21:02:19 <mestery> #startmeeting networking
21:02:19 <openstack> Meeting started Mon Apr 21 21:02:19 2014 UTC and is due to finish in 60 minutes.  The chair is mestery. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:02:20 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:02:23 <openstack> The meeting name has been set to 'networking'
21:02:34 <mestery> #link https://wiki.openstack.org/wiki/Network/Meetings Agenda
21:02:54 <mestery> #topic Announcements
21:03:06 <mestery> I wanted to reiterate the new neutron-specs repository for Juno BPs.
21:03:24 <mestery> If you have a feature which missed Icehouse, you need to file a BP in neutron-specs and get it approved berfore your code will be allowed to merge.
21:03:46 <mestery> #link https://wiki.openstack.org/wiki/Blueprints#Neutron Neutron BP process
21:04:02 <mestery> I encourage all cores (and non-cores) to please add neutron-specs to your review queue and spend some time reviewing BPs.
21:04:12 <mestery> We already have 10+ BPs filed, and some good comments on them from the community.
21:04:25 <mestery> #link https://review.openstack.org/#/q/status:open+project:openstack/neutron-specs,n,z neutron-specs open reviews
21:04:31 <mestery> Any questions on the new BP process?
21:04:39 <amotoki_> I would like to know the relationship between BP approver and neutron-specs approver.
21:04:53 <mestery> amotoki_: So far, they are both neutron cores.
21:05:25 <mestery> amotoki_: wait, did I misinterprate that?
21:05:36 <mestery> amotoki_: Do you mean the BP approver in launchpad and the neutron-specs approver?
21:05:43 <sc68cal> I've seen a couple new bp's registered on the ipv6 side - I'll make sure to tell people tomorrow to get on neutron-specs
21:05:49 <enikanorov_> how detailed bp description should be?
21:05:58 <mestery> sc68cal: Great!
21:05:59 <amotoki_> mestery: i mean BP approver is who is set to approver on launchpad approver.
21:06:17 <marios> mestery: does this include code for which a review is already in progress (e.g. something i am on v 15 for )
21:06:21 <mestery> enikanorov_: It depends on the BP, but enough detail for reviewers to understand the design and be comfortable with the direction./
21:06:42 <sc68cal> don't forget: extra credit for using nwdiag
21:06:45 <mestery> amotoki_: That's a good question. I assume someone who +2s the review in neutron-specs could become launchpad BP approver.
21:07:01 <sc68cal> ;)
21:07:05 <amotoki_> mestery: yeah. it is a good idea.
21:07:06 <mestery> marios: If it's not a bug, it needs a spec in neutron-specs. That's been the process so far.
21:07:26 <mestery> #info A core who +2s a BP in neutron-specs can add themselves as LP approvers for the BP as well.
21:07:43 <mestery> OK, one other announcement
21:07:46 <marios> mestery: well, it *is* a bug, but is also a feature addition (api change... https://review.openstack.org/#/c/62042/)
21:07:46 <amotoki_> For neutron-specs, anyone can approve a blueprint in neutron-specs once we have a rough consensus on it. it is important we don't rush to approve.
21:08:09 <mestery> marios: that looks like a bug fix, no spec required. :)
21:08:20 <mestery> amotoki_: Agreed.
21:08:32 <marios> mestery: thanks for the clarification
21:08:36 <mestery> marios: Sure!
21:08:56 <mestery> OK, lets move on. We can continue the new BP discussion on the ML and also in Atlanta in person for those who will attend the summit.
21:09:01 <mestery> I had one more announcement.
21:09:09 <mestery> markmcclain and I have been talking about the mid-cycle meetup
21:09:17 <mestery> We think it would good to focus this first on nova-network parity
21:09:40 <mestery> So that will be the focus for the 3-day get together.
21:09:48 <mestery> We'll hopefully settle on date and location next week.
21:10:04 <nati_ueno> Sorry late
21:10:09 <mestery> nati_ueno: np, welcome!
21:10:18 * markmcclain sneaks in too
21:10:26 <marios> mestery: is a european location out of the question at this point? (I guess has anyone offered to host in EU yet?)
21:10:28 <mestery> So, I'll continue the discussion on the mid-cycle meetup on the ML.
21:10:51 <mestery> marios: We've pretty much settled on a US location, but I think considering EU for a future one is a good idea!
21:11:14 <marios> sure thanks (sorry silly question should just ask on the mailing list)
21:11:19 <mestery> marios: No worries.
