16:00:02 <hongbin> #startmeeting containers
16:00:08 <openstack> Meeting started Tue Aug  9 16:00:02 2016 UTC and is due to finish in 60 minutes.  The chair is hongbin. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:08 <hongbin> #topic Roll Call
16:00:09 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:11 <openstack> The meeting name has been set to 'containers'
16:00:24 <strigazi> Spyros Trigazis
16:00:26 <vipulnayyar> o/
16:00:29 <jvgrant> Jaycen Grant
16:00:38 <muralia> murali allada
16:00:42 <tonanhngo> Ton Ngo
16:00:48 <mjura> hi all
16:00:55 <mjura> Michal Jura
16:01:02 <hieulq_> Hieu LE o/
16:01:09 <mkrai> Madhuri Kumari
16:01:15 <hongbin> Thanks for joining the meeting strigazi vipulnayyar jvgrant muralia tonanhngo mjura hieulq_ mkrai
16:01:19 <hongbin> #link https://wiki.openstack.org/wiki/Meetings/Containers#Agenda_for_2016-08-09_1600_UTC Today's agenda
16:01:20 <Drago> o/
16:01:25 <hongbin> Anything needs to be added to the agenda?
16:01:53 <hongbin> #topic Announcements
16:02:02 <hongbin> I have no announcement, anyone else has?
16:02:05 <strigazi> We have https://hub.docker.com/r/openstackmagnum/
16:02:11 <muralia> nice
16:02:24 <tonanhngo> +1
16:03:04 <strigazi> hongbin egor and adrian are also owners
16:03:22 <strigazi> I didn't have anyone's else username or email
16:03:22 <tcammann> Vaguely here, got a concurrent meeting.
16:03:29 <yatin_> +1
16:03:47 <hongbin> Thanks strigazi for acquiring this account
16:04:12 <hongbin> If anyone else want to own this repo, feel free to ping strigazi offline
16:04:23 <hongbin> #topic Review Action Items
16:04:26 <hongbin> None
16:04:32 <hongbin> #topic Essential Blueprints Review
16:04:38 <hongbin> 1. Support baremetal container clusters (strigazi)
16:04:43 <hongbin> #link https://blueprints.launchpad.net/magnum/+spec/magnum-baremetal-full-support
16:04:47 <hongbin> strigazi: ^^
16:04:52 <strigazi> k8s-atomic is ready be merged
16:05:05 <strigazi> yuanting did great job
16:05:20 <strigazi> I helped but he deserves most of the credit
16:05:34 <strigazi> I meant yunaying
16:05:49 <strigazi> Testing swarn, it looks in good shape
16:05:58 <strigazi> that's all
16:06:01 <garyduan> Hi Strigazi, do you have any plan to enable dc/os or mesos, swarm on baremetal
16:06:10 <strigazi> mesos is next
16:06:17 <strigazi> then coreos
16:06:28 <strigazi> and then i'd love to add DC/OS
16:06:29 <hongbin> The DC/OS is not even on VM yet
16:06:40 <strigazi> any help is welcome
16:07:05 <eghobo> hongbin: what do you mean?
16:07:05 <tonanhngo> I think Jay is working on DC/OS
16:07:06 <hongbin> Thanks strigazi
16:07:27 <garyduan> Thank you for update, STrigazi
16:07:31 <hongbin> eghobo: We have a BP to support DC/OS, but it is not implemented yet
16:07:40 <eghobo> got it
16:07:57 <hongbin> #link https://blueprints.launchpad.net/magnum/+spec/mesos-dcos
16:08:04 <hongbin> 2. Magnum User Guide for Cloud Operator (tango)
16:08:10 <hongbin> #link https://blueprints.launchpad.net/magnum/+spec/user-guide
16:08:16 <hongbin> tonanhngo: ^^
16:08:20 <tonanhngo> Not much to update this week.  The Bay section was merged.  Next I will do a round of update
16:08:40 <strigazi> I added some notes on the whiteboard
16:08:51 <tonanhngo> to pick up on comments from Ricardo and Spyros that I missed, along with new changes and features
16:09:08 <tonanhngo> Thanks Spyros
16:09:26 <tonanhngo> That's all I have
16:09:37 <hongbin> Thanks tonanhngo
16:09:52 <hongbin> 3. COE Bay Drivers (jamie_h, muralia)
16:09:59 <hongbin> #link https://blueprints.launchpad.net/magnum/+spec/bay-drivers
16:10:01 <hongbin> muralia: ^^
16:10:26 <muralia> I've made some more progress on this. I've moved all the stack creation, update and deletion methods to the common driver folder.
