16:00:32 <hongbin> #startmeeting containers 16:00:33 <openstack> Meeting started Tue Aug 30 16:00:32 2016 UTC and is due to finish in 60 minutes. The chair is hongbin. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:35 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:38 <openstack> The meeting name has been set to 'containers' 16:00:38 <hongbin> #topic Roll Call 16:00:43 <strigazi> Spyros Trigazis 16:00:45 <muralia> murali allada 16:00:46 <jvgrant_> Jaycen Grant 16:00:48 <swatson> Stephen Watson 16:00:49 <tonanhngo> Ton Ngo 16:00:54 <Hieu> Heu LE o/ 16:00:55 <rochaporto> Ricardo Rocha 16:00:57 <rpothier> Rob Pothier 16:01:35 <hongbin> thanks for joining the meeting strigazi muralia jvgrant_ swatson tonanhngo Hieu rochaporto rpothier 16:01:43 <hongbin> #link https://wiki.openstack.org/wiki/Meetings/Containers#Agenda_for_2016-08-30_1600_UTC Today's agenda 16:01:49 <eghobo> o/ 16:01:49 <hongbin> Anything needs to be added to the agenda? 16:02:17 <hongbin> #topic Announcements 16:02:24 <hongbin> 1. The python-magnumclient repo is freezed for preparing Newton release. 16:02:24 <strigazi> could we start with FF? 16:02:43 <hongbin> strigazi: we will discuss that later in the agenda 16:03:25 <hongbin> currently, the client repo is freezed, I am going to submit a release to release python-magnumclient for newton 16:04:06 <hongbin> however, please feel free to let me know if you have client patches for newton 16:04:26 <hongbin> we can do a FFE or a backport accordingly 16:04:41 <hongbin> any question about this? 16:05:18 <hongbin> #topic Review Action Items 16:05:24 <hongbin> none 16:05:31 <hongbin> #topic Essential Blueprints Review 16:05:37 <hongbin> 1. Support baremetal container clusters (strigazi) 16:05:43 <hongbin> #link https://blueprints.launchpad.net/magnum/+spec/magnum-baremetal-full-support 16:05:47 <hongbin> strigazi: ^^ 16:06:37 <strigazi> I was on vacations last week so I don't have anything to report. I'm working on atomic_swarm and mesos. 16:07:22 <muralia> strigazi: if you've noticed, i spilt the k8s ironic support into a new driver. 16:07:24 <strigazi> For mesos maybe easier since it's plain ubuntu and the image might be the same 16:07:26 <muralia> we should do the same 16:07:50 <strigazi> I saw that 16:08:12 <strigazi> that's all 16:08:30 <tonanhngo> muralia: makes sense 16:09:00 <strigazi> if we have more than 90% similarity on the templates 16:09:13 <strigazi> it would create much code duplication 16:09:28 <strigazi> but I'll submit different drives 16:09:31 <strigazi> but I'll submit different drivers 16:10:09 <strigazi> Does anyone has another comment? 16:10:46 <hongbin> I guess no 16:10:50 <hongbin> thanks strigazi 16:10:57 <hongbin> 2. Magnum User Guide for Cloud Operator (tango) 16:11:02 <hongbin> #link https://blueprints.launchpad.net/magnum/+spec/user-guide 16:11:07 <hongbin> tonanhngo: ^^ 16:11:46 <tonanhngo> I am writing the section on Scaling, almost done, should upload later today. Got too many distractions last week. 16:11:56 <tonanhngo> that's about all 16:12:11 <strigazi> scaling of deployment? 16:12:52 <strigazi> tonanhngo ^^ 16:13:47 <strigazi> He might disconnected 16:13:59 <hongbin> I guess he mean scaling of the COE 16:14:03 <strigazi> ok 16:14:07 <hongbin> add/remove nodes 16:14:28 <hongbin> thanks tonanhngo 16:14:29 <strigazi> I think we talked about this again 16:14:39 <hongbin> 3. COE Bay Drivers (muralia) 16:14:46 <hongbin> #link https://blueprints.launchpad.net/magnum/+spec/bay-drivers 16:14:52 <hongbin> muralia: ^^ 16:15:08 <tonanhngo> Sorry I got a call 16:15:16 <muralia> so, as discussed last week, we want to continue supporting server_type, os and coe during baymodel creation along with --drivername. 