15:02:44 <yamahata> #startmeeting neutron_northbound
15:02:44 <odl_meetbot> Meeting started Wed Sep 14 15:02:44 2016 UTC.  The chair is yamahata. Information about MeetBot at http://ci.openstack.org/meetbot.html.
15:02:44 <odl_meetbot> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:02:44 <odl_meetbot> The meeting name has been set to 'neutron_northbound'
15:02:51 <yamahata> #chair mkolesni
15:02:51 <odl_meetbot> Current chairs: mkolesni yamahata
15:02:58 <yamahata> #topic agenda bashing and roll call
15:03:02 <mkolesni> #info mkolesni
15:03:04 <yamahata> #info yamahata
15:03:16 <yamahata> #link https://wiki.opendaylight.org/view/NeutronNorthbound:Meetings meeting agenda
15:03:30 <yamahata> we have usual patch review
15:03:35 <yamahata> no action items
15:03:51 <yamahata> any other items discuss?
15:04:01 <mkolesni> none from me
15:04:03 <yamahata> mkolesni: regarding to odl summit presentation, we can talk privately.
15:04:17 <mkolesni> sure
15:04:22 <yamahata> with bluejeans meeting
15:05:08 <mkolesni> ok
15:05:43 <yamahata> #topic Announcements
15:05:47 <yamahata> No new announcement
15:05:54 <yamahata> odl summit is approaching
15:06:21 <yamahata> Developer design forum will be held at the same time
15:07:02 <yamahata> #link https://wiki.opendaylight.org/view/Events:Carbon_Dev_Forum
15:07:30 <yamahata> I'm preparing a slide for neutorn/northbound planning
15:07:45 <yamahata> #link https://docs.google.com/presentation/d/1606VOorn9pk90ICw99IeT9tIwB_itoGS2jIErp7pGwg/edit odl neutron northbound carbon planning
15:08:06 <yamahata> the wishlist can be edited at trello board
15:08:23 <yamahata> #link https://trello.com/b/LhIIQ8Z0/odl-neutronnorthbound neutron northbound trello board
15:09:00 <yamahata> openstack summit will be held at the end of october. Many ODL'ers would also attend it. I'll also attend it.
15:09:06 <yamahata> any other announcement?
15:10:17 <yamahata> seems nothing else. move on
15:10:35 <yamahata> #topic migrating to new features
15:10:49 <yamahata> Now we're working on stabilizing/filling a gap for newton release.
15:11:02 <yamahata> mainly v2 driver and pseudo-agent port binding, and vpnservice(new netvirt)
15:11:16 <yamahata> Any other features to be included for newton release?
15:11:49 <barak> Does anyone know what is the V2 status with net.scenario tests?
15:12:25 <yamahata> scenario tests? some of scenario tests are known to fail. It's mostly same to v1 driver case.
15:12:58 <yamahata> Those failures should be fixed, though.
15:14:04 <yamahata> For example http://logs.openstack.org/54/359954/9/check/gate-tempest-dsvm-networking-odl-boron-snapshot/36c6d2c/console.html
15:14:13 <yamahata> This is with Boron RC3.5
15:15:18 <yamahata> I'd like to address test_network_v6 and test_network_advanced_server_ops for newton or early ocata
15:15:54 <yamahata> barak: do you have concern about specific tests?
15:16:13 <yamahata> http://logs.openstack.org/54/359954/9/check/gate-tempest-dsvm-networking-odl-boron-snapshot/36c6d2c/console.html#_2016-09-13_00_12_00_377198
15:16:30 <barak> Well, in general I have seen all scenario tests failing on most (if not all ) of the combinations we run
15:16:52 <barak> with ODL of course...
15:18:09 <yamahata> I see. Do you have any plan to look into it?
15:19:06 <yamahata> For now, my hands are full of v2 driver.
15:19:24 <yamahata> Or do you have time constraint?
15:19:47 <yamahata> We can discuss in detail at ODL summit on how to address scenario tests
15:19:54 <barak> May be, depends on other things. Yes I have.
15:21:33 <barak> I won't be there, but there will be some other representatives from our team which I believe you are going to meet :-)
15:23:09 <yamahata> gotcha. Do you have any other issues/concerns?
15:23:13 <barak> Is pseudo agent db going to be the only V2 supported option?
15:24:18 <yamahata> it will supports both v1 and v2.
15:24:46 <yamahata> I'm afraid that v1 case wouldn't be tested well.
15:25:03 <yamahata> V2 driver case would be first.
15:25:20 <yamahata> Do you have requirement for v1 + pseudo agent?
15:26:16 <barak> Just looking into the best stable alternatives, having in mind issues reported about the V1 (race conditions etc)
15:26:43 <yamahata> I understand the situation.
15:27:35 <yamahata> any other issues or move onto reviewsbugs?
15:28:45 <yamahata> okay move on
15:28:52 <yamahata> #topic patches/bugs
15:28:57 <yamahata> Now several patches are under review
15:29:14 <yamahata> #link https://review.openstack.org/#/c/358319/ Journal recovery for basic scenarios
15:29:40 <yamahata> mkolesni: can you please update commit message? then we can merge it, I suppose.