16:10:46 <muralia> im in the process of changing the entry points to load each drivers driver.py file
16:10:57 <muralia> instead of the template_def.py file
16:11:30 <muralia> i'll continue working on that and will have a patch soon
16:12:37 <hongbin> You finished?
16:12:43 <muralia> yes, done
16:12:48 <hongbin> Thanks muralia
16:13:03 <hongbin> #topic Kuryr Integration Update (tango)
16:13:12 <hongbin> tonanhngo: ^^
16:13:38 <tonanhngo> The Kuryr team did not have a meeting again last night, it seems the person from Japan is not available to host
16:14:01 <tonanhngo> I think they will move the meeting time back to the morning slot for the other folks to join
16:14:32 <tonanhngo> The Kuryr docker image is still not working, I saw several attempts to rebuild it but they failed
16:14:43 <tonanhngo> apparently the code refactor is not completed.
16:15:07 <tonanhngo> I went back to the pre-refactor code to build my own Kuryr image
16:15:19 <tonanhngo> got that working
16:15:33 <hongbin> tonanhngo: Do you think Kuryr is stable enough to be integrated?
16:15:46 <tonanhngo> So with that, we do have Kuryr working with our Swarm bay
16:16:08 <tonanhngo> The code for libnetwork is stable, it's just the refactoring
16:16:17 <hongbin> ok
16:16:19 <tonanhngo> to accomodate the new CNI code
16:16:37 <hongbin> It looks their refactoring breaks
16:17:07 <tonanhngo> Yeah, the pieces are just not in place completely yet
16:17:20 <tonanhngo> so for us, we can just use the older version
16:17:37 <tonanhngo> until they are done with refactoring
16:17:43 <hongbin> sound good
16:17:48 <tonanhngo> So now with the prototyping done, I will start the patches
16:18:11 <tonanhngo> I will put 2 images on the openstackmagnum account
16:18:42 <tonanhngo> The Kuryr old image and the OVS agent image, for Fedora Atomic
16:19:08 <tonanhngo> That's all I have for now
16:19:16 <hongbin> Thanks tonanhngo
16:19:29 <hongbin> #topic Other blueprints/Bugs/Reviews/Ideas
16:19:34 <hongbin> 1. Midcycle summarize
16:19:40 <hongbin> #link https://etherpad.openstack.org/p/magnum-newton-midcycle-topics
16:20:08 <Drago> Broken link
16:20:09 <Drago> https://etherpad.openstack.org/p/magnum-newton-midcycle-topics-restored-998
16:20:13 <hongbin> Since some people was not able to attend the midcycle, I want to take this chance to update all of you
16:20:40 <hongbin> Thanks Drago
16:20:50 <hongbin> There are several important decisions in the midcycle
16:21:01 <hongbin> 1. Rename bay to cluster
16:21:11 <hongbin> and rename baymodel to clustermodel
16:21:29 <tonanhngo> Did we open a blueprint on this yet?
16:21:33 <jvgrant> https://blueprints.launchpad.net/magnum/+spec/rename-bay-to-cluster
16:21:44 <hongbin> The old term "bay" and "baymodel" will still be valid for a while but deprecated
16:22:05 <hongbin> Eventually, the old terms will be removed
16:22:24 <hongbin> 2. Introduce the concept of "node-group"
16:22:43 <hongbin> A cluster will have 2 node-group": master and minion
16:23:12 <tonanhngo> Should we write a spec on this? since it's a major change.
16:23:12 <hongbin> But users can add more node-groups to a cluster later
16:23:21 <Drago> https://blueprints.launchpad.net/magnum/+spec/nodegroups
16:23:32 <Drago> See the blueprint for a link to the spec
16:23:39 <tonanhngo> Thanks Drago
16:23:41 <Drago> I am currently drafting it with jvgrant
16:24:19 <hongbin> The bay to cluster renaming and the adding of node-group are the most important decisions
16:24:40 <hongbin> There are several other decisions, which could be found in the etherpad above
16:24:55 <hongbin> So far, any question/concern for the midcycle decisions?