16:15:28 <muralia> so i added the mapping in each driver 16:15:36 <muralia> just like we do today with template_def. 16:16:14 <muralia> because we are reaching feature freeze soon 16:16:31 <muralia> im focusing on server_type, os and coe first before i submit a patch for --driver 16:16:38 <muralia> so everything will stay compatible 16:17:23 <muralia> so with the new mapping, i need to load all drivers using stevedore one by one and create the master mapping. 16:18:03 <muralia> i want to do this once when magnum starts. and then at bay creation time, we can look at the map and decide which driver to use. 16:18:07 <muralia> working on that currently. 16:18:19 <muralia> thats all for now. 16:18:21 <tonanhngo> so would the new option --drivername would replace some of the existing options, like --coe? 16:18:32 <muralia> eventually. 16:18:51 <tonanhngo> and when we add a new driver, we would need to restart the conductor? 16:18:51 <muralia> but we need to keep that for now and deprecate it in a release or 2 16:19:10 <muralia> yes we need to. thats how stevedore works. 16:19:32 <strigazi> the option and the field in clusterTemplate will be replaced? 16:19:38 <adrian_otto> muralia: That restart is not necessarily service affecting, is it? 16:19:58 <strigazi> At this point it is 16:20:13 <adrian_otto> I expect our API can still queue instructions for the conductor when it's down 16:20:13 <tonanhngo> ok, we might want to think about some command to manage drivers to make it more user friendly for the operators. 16:20:23 <muralia> strigazi: yes. i'll change those. adrian_otto: i dont think so, but i need to test it to see 16:20:37 <Drago> Depends if the conductor was doing things at the time, doesn't it? 16:20:44 <strigazi> Drago yes 16:20:56 <strigazi> if there is a bay in creation 16:21:01 <muralia> tonanhngo: we'll be adding a 'driver list'. 16:21:05 <rochaporto> right now if you restart the conductor while it's deploying a bay things go wrong 16:21:27 <Drago> If you do blue-green deployments, I think you'd be fine 16:21:33 <hongbin> a feature to add/remove driver in api without restarting would be cool 16:21:50 <adrian_otto> rochaporto: that makes sense. We would need a way to freeze/thaw the orchestration state. 16:21:58 <tonanhngo> It's really a separate topic, but we also want to make sure the conductor going down won't affect service 16:22:01 <strigazi> when I installed the suse driver 16:22:13 <adrian_otto> but we could look at a way to support draining for conductors 16:22:14 <strigazi> I didn't need to restart 16:22:45 <Drago> tonanhngo, adrian_otto: +1 16:22:49 <tonanhngo> conductor can go down for other reasons 16:23:18 <hongbin> I think there is a patch to add support for adding/removing "service" 16:23:29 <hongbin> where "service" could be magnum-api or magnum-conductor 16:23:40 <muralia> strigazi: really? let me look at that some more then. 16:23:54 <strigazi> doning it again now 16:24:01 <hongbin> a way is to set the conductor "service" to maintain mode, then it won't accept request anymore 16:24:03 <muralia> hongbin: i think that service was something else right? wasnt it related to k8s? 16:24:27 <hongbin> muralia: no, it is not k8s service. it is magnum service 16:24:38 <hongbin> muralia: type magnum service-list, you could see 16:25:42 <strigazi> muralia: verified, I install teh driver again I did list-templates and saw the template definition again 16:26:01 <muralia> nice 16:26:30 <muralia> thats all for my update. I would like this work to get merged in with this release without the --driver support in the cli 16:26:48 <hongbin> thanks muralia 16:26:54 <hongbin> 4. Rename bay to cluster (jvgrant) 16:27:00 <hongbin> #link https://blueprints.launchpad.net/magnum/+spec/rename-bay-to-cluster 16:27:04 <hongbin> jvgrant_: ^^ 16:27:18 <jvgrant_> Lots of good progress this week, majority of external facing changes now done 