21:01:30 <markmcclain> #startmeeting Networking
21:01:31 <openstack> Meeting started Mon Jan 20 21:01:30 2014 UTC and is due to finish in 60 minutes.  The chair is markmcclain. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:01:32 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:01:34 <openstack> The meeting name has been set to 'networking'
21:02:11 <safchain> hi
21:03:03 <markmcclain> #link https://wiki.openstack.org/wiki/Network/Meetings
21:03:28 <markmcclain> #topic announcements
21:04:26 <markmcclain> #info HUGE Thank You to anteaya for organizing the Montreal code sprint
21:04:48 <anteaya> o/
21:04:57 <anteaya> you are welcome
21:05:11 <anteaya> I think it made a big difference to communication
21:05:16 <emagana> markmcclain: +1
21:05:17 <mlavalle> anteaya: you are a great host. Thanks
21:05:29 <anteaya> thanks to everyone who attended and suported the work
21:05:39 <markmcclain> We got a lot done and still have several things to merge once we get to the bottom of the gate problems
21:05:40 <anteaya> jlibosva: thanks for being so available
21:05:47 <anteaya> mlavalle: thanks
21:05:56 <markmcclain> #topic bugs
21:06:46 <markmcclain> We're currently unable to merge any Neutron changes because the isolated job fails >70% of the time
21:07:05 <markmcclain> We've had several people looking at it
21:07:20 <salv-orlando> Considering that on wednesday it was a lot better, it should be easy to find out what to rever
21:07:24 <salv-orlando> revert or fix
21:07:46 <salv-orlando> nothing really suspicious got into Neutron (well nothing got into neutron. period.)
21:08:18 <salv-orlando> so I think people are already looking into tempest/devstack to see if we let slip in some change which broke neutron or highlighted some other fault in neutron
21:08:28 <markmcclain> Right the tricky part well finding the review in another project that is causing i
21:08:32 <salv-orlando> another thing to consider is that the patches we're trying to push are all broken ;)
21:08:39 <salv-orlando> which might be since most are mine....
21:08:56 <sc68cal> :)
21:08:56 <markmcclain> given that translations failed probably not the case
21:09:16 <enikanorov_> there are several old issues appeared in a high failure rate
21:09:46 <markmcclain> Everyone need to be aware that this will impact what can land before the I-2 cut
21:09:46 <enikanorov_> plus that gettextutils issue
21:10:23 <markmcclain> I'd like merge that but the other problems are preventing us from getting even that fix in
21:11:24 <enikanorov_> as said, we have to inspect tempest and devstack commits
21:11:40 <markmcclain> We should be coordinating the irc channel as folks find relevant info
21:11:42 <enikanorov_> salv-orlando: wmwaer minesweeper is also failing each time, could it give some clue?
21:12:06 <markmcclain> Besides the unknown gate blocker... Any other bugs to discuss?
21:12:28 <anteaya> staying up to date on the challe logs will help: http://eavesdrop.openstack.org/irclogs/%23openstack-neutron/
21:12:35 <anteaya> *channel
21:12:40 <salv-orlando> enikanovor_: I have been looking at the logs… it looks like bug 1253896, which should not effect vmware. I can't help but think the changes I and mtreinish did might have an impact
21:12:50 <banix> markmcclain: Will this also impact the I-2 requirement for new plugins?
21:12:52 <salv-orlando> we pretty much did changes which affect most, if not all, tempest modules
21:13:21 <enikanorov_> even your <tab> are amking typos ;)
21:13:37 <enikanorov_> i see, ok
21:13:49 <markmcclain> banix: the testing requirement will not change
21:14:12 <markmcclain> As the deadline is just the first time we publish the compliance list
21:14:38 <markmcclain> It doesn't become final until the feature freeze in I-3
21:15:12 <markmcclain> #topic nova-parity
21:15:17 <beagles> hi
21:15:27 <banix> markmcclain: I see; So even if a new plugin doesn't get merged in I-2, it still can get merged bi I-3?
21:16:04 <markmcclain> banix:yes
21:16:07 <beagles> So work continues on "how-to" get semantically similar/equivalent deployments to the nova-network manager
21:16:38 <salv-orlando> ping?
21:16:53 <salv-orlando> sorry I thought I had been disconnected
21:17:36 <beagles> there was some very indepth discussion last week on the differences in how the accessibility of floating IPS work as well as some conversations on the challenges of doing a "multi-host" thing
21:17:50 <markmcclain> beagles:cool
21:18:33 <beagles> I started some "black magic" config-hacks but got caught in some blind alleys... which reminds me that some things are harder to back out of than others :)
21:19:02 <Swami> beagles: Do you have a link to the meeting minutes regarding the floating iP discussion
21:19:34 <beagles> I filed bugs for all of the missing API implementations so they are out there
21:19:47 <beagles> Swami: they were vocal, but thanks for reminding me that I need to summarize
21:19:52 <markmcclain> Yeah I think multi host will come up later in the week when the sub team meets
21:20:00 <markmcclain> Thanks for filing the bugs
21:20:32 <Swami> beagles: Yes it would be nice if you could summarize the meeting minutes, since it would be usefull in our multi-host discussions
21:20:34 <markmcclain> beagles:anything else?
