21:02:26 <markmcclain> #startmeeting Networking
21:02:27 <openstack> Meeting started Mon Jan 27 21:02:26 2014 UTC and is due to finish in 60 minutes.  The chair is markmcclain. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:02:27 <armax> hello
21:02:28 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:02:30 <openstack> The meeting name has been set to 'networking'
21:03:06 <markmcclain> #link https://wiki.openstack.org/wiki/Network/Meetings
21:03:11 <markmcclain> #topic Announcements
21:03:23 <markmcclain> #info Feature Proposal Deadline is Feb 18th
21:03:43 <salv-orlando> a slightly late aloha to everybody
21:03:57 <markmcclain> I-3 features must in review by end of day on the 18th
21:04:51 <markmcclain> this will enable the team time to review
21:05:00 <markmcclain> #link https://launchpad.net/neutron/+milestone/icehouse-3
21:05:11 <markmcclain> #topic Bugs
21:05:20 <markmcclain> #link https://bugs.launchpad.net/neutron/+bugs?search=Search&field.importance=Critical&field.status=New&field.status=Confirmed&field.status=Triaged&field.status=In+Progress
21:05:36 <markmcclain> We have 6 critical bugs we're tracking
21:06:05 <Sukhdev> Hi
21:06:44 <markmcclain> salv-orlando has been working on https://bugs.launchpad.net/neutron/+bug/1273386
21:07:20 <salv-orlando> sdague and russellb are actually doing the actual stuff (upgrading kernel in node pool and devstack)
21:07:29 <salv-orlando> I just limited myself to reading the logs ;)
21:07:52 <russellb> yeah, it's a WIP ... https://review.openstack.org/#/c/69445
21:08:08 <russellb> we're going to start with doing it as a new experimental job so we can make sure the new kernel doesn't make the world explode
21:08:38 <russellb> i have a few more infra pieces to change before we can light it up
21:09:21 <marun> is there any indication if a recent patch is causing this kernel issue to manifest?
21:09:41 <salv-orlando> marun: tried a few reverts with no luck, issue still occurring
21:09:47 <marun> salv-orlando: :(
21:10:13 <marun> still, I suppose it's good motivation to stop running an ancient kernel in the gate.
21:10:25 <salv-orlando> we're shooting for a kernel issue now, but it should not be excluded an issue with utilities such as iputils
21:10:31 <russellb> salv-orlando: can you update https://review.openstack.org/#/c/69448/ ?
21:11:25 <salv-orlando> russellb: sure
21:12:25 <russellb> so the kernel bug has been happening for as long as we have history in logstash (2 weeks)
21:12:46 <russellb> so could have been going much longer
21:12:56 <markmcclain> that is my guess
21:13:38 <salv-orlando> I checked a few occurrences (well just two) of bug 1254890 before 2014/01/15 and did not find kernel crashes
21:13:50 <russellb> salv-orlando: ah..
21:14:04 <banix> bye
21:15:37 <salv-orlando> anyway, I don't want to sound rude, but perhaps it would be sufficient to say that everybody interested in helping us could jumo in either #openstack-infra, #openstack-qa, #openstack-neutron and #openstack-gate
21:15:43 <banix> sorry wrong window.
21:15:51 <salv-orlando> as I guess there is also a need to move forward with the meeting agenda
21:16:25 <markmcclain> yeah coordinating on IRC will be super helpful
21:16:54 <markmcclain> moving onto other bugs
21:16:55 <markmcclain> https://bugs.launchpad.net/neutron/+bug/1270212
21:17:40 <markmcclain> There is a fix for this bug, but it is currently blocked by isolated job failures
21:17:56 <samuelbercovici> hi
21:17:59 <markmcclain> Any of the the other bugs we need discuss?
21:18:20 <salv-orlando> just to say darragh pushed and had merged a patch in tempest
21:18:30 <salv-orlando> related to bug 1253896 which raises the ping timeout to 120 seconds
21:18:42 <markmcclain> ok good to know
21:18:58 <salv-orlando> this will avoid the issue we sometimes see with the 1st dhcpdiscover being missed, and ping timing out because of this
21:19:18 <markmcclain> nice
21:19:19 <salv-orlando> I'm sorry I don't know Darragh's IRC handle
21:19:22 <sdague> is there any understanding on why the isolated jobs suddenly went to 70% fail rates?
