: not yet but in the works. can't give a time as all the folks working on it are done for the day 16:42:34 <yamahata> #action yamahata raise iaware issue at design forum and/or talk with Keith. 16:42:48 <vthapar> flaviof: ^clustering. 16:43:07 <flaviof> vthapar cool. thanks! 16:44:07 <flaviof> we are planning on using HAProxy on northbound of ODL, but I have never really played with that in O/S 16:44:17 <yamahata> I forgot to mention that stable/beryllium branch was cut. 16:44:28 <yamahata> #info stable/beryllium branch was cut. 16:45:01 <flaviof> yamahata ack. with that, I think we can use stable/be for the CI tests. 16:45:44 <flaviof> next question: Rally testing… any updates from Rui? 16:46:14 <yamahata> flaviof: unfortunately no updates except patch respining 16:46:24 <flaviof> yamahata ack, np! 16:46:56 <flaviof> last one from me: in the near future I’ll be less involved with opendaylight project 16:47:07 <flaviof> if any of you who have had some history with the project and would like to become 16:47:09 <flaviof> a comitter, please let me or Isaku know. 16:48:15 <flaviof> vthapar you had questions? 16:48:46 <vthapar> yes. 16:49:12 <vthapar> 1. Regarding the v2 driver... I'd like it done fore BGPVPN as well as L2GW plugin [which isn't started yet]. 16:49:40 <vthapar> BGPVPN is currently sitting in networking-bgpvpn, but I think this would be a good reason and moment to bring it into networking-odl. 16:50:00 <vthapar> the one in n-bgpvpn can be considered v1_0 and this one v2_0. 16:50:11 <vthapar> wanted to know how easy or difficult would be to do this? 16:50:19 <vthapar> and if anyone has any objections to it? 16:50:58 <flaviof> vthapar no objections on my part but, isn't networking-bgpvpn used outside ODL? 16:51:02 <yamahata> IIRC, dirver for bgpvpn was commited to n-bgpvpn for release management ant api stability. 16:51:22 <yamahata> Now it's okay with them. 16:51:35 <vthapar> flaviof: yep, but am talking about bgpvpn odl driver specifically. 16:51:39 <yamahata> For me, it's okay to include driver for bgpvpn. 16:51:41 <vthapar> yamahata: yep 16:51:52 <yamahata> Is it okay with them? 16:51:57 <vthapar> with liberty release at least they have a stable enough API. 16:52:29 <vthapar> I expect them to be. I am not sure how easy o difficult it would be to have the out-of-sync fixes if it is not in n-odl. 16:52:37 <flaviof> vthapar ack. make sense to ne in networking-odl then. 16:52:37 <vthapar> ^easy or difficult. 16:53:14 <vthapar> for Mitaka at least I will continue to support driver in n-bgpvpn too. 16:53:17 <asomya> vthapar: will have to scope the bgpvpn driver in it's current for to answer the easy/difficult question 16:53:42 <vthapar> asomya: fair enough. 16:53:51 <yamahata> which timeframe do you have in mind? Mitaka or N? 16:54:14 <vthapar> I would prefer Mitaka and willing to put in effort to get it done 16:54:46 <vthapar> L2GW may not be ready by Mitaka coz still needs some work on plugin side. 16:54:49 <flaviof> vthapar a potential trouble there is that v2 may not be ready for Mitaka. 16:55:25 <vthapar> flaviof, np. that is why I would continue supporting existing one in n-bgpvpn, this one can be considered experiantial. 16:55:52 <vthapar> if we can't make it, we'll go in N. but would prefer to get started sooner than later. 16:56:09 <yamahata> So you plan to continue to have old(v1) in n-bgpvon and new(v2) in n-odl? 16:56:29 <vthapar> yamahata: at least till Mitaka release. 16:56:31 <flaviof> vthapar k. as long we do not overload v2 before we get parity with v1 16:57:09 <vthapar> flaviof: agree on that. this comes after L3 and SG and when all say ready to bring bgpvpn in. 16:57:48 <flaviof> cool beans 16:57:56 <vthapar> my second question was regarding how we're handling updates in NN today. 16:58:22 <vthapar> basically the updateDelta method in all Neutron*Northbound methods 16:58:59 <vthapar> I believe proper fix for it requires mech driver to send the complete object for update rather than just delta, right? 16:59:24 <vthapar> then we can just dump the object into MDSAL without having to add back some crucial fields. 16:59:39 <vthapar> right? 17:00:11 <flaviof> vthapar i thought that the updateMD methods did that for us, but this may be a case by case question 17:00:59 <vthapar> flaviof, I still some comments to that in code. at least in Network, Subnet and Port they're there. 17:01:57 <yamahata> vthapar: What's concern? do you have specific issues with bgpvpn or l2gw? 17:02:32 <yamahata> or general code quality/consistency? 17:02:37 <vthapar> yamahata: not really, for those two we don't use any updateDelta at all and in driver code I already send the complete object. 17:03:00 <vthapar> yep. I think it might be required for making NN fully transparent. 17:03:07 <vthapar> for reference: https://github.com/opendaylight/neutron/blob/master/northbound-api/src/main/java/org/opendaylight/neutron/northbound/api/NeutronPortsNorthbound.java#L251 17:03:51 <vthapar> the comment in there suggests that it is a patch code that should be addressed. 17:04:07 <yamahata> In that direction, the direction is to switch to RESTConf from custom neutron northbound. 17:04:32 <vthapar> I haven't looked deeper, but suspect it could also lead to data loss. 17:04:45 <yamahata> NN will maintain only yang mode with full transparency. 17:05:00 <vthapar> yamahata: so the driver will directly use restconf? 17:05:05 <yamahata> That would be a bit long term. 17:05:12 <yamahata> vthapar: yes. 17:05:40 <yamahata> It exposes yang model version directly. there maybe compatibility issue. 17:06:14 <vthapar> okay, good enough. 17:06:23 <yamahata> Does it make sense? At this point, just an idea. need to reach consensus before decision. 17:06:33 <flaviof> yes. the compatibility is the more dangerous issue by going straight restconf. but the automation would eliminate a lot of manual code. 17:07:18 <vthapar> I am in favor of this. I think you'd asked me to work on bgpvpn driver so it can sue restconf directly. plan to do that, haven't had time yet. 17:07:55 <vthapar> if out-of-sync is not ready by mitaka for bgpvpn driver, I'll see if I can write it to use restconf. 17:07:55 <flaviof> yamahata i think it is a valid idea. it would be awesome to have yangtools spitting out some python that would be used by networking-odl 17:08:14 <vthapar> flaviof: +1 17:08:28 <yamahata> we should understand the implication for compatibility. given openstack rolling upgrade is requirement. 17:08:37 <flaviof> ack 17:09:32 * yamahata 10minutes over... 17:10:03 <vthapar> am done. thanks for your patience :) 17:10:06 <yamahata> vthapar: more questions? 17:10:08 <yamahata> oh. 17:10:16 <yamahata> any other question or let's close the meeting 17:10:47 <yamahata> okay thanks everyone 17:10:49 <yamahata> #topic cookies 17:10:55 <yamahata> #endmeeting