21:21:28 <beagles> no, that's pretty much it for what happened last week. While it isn't necessarily top priority, I will need to sync up with mlavalle regarding the APIs that are used from nova and gauge what the coverage is like
21:21:42 <beagles> in our tempest api tests that is
21:21:48 <markmcclain> Great... Thanks for updating
21:22:07 <markmcclain> #topic tempest
21:22:09 <mlavalle> beagles: ping me wednesday or thursday, please
21:22:17 <beagles> mlavalle, cool
21:22:19 <markmcclain> mlavelle:hi
21:22:35 <mlavalle> hi, I want to highlight the progress of last week in Montreal
21:22:58 <mlavalle> the most important thing is that we trained a few folks on writing api tests
21:23:15 <mlavalle> and all of them were able to push code at the end of the week
21:23:32 <Sukhdev> mlavalle: you rock!! :-)
21:23:44 <mlavalle> this is key to scale up our efforts to develop tests for Neutron
21:24:03 <mlavalle> and I will follow up with each one of them to make sure they continue making progress
21:24:21 <mlavalle> their patch sets were added to the list in the agenda
21:24:29 <markmcclain> Awesome
21:24:38 <enikanorov_> mlavalle: hope their patches leverage new network client
21:24:52 <mlavalle> and as a bonus, we are finding bugs in Neutron and the docs
21:25:24 <mlavalle> enikanorov: I have that in mind
21:25:41 <markmcclain> Better to find them now than in the freeze period
21:25:47 <Sukhdev> mlavalle: I will be sending out an email on ML to discuss the findings we had regarding API doc issues
21:25:49 <mlavalle> it's not necessarily happening right now, but as we progress in the review cycle, we will get there
21:25:49 <markmcclain> Anything else?
21:25:58 <jorisroovers> enikanorov, they do AFAIK
21:26:03 <mlavalle> that's it
21:26:07 <enikanorov_> good
21:26:15 <markmcclain> mlavelle thanks for the update
21:26:33 <SumitNaiksatam> mlavalle: when do think this will merge: https://review.openstack.org/#/c/67652/1
21:27:07 <mlavalle> I'll take a look at it and ping you later this week
21:27:16 <SumitNaiksatam> mlavalle: great thanks
21:27:17 <markmcclain> salv-orlando: want to update on parallel testing?
21:27:18 <salv-orlando> Parallel, fully 'isolated' testing
21:27:34 <SumitNaiksatam> mlavalle: it blocks some of the plugins
21:27:42 <markmcclain> Thanks for the correction :)
21:28:18 <salv-orlando> A subset of the people in montreal tackled this issue. I, mtreinish and sylvain afchain tackled tempest changes for making jobs working in full isolation
21:28:38 <salv-orlando> we pushed and merged most of the patches. Trouble is that they might be the cause of the gate failures ;)
21:28:44 <enikanorov_> :)
21:29:20 <salv-orlando> rossella_s, carl_baldwin and obondarev tested the patches that we already have under review, and developed other neutron optimisations and/or tempest fixes to make the job work
21:29:36 <markmcclain> Awesome
21:29:40 <salv-orlando> We hope all these patches will eventually stop failing the gate
21:30:24 <salv-orlando> I forgot to mention all the other people who helped us with reviews, marun and nati-ueno and emagana first of all.
21:30:31 <markmcclain> Yeah I can't wait until we get them all merged
21:30:42 <salv-orlando> where first of all does not mean one in particular first of all, but all of them!
21:31:19 <salv-orlando> I actually can't wait for the isolated jobs to start working again.
21:31:33 <salv-orlando> then I'll start worrying about the rest ;)
21:32:22 <markmcclain> Yeah... Will be nice to track down the problem
21:32:29 <markmcclain> Anything else?
21:32:50 <banix> Is this also being looked at: https://bugs.launchpad.net/tempest/+bug/1270244
21:33:16 <banix> i see it is.
21:34:03 <markmcclain> banix: yep is on our radar
21:34:45 <markmcclain> The other item worked on at the sprint was grenade testing
21:35:01 <markmcclain> Lots of progress was made there too
21:35:17 <markmcclain> #topic IPv6
21:35:22 <sc68cal> Hello
21:35:26 <markmcclain> hi
21:35:28 <marun> 500460
21:35:54 <sc68cal> Last week we got approval for the patch that disables hairpinning in libvirt when Neutron is enabled, but it failed at the gate
21:36:20 <markmcclain> Might have to retest when we solve the current gate issues
21:36:43 <sc68cal> Agreed - so at this point we're sort of stuck in limbo on it
21:37:15 <markmcclain> Yeah I think that is going to a be a consistent theme
21:37:23 <sc68cal> Subnet mode patch requires me to address reviewer comments, however some traffic on the ML last week indicates that others have ideas about moving in a different direction, or prosing additional attributes
21:37:55 <markmcclain> Ok.. So that everyone should keep following it this week
21:38:17 <markmcclain> Anything else?