21:19:32 <sdague> because they were working 2 weeks ago
21:20:14 <markmcclain> no still have not identified the change that caused the breakage
21:20:56 <salv-orlando> I reverted a few commits, and could not single out a specific change. I'm not even sure it is a neutron change.
21:21:18 <salv-orlando> just saying, every time the neutron isolated job fails, the kernel bug is there.
21:21:33 <salv-orlando> but I don't have data to prove 100% correlation
21:22:03 <marun> salv-orlando: and does an upgraded kernel fix things?
21:22:57 <salv-orlando> marun: I found out this only on friday evening. Still did not have time to set up a new repro env. Other people from the community are helping here. See the patch russellb posted earlier during the meeting.
21:23:14 <marun> salv-orlando: ok
21:23:28 <markmcclain> ok.. we talk more about the kernel bug in the channel after the meeting
21:23:38 <markmcclain> let's keep moving for now
21:23:49 <markmcclain> #topic Docs
21:23:54 <markmcclain> emagana: hi
21:23:57 <emagana> hello!
21:24:08 <emagana> markmcclain: Still working on HA at the moment
21:24:25 <emagana> Will update the wiki with the BP and Gerrit Commit
21:24:26 <markmcclain> ok
21:24:43 <markmcclain> any other doc items to highlight?
21:25:03 <emagana> There is a meeting organized by Doc team on 1/29
21:25:09 <emagana> I am planning to attend it
21:25:19 <markmcclain> great
21:25:21 <Sukhdev> I had sent an email on Neutron API - this seems to fall into docs area
21:25:27 <emagana> nothing else from by side
21:25:37 <markmcclain> emagana: thanks for updating
21:25:57 <markmcclain> Sukhdev: the API docs do fall into that area
21:26:01 <EmilienM> emagana: let me know if you need quick reviews
21:26:02 <markmcclain> #topic Nova Parity
21:26:06 <beagles> hi
21:26:09 <emagana> EmilienM: will do!
21:26:12 <markmcclain> beagles: looks like you made lots of agenda updates
21:26:47 <beagles> yes.. but unfortunately the were of the form of "nothing changed" and "not done" kind of deals
21:27:04 <beagles> last week was hectic for lots of folks for a variety of reasons (good ones)
21:27:27 <beagles> but we should be able to get some of this stuff nudged along this week
21:27:33 <markmcclain> great
21:27:42 <Sukhdev> markmcclain, emagana: perhaps the docs need to be updated to reflect the correct behavior of API usage
21:28:23 <annegentle> Sukhdev: have you logged a bug like these? https://bugs.launchpad.net/openstack-api-site/+bugs
21:28:36 <beagles> in the case of the bugs against API implementation we'll need to figure out how they fit in the current icehouse schedule.. while the code lives in nova repo, it will need a neutron resource to implement them
21:28:51 <annegentle> Sukhdev: you can use the tag netconn-api
21:28:52 <markmcclain> beagles: right
21:29:07 <Sukhdev> annegentle: will do - thanks
21:29:12 <emagana> Sukhdev: all API miss configuration should follow up by bug in launchpad, feel free to add any to the wiki
21:29:14 <beagles> arosen has several irons in the fire already for that code so we'll likely be triaging in the truest sense of the word :)
21:29:47 <markmcclain> true
21:29:56 <annegentle> emagana: have you been able to meet with NickChase or others on the separate Networking guide start?