15:29:49 <yamahata> or do you plan to respin it?
15:30:02 <mkolesni> yamahata: sure ill update the commit message
15:30:22 <yamahata> #link https://review.openstack.org/#/c/359501/
15:30:22 <yamahata> Cloud admin script auto-config hostconfig defaults
15:30:35 <yamahata> Now Federico will take care of pseudo agent port binding stuff.
15:30:57 <yamahata> let's review it and make progress on it.
15:31:21 <yamahata> Some tasks need to be done for pseudo agent port binding.
15:31:48 <yamahata> Okay, now my patches.
15:31:59 <yamahata> #link https://review.openstack.org/#/q/topic:db-datetime-second-precision
15:32:04 <mkolesni> i did a review there
15:32:09 <yamahata> mkolesni: thanks.
15:32:28 <yamahata> Basically which patches are okay(except resping/review)?
15:32:35 <yamahata> I see your -2
15:32:55 <yamahata> can we respin patches with -1?
15:33:06 <mkolesni> well yeah that one is contrary to the basic idea of the journal
15:33:19 <mkolesni> so it doesnt make sense to have
15:33:30 <yamahata> The direct sync one?
15:33:37 <mkolesni> yeah
15:33:53 <yamahata> I think it isn't contrary.
15:33:56 <mkolesni> its basically turns v2 driver into v1
15:34:01 <yamahata> No.
15:34:06 <mkolesni> so just use v1 driver :)
15:34:09 <yamahata> It's just for throttling.
15:34:28 <yamahata> The main goal of v2 driver is to maintain the order of operation.
15:34:35 <mkolesni> i think we lose a lot of the advantage there
15:34:55 <yamahata> It doesn't matter where to trigger sending rest request to ODL.
15:35:06 <mkolesni> if we need to tackle perf issues theres probably a better way
15:35:11 <yamahata> The problem I'm seeing is huge back log in journal.
15:35:18 <yamahata> I see.
15:35:29 <mkolesni> its a problem if it takes a lot of time to clean up
15:35:44 <mkolesni> but i think your other changes are adressing some of it
15:35:59 <yamahata> Okay for now, I'll respin other patches.
15:36:17 <yamahata> and let's continue discussion on huge backlog issue.
15:36:17 <mkolesni> we can discuss the plan for next cycle on the summit and talk about moving the dependency analysis
15:36:24 <yamahata> Sure.
15:36:36 <mkolesni> because i think that would be a major gain
15:36:48 <mkolesni> in terms of performance and reliability
15:37:03 <mkolesni> but the other patches do make sense given the current situation
15:37:40 <mkolesni> except the one where you change the oprdering to be by create_date which i think has many reprecautions that you didnt take into account
15:38:09 <yamahata> mkolesni: sure. That's reason why we need patch review.
15:38:23 <mkolesni> i know you think its better that way but it introduces a big change in behavior
15:38:51 <yamahata> I agree that it's big change.
15:39:04 <mkolesni> so im not sure if we should go ahead with that one
15:39:20 <mkolesni> the one about the autoincrement does make sense if you want better percision
15:39:33 <mkolesni> and so some of the fixes to dependency validation
15:39:53 <mkolesni> so thats basically my 2 cents :)
15:39:59 <yamahata> I have those patches when trying to make tempest test pass decently.
15:40:06 <yamahata> That's the motivation.
15:40:09 <mkolesni> yeah i know
15:40:24 <mkolesni> but im not sure what problem the created at is trying to solve
15:40:29 <yamahata> Even tests outputs success, I'm seeing issues behind.
15:41:03 <yamahata> those issues should be solved somehow. Otherwise we'll get many complains.
15:41:17 <mkolesni> well it is a tech preview of sorts
15:41:44 <mkolesni> i dont think it will be stable yet in Newton, we dont have enough time and man power to achieve that
15:42:14 <mkolesni> but i do think its ready for consumption, and then well see what bugs are people hitting
15:42:26 <yamahata> Right. Hopefully at ODL summit, we can have common idea of what will be delivered etc.
15:42:32 <mkolesni> sure
15:43:01 <mkolesni> btw if you want greater performance perhaps it's better to see if we can somehow run multiple journals on same host
15:43:11 <mkolesni> ie tie it with api/rpc workers
15:43:20 <mkolesni> so it utilizes multi threading
15:43:50 <yamahata> Yeah. we should do benchmark and figure out where's bottleneck first.
15:44:11 <mkolesni> i think youll be amazed how much it will improve :)
15:44:57 <yamahata> :-)
15:45:41 <yamahata> Okay, any other patches/bugs to discuss?
15:46:09 <mkolesni> none from me
15:46:15 <yamahata> move on
15:46:17 <yamahata> #topic open mike
15:46:22 <yamahata> any thing else to discuss?
15:47:02 <mkolesni> nothing here
15:47:10 <yamahata> Today we'll have bluejeans meeting too for ODL summit logistics.
15:47:17 <yamahata> #link https://bluejeans.com/223556372/
15:47:37 <yamahata> thanks. see you at bluejeans
15:47:43 <yamahata> #topic cookies
15:47:47 <yamahata> #endmeeting