21:11:28 <mestery> #topic Bugs
21:12:12 <mestery> Looks like there are issues with the stable queue at this point: http://lists.openstack.org/pipermail/openstack-dev/2014-April/033225.html
21:12:30 <mestery> It's unclear to me if this is mostly an oslo issue or also a neutron issue.
21:13:04 <markmcclain> I was planning on digging into these tomorrow unless someone wants to start earlier
21:13:37 <mestery> markmcclain: I will also dig into this, but won't have time until tomorrow either.
21:13:50 <mestery> So unless someone beats markmcclain and I to this, we'll hopefully get an understanding of this better tomorrow.
21:13:59 <mestery> Are there other bugs the team shoudl be aware of now?
21:14:17 <enikanorov_> there is one
21:14:22 <enikanorov_> just for your info
21:14:25 <mestery> I see this one: https://bugs.launchpad.net/neutron/+bug/1307344
21:14:26 <uvirtbot> Launchpad bug 1307344 in neutron "neutron migrations are not idempotent or invariant to enabled services" [Critical,In progress]
21:14:38 <enikanorov_> tempest scenario test of lbaas is failing pretty often
21:15:03 <mestery> enikanorov_: Any progress on that one to track a root cause?
21:15:03 <enikanorov_> not exactly in neutron, but i just would like you to know that we're taking care of it
21:15:11 <mestery> enikanorov_: Thanks!
21:15:30 <mestery> #topic Docs
21:15:32 <enikanorov_> yes, at least we know what has triggered it, but to sure what to do with it... probably revertthe patch
21:15:38 <mestery> #undo
21:15:39 <openstack> Removing item from minutes: <ircmeeting.items.Topic object at 0x38f6410>
21:15:50 <marun> enikanorov_: feel free to work with me to create a functional test to reproduce ;)
21:15:53 <mestery> enikanorov_: OK, cool. Thanks for working on that!
21:16:15 <enikanorov_> marun: thanks, will think about it
21:16:22 <mestery> Any other bugs the team should discuss?
21:17:05 <mestery> #topic Docs
21:17:09 <mestery> emagana: Howdy!
21:17:10 <emagana> Hi Folks!
21:17:14 <mestery> emagana: Any updates for us this week?
21:17:28 <emagana> Good progress by the team
21:17:41 <emagana> a lot of bugs have been closed!!!
21:17:46 <mestery> Great!
21:18:17 <emagana> I have updated the wiki with the remaining work
21:18:33 <emagana> also (as always) chasing people to take ownership of some bugs  :-)
21:18:37 <mestery> Thanks emagana and team for the work on the docs issues!
21:18:44 <amotoki_> Re: https://bugs.launchpad.net/openstack-manuals/+bug/1293897
21:18:46 <uvirtbot> Launchpad bug 1293897 in openstack-manuals "Create new IPv6 attributes for Subnets" [Medium,Confirmed]
21:18:49 <emagana> There is one that I want to discuss
21:18:49 <amotoki_> iwe can defer it to Juno.
21:18:58 <marios> question: when a bug has !DocImpact (API change/addition for example) - is this picked up by emagana and team?
21:19:12 <emagana> marios: Yes!
21:19:18 <marios> emagana: perfect, thanks
21:19:45 <marios> emagana: is this picked up by some existing process or should folks report these/ping you/someone else/mailing list etc
21:19:50 <mestery> amotoki_: Yes, that one can be deferred to Juno for sure since we hid those attributes in Icehouse.
21:20:21 <emagana> marios basically it creates a new bug in launchpad tagged as Neutron
21:20:59 <emagana> This one: https://bugs.launchpad.net/openstack-manuals/+bug/1296997
21:21:00 <uvirtbot> Launchpad bug 1296997 in openstack-manuals "Add script to migrate ovs or lb db to ml2 db" [High,Triaged]
21:21:22 <emagana> steven weston owns this one but this is very important
21:21:43 <mestery> emagana: Agreed. Thank you sweston for taking ownership of that one!
21:21:46 <sweston> Yes, that one is mine.  It is documented well in the code, I'm not sure what else I can add to it.
21:22:08 <emagana> sweston: We need to move the documentation from the code to the Docs
21:22:22 <amotoki_> sweston: even if it is same as the description in the code, it is worth documented.
21:22:28 <emagana> sweston: we can work offline on this
21:22:35 <mestery> emagana: Regarding that one, someone has already hit an issue with OVS -> ML2 migration: http://lists.openstack.org/pipermail/openstack/2014-April/006767.html
21:22:35 <sweston> emagana: gotcha, yes let's do that.