21:30:07 <beagles> but nothing new besides
21:30:17 <markmcclain> beagles: thanks for updating
21:30:17 <annegentle> (woops, sorry the topic already changed, emagana just let me know outside of the meeting, that's fine)
21:30:25 <emagana> annegentle: ok
21:30:32 <markmcclain> #topic Tempest
21:30:34 <markmcclain> mlavalle: hi
21:30:36 <mlavalle> Hi
21:30:59 <markmcclain> looks like we have a bunch of open reviews
21:31:02 <mlavalle> I want to highlight that all the tests in the api gap analysis have been claimed
21:31:10 <markmcclain> that's good news
21:31:16 <mlavalle> every topic has an owner now
21:31:42 <mlavalle> all the people trained in Montreal are actively contributing patchsets
21:32:01 <mlavalle> so my priority over the ned few weeks is now code review and help them merge their code
21:32:05 <markmcclain> that's great news
21:32:33 <markmcclain> Any tempest questions for mlavalle?
21:32:43 <HenryG> Special thanks to mlavalle for training several people on how to write tempest tests.
21:32:47 <mlavalle> Do we have a limitation on the Neutron side to start merging Tempest code?
21:33:18 <mlavalle> HenryG: Thanks for your patchste> I reviewed yesterday
21:33:45 <markmcclain> mlavalle: if the code passes gating then it should be eligible for merging
21:34:00 <markmcclain> if you're concerned that we need some neutron cores to take a look
21:34:07 <markmcclain> we can work improving coverage
21:34:08 <mlavalle> ok, I'll take that message to the Tempest meeting on Thursday
21:34:29 <mlavalle> and as the code gets ready I will help for help for some reviews
21:34:39 <mlavalle> ask^^^^
21:34:57 <mlavalle> that's all I have
21:35:04 <markmcclain> mlavalle: thanks for update
21:35:15 <markmcclain> #topic L3
21:35:21 <markmcclain> This is a new report this week
21:35:33 <carl_baldwin> Not much yet.
21:35:41 <carl_baldwin> Initial focus will be around building community around DVR/HA routers and building consensus.
21:35:50 <carl_baldwin> I've added a link to the subteam wiki from the Neutron meeting wiki page.
21:36:08 <carl_baldwin> Please review the link and contact me if you desire to participate as a part of this sub team.
21:36:22 <carl_baldwin> One question is how should this sub team relate to and interact with the established DVR sub team?
21:36:52 <markmcclain> I think that DVR is a specific use case
21:37:07 <mestery> markmcclain: +1 to that
21:37:14 <markmcclain> whereas the L3 subteam is coordinating the bigger picture
21:37:48 <lifeless> how far away is DVR/HA routers? We're discussing that as a deliverable for the next tripleo iteration...
21:38:06 <lifeless> and whether we do it within neutron or by layering something on top ?
21:38:23 <markmcclain> lifeless: my current feeling is that it's not ready for merging in Icehouse
21:38:26 <Swami> We are still in the process of refining the design
21:38:40 <mestery> lifeless: There is a meeting in 2 weeks around this in the bay area, there was a thread on the ML announcing it.
21:38:46 <Swami> may not be ready for Icehouse. We may have a WIP code by Icehouse 3 but will not be ready for Icehouse
21:39:01 <carl_baldwin> HA routers has some code up but I have not reviewed its progress yet.
21:39:05 <lifeless> jog0: ^ you might like to pop your head into that meeting
21:39:50 <carl_baldwin> k, that is all I have.
21:39:53 <dkehn> NobodyCam: saw that, did something make it into the master
21:40:01 <dkehn> sorry wrong win
21:40:03 <Swami> For DVR discussion we are planning to have a face to face meeting in the Bay on the Week of Feb 10th. If interested please send me an email.
21:40:14 <markmcclain> carl_baldwin: thanks for leading this subteam
21:40:21 <carl_baldwin> Glad to help.