16:27:28 <jvgrant_> https://review.openstack.org/#/c/362660/ - conf options 16:27:39 <jvgrant_> is the remaining review for anything a user sees 16:27:56 <jvgrant_> working on all the internal references, objects, and db now. 16:28:07 <jvgrant_> https://review.openstack.org/#/c/359470/ - ClusterTemplate db and internal references 16:28:19 <jvgrant_> https://review.openstack.org/#/c/360054/ - Certificate object and references 16:28:28 <jvgrant_> https://review.openstack.org/#/c/362485/ - Functional test updates 16:28:39 <jvgrant_> trying to break them up as small as i can for reveiw 16:28:51 <strigazi> :) 16:29:05 <jvgrant_> after a few more of these merge there will be one final patch with the bay db and objects changes 16:29:13 <jvgrant_> after that we should be done :) 16:29:40 <jvgrant_> big thanks to swatson for the client and conf changes, and everyone who has given good feedback in reviews 16:29:51 <swatson> I should add that consensus from Oslo and Nova is that we mention in release notes what all the CONF changes were 16:30:02 <swatson> Deprecating in oslo config is just 1 step 16:30:24 <jvgrant_> that is all 16:30:39 <hongbin> sweston: yes, a release note to document the conf changes is nice 16:31:19 <swatson> hongbin: Is there a way to make sure we don't miss noting the changes in an official release? 16:32:19 <hongbin> sweston: AFAIK, there is no automate way to detect missing of release note 16:32:31 <hongbin> sweston: we rely on reviewers to catch that 16:33:04 <swatson> hongbin: Is there a review created specifically for release notes? Or some wiki to update? 16:33:04 <hongbin> sweston: I will change the release note again when I am doing a release 16:33:11 <swatson> hongbin: OK 16:33:13 <hongbin> s/change/check/ 16:33:54 <hongbin> #link https://github.com/openstack/magnum/blob/master/doc/source/dev/reno.rst 16:34:30 <swatson> I'll check it out 16:34:32 <hongbin> sweston: the link above is the guide for release note 16:34:51 <hongbin> thanks jvgrant_ for the hard work on renaming :) 16:35:08 <tonanhngo> massive change ! 16:35:18 <hongbin> #topic Kuryr Integration Update (tango) 16:35:27 <hongbin> tonanhngo: ^^ 16:35:31 <tonanhngo> I attended the Kuryr meeting yesterda 16:35:48 <tonanhngo> The team is splitting up the refactoring into two releases 16:36:17 <tonanhngo> first one should be done this week, then followed by the second one in a few weeks to add the RPC/Rest server 16:36:41 <tonanhngo> For our purpose, I think I will just wait for the second release 16:37:00 <tonanhngo> since it would involve more changes in our integration 16:37:28 <tonanhngo> I am continuing debugging the Swarm deployment using the Mitaka version 16:38:01 <tonanhngo> Currently there is some problem with the Neutron agent on the VM creating a VIF 16:38:24 <tonanhngo> So I am comparing with the working environment to see what's missing 16:38:45 <tonanhngo> I am adding another patch to handle the configuration in Magnum 16:39:11 <tonanhngo> That's about all for now 16:39:43 <hongbin> any question for tonanhngo ? 16:40:16 <hongbin> ok, advance topic 16:40:19 <hongbin> #topic Other blueprints/Bugs/Reviews/Ideas 16:40:26 <hongbin> 1. Decide the time for Magnum feature freeze 16:40:32 <hongbin> #link http://releases.openstack.org/newton/schedule.html Newton release scheduler 16:41:04 <hongbin> right now, the client repo is freezed 16:41:13 <hongbin> we need to decide when to freeze the server repo 16:41:51 <adrian_otto> we have one feature we really want to land before freeze 16:41:55 <hongbin> the deadline to release magnum server is Sep 26-30, which is the final release candidate 16:42:01 <adrian_otto> vijendar: that's your patch 16:42:28 <muralia> the version info patch 16:42:32 <muralia> is that it? 16:42:35 <tonanhngo> link? 16:42:41 <strigazi> https://review.openstack.org/#/c/359249/ 16:42:55 <strigazi> tonanhngo ^^ 16:43:12 <adrian_otto> Yes, https://review.openstack.org/359249 16:43:37 <hongbin> ok, we can merge that one before the freeze 16:43:41 <adrian_otto> one of our product integrations depends on that one. It looked close to merge when I reviewed it yesterday, but just FYI... we carea bout that one a lot. 16:44:26 <tonanhngo> ok, we can give it priority in reviewing 16:44:28 <strigazi> the patch looks good, but a merge conflict was created today 16:44:38 <vijendar> adrian_otto hongbin I have updated patch with your review comments… please take a look again 16:44:40 <strigazi> it's solved 16:44:48 <adrian_otto> tx 16:44:57 <hongbin> vijendar: will do that 16:45:15 <vijendar> strigazi: I rebased this morning 16:45:23 <hongbin> besides the version info patch, any other patches that needs to be included in Newton? 16:45:26 <vijendar> thanks adrian_otto hongbin strigazi 16:45:39 <jvgrant_> i'd like all the bay to cluster stuff to make it in 16:45:47 <hongbin> jvgrant_: ack 16:45:48 <jvgrant_> it is close 16:46:28 <muralia> same with the driver work. maybe a weeks worth of work more. mostly for testing 16:46:40 <hongbin> muralia: ack 16:46:47 <rochaporto> +1 drivers 16:47:04 <hongbin> any others? 16:47:49 <tonanhngo> please take a look at the user guide to see if any particular section needs to be filled out 16:47:54 <hongbin> #agreed try to merge bay info, drivers, cluster renaming patches before freezing the magnum server repo 16:48:03 <strigazi> are we going to create a release this week? 16:48:17 <strigazi> as mentioned by Doug in the ML? 16:48:37 <hongbin> strigazi: yes, doug wants a release by this week 16:48:49 <hongbin> strigazi: however, that is not necessary to be the final release 16:49:05 <hongbin> strigazi: we can do another release before the deadline 16:49:15 <tonanhngo> It usually takes several iterations 16:50:05 <strigazi> we might want a release soon for distro packagers work on 16:50:42 <hongbin> strigazi: by this week? 16:50:58 <strigazi> At CERN we build the packages our own so we don't depend on that much 16:51:28 <strigazi> it's not clear to me, some projects have releases already and the packaging started 16:51:50 <strigazi> if we can do many iterations 16:52:00 <strigazi> let's do one this week 16:52:12 <hongbin> ok 16:52:27 <strigazi> and be sure to include the patches mentions in the next iterations 16:52:32 <strigazi> and be sure to include the patches mentioned in the next iterations 16:53:10 <strigazi> @all what do you think? 16:53:38 <muralia> im ok with that 16:53:43 <hongbin> wfm 16:53:51 <tonanhngo> sounds reasonable 16:54:29 <strigazi> it's mostly needed for the dependencies 16:54:48 <strigazi> the new patches don't introduce new dependencies anyway 16:55:36 <adrian_otto> wfm too 16:55:47 <hongbin> yes, also keep in mind that the requirement repo will freeze at some time 16:56:04 <hongbin> after the requirement repo is freezed, new dependencies cannot be added 16:56:42 <hongbin> another thing to keep in mind 16:56:51 <hongbin> the string freeze deadline is this week 16:57:10 <hongbin> so after this week, we are not allowed to change the translation string 16:57:34 <hongbin> #topic Open Discussion 16:58:02 <tonanhngo> Quick question for the team 16:58:24 <tonanhngo> a colleague of mine wants to add a driver for Ubuntu on Swarm 16:58:56 <tonanhngo> I will work with him. Should this go into contrib? or is there enough interest to put it as a supported driver? 16:59:29 <strigazi> To start iterating on it we can start from contrib 16:59:58 <tonanhngo> ok. We will use 16.04, any concern? 17:00:11 <strigazi> +1 17:00:12 <hongbin> time is up 17:00:18 <tonanhngo> thanks 17:00:21 <hongbin> all, thanks for joining the meeting 17:00:25 <hongbin> #endmeeting