21:22:44 <emagana> sweston: great!
21:22:50 <emagana> mestery: nothing else commander
21:22:54 <mestery> emagana: I just saw that on list, not sure if it's due to lack of documentation etc.
21:22:56 <mestery> emagana: OK, thanks!
21:23:16 <mestery> #topic Juno Design Summit
21:23:32 <mestery> Thank you to all who submitted session proposals for Atlanta!
21:23:38 <enikanorov_> what is oversubscription rate? :)
21:23:40 <mestery> We are oversubscribed about 2.5:1 at this point.
21:23:44 <sc68cal> sorry just missed mention of 1293897
21:23:53 <mestery> So, not everyone will get time.
21:24:14 <mestery> I wanted to make sure that people know that even if your proposal doesn't get floor time in Atlanta, that doesn't mean it can't get into Juno.
21:24:17 <sc68cal> I can assign myself it for the juno release
21:24:22 <mestery> Thanks sc68cal!
21:24:34 <mestery> I'm working with the core team to come up with a schedule now, and it's looking pretty good.
21:24:50 <mestery> Neutron sessions will be all-day Wed, Thur morning until lunch, and Fri 10:30-5:30.
21:25:18 <mestery> I may be reaching out to people to collapse some sessions together, I've done this for a few things already.
21:25:32 <mestery> So please bear with me on that as I finish settling the schedule this week.
21:25:44 <mestery> Any questions on the Juno Design Summit?
21:26:04 <pcm_> Would be good to know Fri schedule, in case need to adj flight times for leaving.
21:26:32 <mestery> pcm_: Yes. We will have sessions until 4:30, I'm planning the last session as a wrapup session similar to Hong Kong.
21:26:46 <mestery> We can use that one for additional discussions if needed, or end a bit early if people are burned out by that time. :)
21:26:58 <pcm_> mestery: ok cool, then my 7pm flight will be ok.
21:27:09 <mestery> pcm_: I think that will be perfect, yes. :)
21:27:28 <amotoki_> AFAIK, Tuesday is for cross-project design summit sessions.
21:27:43 <mestery> amotoki_: Yes! The schedule for that was released already. There are a lot of good sessions in there!
21:27:44 <markmcclain> amotoki_: correct the TC has scheduled it
21:28:17 <mestery> I hope many Neutron developers can attend a mix of those sessions so we are represented well in the cross-project sessions.
21:28:53 <mestery> #topic Nova Parity
21:29:04 <mestery> Nova Parity is our most critical item for Juno from a project perspective.
21:29:21 <mestery> markmcclain has volunteered to lead this effort, as it branches across many things, including across projects.
21:30:05 <mestery> There will be plenty of work here in Juno, so anyone who wants to participate this is encouraged to help out!
21:30:25 <mestery> We will also have a handful of Summit sessions in this area as well.
21:31:03 <mestery> markmcclain: Anything to add here at this point?
21:31:39 <markmcclain> Not too much other than I think we'll setting up a regular subteam meeting just to keep everyone in the loop on this work
21:31:49 <mestery> +1 to that markmcclain!
21:32:03 <mestery> Any questions on Nova Parity?
21:32:48 <mestery> #topic Tempest
21:32:53 <mestery> mlavalle: hi!
21:32:57 <mlavalle> hi
21:33:17 <mlavalle> we are on the home strecth to finsih the api tests for icehouse
21:33:30 <mlavalle> we have merged 19 tests out of 28 that we are tracking
21:33:39 <mlavalle> this past week there were no new merges
21:33:43 <mestery> mlavalle: Awesome!
21:33:55 <mlavalle> that's kind of logical, since we are working on the last ones
21:34:00 <mestery> mlavalle: It looks like there are a handful of patches which are close to merge though.
21:34:13 <mestery> With one core tempest +2
21:34:20 <enikanorov_> mlavalle: i wonder if there is any guidelines for scenario tests
21:34:26 <mlavalle> yes and I pijted them out to the tempest cores this part Thurdsady
21:34:33 <mlavalle> and I will do it again this week
21:34:35 <mestery> Thanks mlavalle!