21:40:28 <markmcclain> #topic IPv6
21:40:31 <markmcclain> sc68cal: hi
21:40:32 <sc68cal> Hello
21:40:44 <jog0> lifeless: looking at email
21:40:55 <sc68cal> I am currently working to address a concern by a Nova core who has -2'd the IPv6 hairpin fix
21:41:25 <sc68cal> There appears to have been a race condition with using the utils.is_neutron function - so I am pivoting to use a config flag in nova.conf to enable/disable hairpin behavior on bridge interfaces
21:41:33 <jog0> sc68cal: that was me
21:41:43 * jog0 hates race conditions
21:42:26 <sc68cal> Hopefully we can avoid any race conditions by just using a config flag, lift the -2 from the core reviewer, and get this closed
21:42:38 <sc68cal> #link https://review.openstack.org/#/c/62599/
21:42:59 <jog0> sc68cal: I'll lift the -2 once I see the unit tests passing reliably (re: run recheck no bug a few times)
21:42:59 <sc68cal> We also have reached a consensus on the two new attributes to add to Subnets, for v6 options
21:43:14 <sc68cal> jog0: Thank you
21:43:38 <sc68cal> #link http://lists.openstack.org/pipermail/openstack-dev/2014-January/025573.html IPv6 subnet attributes
21:43:52 <sc68cal> Input from core devs would be appreciated
21:44:20 <sc68cal> I plan on updating the review I had that added one new attribute, to add the second
21:44:38 <markmcclain> ok.. I'll weigh in on the mailing list
21:44:53 <markmcclain> Anything else to raise?
21:45:00 <sc68cal> Not at this point - thank you.
21:45:21 <markmcclain> sc68cal: thanks for the update
21:45:31 <markmcclain> #topic ML2
21:45:37 <markmcclain> rkukura and mestery: hi
21:45:41 <mestery> hi
21:45:53 <rkukura> hi - go ahead mestery
21:45:55 <mestery> So, we've started tracking new ML2 MechanismDrivers which have been proposed in our weekly IRC meeting.
21:46:13 <mestery> We encourage authors of those patches to please join the weekly ML2 IRC meeting to become aware of any ML2 changes affecting them.
21:46:26 <markmcclain> great
21:46:40 <mestery> We're also finalzing some plans on port-binding now which will affect all MDs.
21:46:50 <mestery> Hope to have this finalized by this week's meeting now.
21:47:00 <mestery> Those are the bigs things with ML2.
21:47:08 <markmcclain> ok
21:47:33 <markmcclain> Any questions for ML2?
21:47:34 <mestery> That's all for now, more detail in our Wednesday meeting
21:47:45 <markmcclain> mestery: thanks for updating
21:47:49 <rkukura> I've also been working with the PCI passthru / SR-IOV team who will be adding a new MechansimsDriver
21:48:37 <markmcclain> rkukura: cool
21:48:45 <markmcclain> #topic VPN
21:48:51 <markmcclain> nati_ueno: hi
21:49:01 <markmcclain> we have this review: https://review.openstack.org/#/c/58897/
21:49:04 <nati_ueno> ok rajesh is working on agent implemenation
21:49:05 <nati_ueno> also
21:49:16 <nati_ueno> yes. Hopefully, We can get two cores for that one
21:49:18 <markmcclain> ok was going to ask about the implementation
21:49:40 <markmcclain> remember before we said that we wanted a working implementation to review alongside the API
21:49:54 <nati_ueno> Ah OK
21:50:13 <markmcclain> we can do code/API design reviews
21:50:25 <nati_ueno> I got it. We will request review after we have working implementation.