21:38:23 <sc68cal> We have a meeting tomorrow, so we will determine the next step by the end of that meeting. At this point we need to put code in the review, instead of continuing to discuss arch
21:38:43 <sc68cal> I have zero desire to see Neutron have a non-working IPv6 in Icehouse
21:38:57 <sc68cal> but I worry we are running very tight timelines
21:39:24 <markmcclain> Make to post results of the outcome so that others who can't make the meeting know and can comment on final direction
21:39:31 <sc68cal> will do - that's it for me
21:39:49 <markmcclain> Thanks for updating
21:40:08 <markmcclain> #topic ml2
21:40:22 <markmcclain> rkukura or Mestery: hi
21:40:35 <rkukura> I think mestery is offline
21:40:48 <rkukura> we've got a lot of work for i-3
21:41:15 <rkukura> several new drivers in review or coming soon
21:41:33 <rkukura> PCI-passthru / SRIOV
21:41:42 <markmcclain> Yeah I'm a bit concerned given about it
21:41:55 <rkukura> working on updating portbinding for vif_security
21:42:41 <rkukura> A couple of the efforts will have broad impact on drivers, but at least drivers are independent from each other
21:43:17 <markmcclain> Yeah the independence is nice
21:43:32 <rkukura> I think we've got the general approach for port binding to support vif security and provide the MechanismDrivers all the info they need worked out
21:43:49 <markmcclain> Great
21:43:52 <rkukura> We need good progress on code this week
21:44:59 <markmcclain> Agreed... Any questions on ML2?
21:45:01 <rkukura> Also need to decide on the TypeDriver refactor that allows more flexible attributes for network segments
21:45:45 <markmcclain> Ok.. What's the time frame for that decision?
21:45:55 <jorisroovers> ls
21:46:16 * jorisroovers (sorry, wrong window)
21:46:18 <rkukura> should discuss it in more detail at this Wednesday's subteam meeting
21:46:47 <markmcclain> Ok.. Thanks for updating
21:47:01 <markmcclain> #topic lbaas
21:47:58 <enikanorov_> hi
21:48:04 <markmcclain> enikanorov_ want to provide an update from last weeks meeting?
21:48:10 <enikanorov_> yes
21:48:28 <enikanorov_> we have several pathes on review
21:48:43 <enikanorov_> major - ssl extension, loadbalancer instance
21:49:03 <enikanorov_> minor - adding ability to start multiple providers with the same driver (service flavor analog)
21:49:28 <enikanorov_> another lbaas driver has been posted on review recently
21:49:54 <markmcclain> Great.. That driver needs testing before we can merge
21:50:12 <enikanorov_> sure, the author was informed about that
21:50:29 <enikanorov_> still we have to decide on vendor extension framework
21:51:02 <enikanorov_> i think that is a really necessary thing to have. currently each vendor tries to extend core lbaas api
21:51:18 <markmcclain> Yeah...we will need to discuss that on ML since we're running out of time today
21:51:22 <enikanorov_> with what only vendor's driver going to support
21:51:33 <enikanorov_> so i'm going to post a WIP patch later this week
21:51:46 <enikanorov_> that's all from me
21:51:47 <markmcclain> Ok that sounds good
21:51:55 <markmcclain> Thanks for the update
21:52:12 <markmcclain> #topic docs
21:52:24 <emagana> hi all!
21:52:37 <markmcclain> Sorry I accidentally skipped you
21:52:46 <emagana> One important topic to share is about HA Neutron
21:52:52 <emagana> markmcclain: No problem!
21:53:09 <emagana> I have started a mailing list discussion: http://lists.openstack.org/pipermail/openstack-docs/2014-January/003689.html
21:53:44 <markmcclain> Cool
21:54:07 <emagana> Feedback is required, specially from those who have deployed HA is both active/standby  and active/active
21:54:58 <markmcclain> thanks for sharing the link
21:55:03 <emagana> So far the idea is to deprecated some of the staff in standby/active and pushing the default to be active/active.. BTW. Many thanks to EmilienM for the great discussion on this
21:55:42 <emagana> nothing more so far from me, if anyone wants to add something, please go ahead
21:55:57 <markmcclain> thanks for the update
21:56:09 <markmcclain> #topic Open Discussion
21:56:20 <samuelbercovici> hi
21:56:35 <markmcclain> Hi
21:56:59 <samuelbercovici> looks like i missed by an hour
21:57:27 <salv-orlando> techinically 56 minutes...
21:57:38 <samuelbercovici> :-p
21:58:13 <markmcclain> Any other items this week?
21:58:30 <jlibosva> I'd like to ask about SG, especially ask core reviewers to look at https://review.openstack.org/#/c/21946/
21:59:06 <markmcclain> Thanks to everyone for stopping in the week and remember to share info on the gate bugs on
21:59:14 <markmcclain> IRC
21:59:46 <marun> 299429
21:59:48 <markmcclain> jlibosva there was some discussion on the approach
22:00:14 <markmcclain> I know salv-orlando and I have both reviewed
22:00:24 <salv-orlando> I reviewed the nova-side patch.
22:01:30 <markmcclain> Ok we're at time
22:01:35 <markmcclain> #endmeeting