21:34:50 <marun> enikanorov_: it should be something that can't be tested in a functional test
21:34:54 <mlavalle> every Thursday I bring up to their attention the ones that only require one more +2
21:35:10 <enikanorov_> mlavalle: i see two general issues with them - consume much time with each scenarios being added and 2) - depend on many sw components
21:35:26 <mlavalle> I also proposed a session for the design summit
21:35:48 <mestery> mlavalle: Yes, I saw that, it's an important session topic I've actually preapproved that one already. :)
21:36:11 <mlavalle> hoping to get it apporved. The tempest cores are aware and I think that if it gets approved it should be a joint session
21:36:26 <enikanorov_> mlavalle: are you aware of scenario testing for adv services other than lbaas?
21:36:49 <mlavalle> enikanorov: I am planning to pay more attention to scenario in Juno
21:37:27 <mestery> Thanks for the update mlavalle!
21:37:30 <mlavalle> in fact, I am planning to send a message this week to the ML asking for suggestions of scenario tests that should be implemented
21:38:02 <mlavalle> and we will start tracking them like we are tracking api tests
21:38:04 * sc68cal needs to get people on v6 side
21:38:08 <mlavalle> I'm done
21:38:13 <enikanorov_> mlavalle: just thinking if scenarios should be a separate job...
21:38:21 <Sukhdev> mlavalle: Good…Look forward to discussion on ML
21:38:36 <mestery> Yes, thanks for continuning to drive this important area mlavalle!
21:38:48 <mlavalle> enikanorov: let's talk after the meeting
21:38:56 <mestery> #topic L3
21:39:04 <mestery> I know carl_baldwin may not make it today, but I saw Swami here.
21:39:12 <mestery> Swami: Did you want to update a bit on DVR in this slot?
21:40:00 <mestery> In lieu of the L3 team providing an update here, carl_baldwin has updated the agenda with the L3 sub-team update.
21:40:14 <mestery> I see more DVR patches being pushed, which is great!
21:40:25 * mlavalle has to run to another meeting. see you all in the irc channel
21:40:34 <mestery> #topic IPv6
21:40:37 <mestery> sc68cal: Hi
21:40:39 <sc68cal> Hellloooooooooo
21:40:51 <sc68cal> So - big first thing is getting all our BP's into neutron-specs
21:41:00 <Swami> Sorry we have pushed the L3 agent today for the DVR
21:41:18 <mestery> Thanks Swami, saw that, nice!
21:41:26 <sc68cal> seeing a couple new ones that baoli added to the meeting agenda, will instruct tomorrow to move to neutron-specs
21:41:29 <mestery> sc68cal: You have been one of the neutron-specs pioneers, thanks!
21:41:40 <sc68cal> Gotta keep my stackalytics stats up
21:42:02 <sc68cal> Seeing some bugs in nwdiag - but sort of proud of the diagram for the provider-nets-slaac
21:42:07 <amotoki_> sc68cal: it would be helpful if we have links to reivews (both spec and patches)
21:42:09 <sc68cal> it's pretty nice overall
21:42:12 <sc68cal> https://review.openstack.org/#/c/88043/
21:42:45 <sc68cal> amotoki_: where do you want the links?
21:43:06 <amotoki_> sc68cal: the meeting wiki page is a good place to share.
21:43:11 <sc68cal> oh in this meeting's agenda? ah gotcha
21:43:34 <amotoki_> sc68cal: yes, everyone see this page every week :-)
21:43:45 <mestery> :)
21:43:55 <sc68cal> Alright. well then I yield my time since I am not prepared
21:44:02 <mestery> Thanks sc68cal. :)
21:44:09 <mestery> #topic Open Discussion
21:44:18 <mestery> I wanted to highlight one thing here: Code Reviews
21:44:25 <mestery> We have a pretty large queue at the moment.
21:44:38 <mestery> So I encourage core and non-cores to help weed through that with reviews in the coming week.
21:44:47 <mestery> Also, marios brought up the idea of a weekly Neutron review meeting.
21:44:54 <mestery> Does anyone have thoughts on that one way or another?
21:45:06 <sc68cal> I like it- the turnaround time on reviews has always been frustrating
21:45:21 <marios> seems like there are folk for this, but if core devs aren't behind it/available then it's a non starter really
21:45:22 <pcm_> +1
21:45:35 <mestery> marios: I'm all for this actually, I think it makes sense.
21:45:36 <marun> I think we need more than that
21:45:40 * pcm_ +1 on the idea
21:45:49 <rkukura> +1
21:45:52 <mestery> marun: Please explain. :)
21:46:00 <marun> simply adding more resources to reviews isn't enough - we need to actually target our efforts
21:46:01 <Swami> +1
21:46:28 <marun> but by all means, pursue the same strategy more intensely
21:46:35 <mestery> marun: Do you mean specific areas of code?