21:50:29 <markmcclain> but it's nice to actually test end to end
21:50:35 <markmcclain> ok sounds good
21:50:36 <nati_ueno> I agree
21:50:51 <nati_ueno> I think we can make client and agent working until FF deadline,
21:51:02 <nati_ueno> so hopefully, we can make ssl-vpn in icehouse
21:51:06 <markmcclain> ok
21:51:11 <markmcclain> #topic FWaaS
21:51:32 <markmcclain> I know we said that we wanted to wait until after I-2 before activating FWaaS in the gate
21:51:41 <SumitNaiksatam> hi
21:51:53 <SumitNaiksatam> yeah, so SridarK has been working on testing that
21:51:54 <markmcclain> until the isolation tests are working I think we probably want to defer
21:52:04 <SumitNaiksatam> markmcclain: ok sure
21:52:12 <SumitNaiksatam> the update is promising though
21:52:21 <markmcclain> great
21:52:25 <SumitNaiksatam> so far he was able to run FWaaS and VPNaaS concurrently
21:52:37 <markmcclain> ah cool
21:52:40 <SumitNaiksatam> we had tried this before as well, but he did more manual tests now
21:52:51 <SumitNaiksatam> we are not making a final call on this yet
21:52:59 <SumitNaiksatam> but i think we are looking good to propose this
21:53:08 <markmcclain> ok
21:53:11 <SumitNaiksatam> markmcclain: so whenever its a good time, we can consider this
21:53:38 <markmcclain> ok.. let's revisit this next week
21:53:39 <SumitNaiksatam> other than that, we are still sitting on our two main patches - one for service type framework support, and
21:53:50 <SumitNaiksatam> the other one for the service insertion
21:54:09 <markmcclain> looks like both need some review attention
21:54:22 <SumitNaiksatam> yeah, they haven't gotten much core tlc yet, but hopefully soon
21:54:38 <SumitNaiksatam> after all the other bigger issues with the gate and tests are resolved
21:54:49 <markmcclain> yeah.. I think that makes sense
21:54:55 <SumitNaiksatam> thats it from me
21:55:00 <markmcclain> Thanks for updating
21:55:04 <markmcclain> #topic LBaaS
21:55:07 <enikanorov_> hi
21:55:37 <enikanorov_> we get 3 major items on review - loadbalancer instance, ssl and l7 resources for lbaas api
21:55:50 <enikanorov_> agreed to present implementation of l7 and ssl with haproxy
21:56:35 <enikanorov_> yet to decide the way of securing communication between the server and the agent
21:56:49 <enikanorov_> i've sent out an email to openstack-dev on this
21:57:16 <markmcclain> ok
21:57:28 <enikanorov_> also, it would require to add ability to pass file name to CLI to upload the cert
21:57:47 <markmcclain> right
21:57:57 <markmcclain> Anything else?
21:58:11 <enikanorov_> anyone welcome to review! :)
21:58:28 <markmcclain> #topic General Discussion
21:58:39 <samuelbercovici> hi
21:58:54 <samuelbercovici> there is small change: https://review.openstack.org/#/c/64139/
21:59:24 <enikanorov_> right, we had a big discussion with jaypipes on that
22:00:00 <enikanorov_> the bottom of the question is do we want service flavors in this form
22:00:36 <markmcclain> I am not a big fan of this approach
22:00:44 <salv-orlando> I think when I talked with a few other people on irc and gerrit about this specific patch, the bottom line is that it was a work-around to circumvent a limitation of the current provider mechanisms, which would need what enikanorov is saying for a proper fix
22:00:48 <enikanorov_> markmcclain: could you suggest the alternative?
22:01:19 <markmcclain> I'd like to see a proper flavor framework
22:01:34 <enikanorov_> i mean that once we decided that providers are configuration-based object it would be hard to base db-based objects on top of providers
22:01:42 <salv-orlando> and I have expressed my opinion that we should strive to stop accepting workarounds which result in technical debt (because anything for which we say "it will be fixed later on" is technical debt I think)
22:01:56 <salv-orlando> so to cut a long story short I agree with markmcclain.
22:01:56 <markmcclain> salv-orlando: +1
22:02:11 <emagana> I have a question about this commit: https://review.openstack.org/#/c/61663/  Is there anyone against this fix? Specially markmcclain and mestery, any ML2 concerns?
22:02:25 <mestery> emagana: Checking
22:02:33 <enikanorov_> markmcclain: salv-orlando: i'll be glad to move discussion to neutron channel
22:02:48 <mestery> emagana: Nope, I have even given it +2 in the past :)
22:03:22 <emagana> mestery: if you provide your +2 again I will approve it
22:03:25 <markmcclain> emagana: I'd like to see rechecked by jenkins before approving
22:03:41 <emagana> markmcclain: agree
22:03:49 <samuelbercovici> enikanorov_: markmcclain: salv-orlando: can we move to neutron channel to continue discussing this?
22:03:51 <markmcclain> enikanorov_: that works
22:04:15 <markmcclain> We're at time for this week.. thanks for stopping by talk to everyone on the mailing list or IRC
22:04:18 <markmcclain> #endmeeting