21:46:37 <marun> I'll have words to say about that at summit
21:47:24 <marun> No, I mean attaching people to reviews for the entire cycle, making sure we're targeting things based on priority, and making sure turnaround by both submitters and reviewers is a priority
21:47:41 <marun> entire cycle -> life of the change
21:47:59 <mestery> marun: I like that idea a lot.
21:48:08 * pcm_ thinks having an IRC will help disseminate knowledge about reviewing to non-cores
21:48:09 <banix> marun: that makes a lot of sense
21:48:27 <sc68cal> would be nice to have a sherpa for a blueprint
21:48:30 <mestery> marun: Is this something we should be pursuing prior to the summit with a discussion on the ML? Or do you already have a proposal here?
21:48:38 <marios> marun: that makes sense. one of the issues i've had is differing advice/opinions of cores but which are expressed very widely apart timewise
21:48:59 <marios> marun: could having a review hour be the vehicle with which reviews are assigned to cores
21:49:15 <marios> (e.g. when they can't trivially be reviewed/closed/-1 there and then)
21:49:31 <mestery> I think if we go down this path, we need to assign reviews to cores at BP creation time, rather than gerrit push time.
21:49:34 <marun> so, we have work to do in any case
21:49:42 <marun> and it will have to evolve
21:49:56 <marun> I do think more attention to reviews is good, so I'm not saying that a weekly meeting is bad
21:50:08 <mestery> I look forward to your lively discussion in Atlanta marun. :)
21:50:25 <marun> /soapbox :)
21:50:42 <enikanorov_> i have a question on a new feature that was added recently
21:50:42 <marios> not all of us will be in atlanta though. is there something you could add to the existin mailing list discussion (or if i have misunderstood and it is in fact a different issue?)
21:50:48 <enikanorov_> nova notification
21:50:48 <marios> marun: ^^
21:51:12 <enikanorov_> right now it is implemented by calling nova via nova client
21:51:26 <marun> marios: I'll aim to raise the issue on the mailing list pre-summit, and there will be an etherpad as part of the summit discussion that remote participants can edit
21:51:30 <enikanorov_> is there any plans to do it via rpc or implement bulk notifications?
21:51:43 <amotoki_> enikanorov_: are you talking in #-neutron channel?
21:51:44 <marios> marun: thanks will look out for it
21:52:02 <enikanorov_> amotoki_: i'm multithreaded :Р
21:52:17 <amotoki_> :-)
21:52:18 <markmcclain> enikanorov_: there is not a plan to change neutron to Nova communication to use RPC
21:52:36 <markmcclain> arosen1 would probably know more about bulk callbacks
21:52:55 <marun> enikanorov_: I'm pretty sure the decision to not use rpc was entirely deliberate
21:53:02 <arosen1> it was.
21:53:13 <enikanorov_> performance concerns come into mind then
21:53:14 <marun> enikanorov_: the nova team is very happy with throughput of their wsgi api
21:53:23 <enikanorov_> ha
21:53:30 <arosen1> dansmith: would be the one to talk to about why we used the rest interface rather than rpc.
21:53:35 <enikanorov_> ok, i'l tell that to our performance testers!
21:54:25 <marun> enikanorov_: It would depend on the configuration of course, but it sounds like there is a disconnect between the configuration that nova folks like dansmith are expecting and what your testers are using.
21:55:09 <enikanorov_> marun: our testers are concerned about the case of lots of VMs spawned in short period of time
21:55:50 <marun> enikanorov_: there are _always_ going to be hard scalability limits
21:55:59 <dansmith> enikanorov_: nova supports bulk notifications via this interface
21:56:00 <enikanorov_> that's not something proven by numbers yet
21:56:15 <dansmith> enikanorov_: if neutron isn't sending them, then that's an "easy" place for improvement
21:56:30 <enikanorov_> ok, good to know
21:56:33 <enikanorov_> thanks!
21:57:19 <mestery> OK, we're coming to the top of the hour here.
21:57:45 <arosen1> enikanorov_:   fwiw i was doing some performance testing last week and i was able to boot 100 instances with 5 nics each  with nova boot <snip> --num-instances without any errors.
21:58:39 <enikanorov_> arosen1: good to know. i'd like to discuss this feature more with you
21:59:09 <mestery> OK, lets continue this and other discussions on the ML.
21:59:18 <mestery> Thanks for attending the Neutron meeting this week folks!
21:59:24 <mestery> #endmeeting