16:25:08 <rochaporto> is it an addition or a new api version?
16:25:28 <garyduan> Hi, these 2 bps seems to big change. Do you have any plan or schedule on when to complete them?
16:25:52 <garyduan> considering that it might impact the implementation of other features
16:26:00 <hongbin> rochaporto: cluster will be introduced in v1, node-group might be introduced in v2 (I guess)
16:26:08 <muralia> +1
16:26:11 <Drago> +1
16:26:27 <hongbin> garyduan: ASAP
16:26:44 <hongbin> garyduan: Those two BPs should have the most priority
16:26:49 <jvgrant> the bay to cluster change will be done in steps, i should have a patch out today for the new cluster api
16:27:01 <jvgrant> they will use the same backend as the bay api's
16:27:22 <strigazi> jvgrant: it's a patch in the client?
16:27:53 <jvgrant> strigazi: doing the magnum-api first then the client, so i can use the new api's in the client
16:28:12 <rochaporto> so bay to cluster renaming makes it to newton, but not nodegroups?
16:28:12 <tonanhngo> We should stage the patches to avoid disruption.
16:28:47 <strigazi> if jvgrant's patch is only additions it's ok
16:28:47 <tonanhngo> We do have users trying out Magnum, so it may create confusion
16:29:03 <muralia> rochaporto: yes, will not be part of the newton release.
16:29:16 <jvgrant> it will appear to be additions for users in v1
16:29:32 <jvgrant> all previous commands will continue to work as well
16:29:36 <strigazi> I think it's ok
16:30:27 <hongbin> Any further concern/question?
16:30:42 <rochaporto> feature freeze is aug 29th?
16:31:12 <hongbin> rochaporto: I think we have a bit flexiable about the feature freeze date
16:31:29 <hongbin> I will discuss with the team about when is the best date to freeze
16:31:37 <rochaporto> sounds good
16:31:42 <garyduan> It sounds that nodegroup features will be implemented in next release. Is it correct?
16:31:50 <muralia> yes
16:32:20 <hongbin> I think you can start the implementation in this release, if nothing is blocking
16:32:43 <hongbin> However, get everything done in this release is almost impossible
16:33:41 <hongbin> OK. Advance topic
16:33:52 <hongbin> 2. Enable usage of test account files by default (vipulnayyar)
16:33:59 <hongbin> #link https://bugs.launchpad.net/magnum/+bug/1610186
16:33:59 <openstack> Launchpad bug 1610186 in Magnum "Use accounts.yaml file as default method to create users for functional tests" [Low,Triaged] - Assigned to Vipul Nayyar (vipulnayyar)
16:34:04 <hongbin> vipulnayyar: ^^
16:34:10 <vipulnayyar> So, wanted to work on making use of accounts.yaml file default for functional testing. Especially for tests where tenant sharing is required. I'd suggest to have a read up on the bug description if you haven't already. Just wanted to parade this around to get suggestions, if any or some caveats to keep in mind.
16:35:32 <strigazi> Did any other project have similar issues?
16:36:08 <strigazi> Can you describe a little the refactoring?
16:37:06 <vipulnayyar> We use dynamic_credentials in functional tests generated by Tempest. All the users created have separate dynamic tenants.
16:37:29 <vipulnayyar> For tests requiring users in same tenant, we need to use accounts.yaml file.
16:37:55 <vipulnayyar> Ryt now use_dynamic_credentials is used as true implicitly
16:38:06 <vipulnayyar> We need to use it as explicitly false.
16:38:35 <vipulnayyar> AS for refactoring, we just need to add pass a force_tenant_isolation parameter as True to Tempest credentials creation api
16:38:58 <tonanhngo> Is that the recommended approach for Tempest?
16:39:24 <vipulnayyar> Based on the Tempest doc and api that I read, I believe it is the right approach.
16:39:39 <strigazi> link?
16:39:43 <vipulnayyar> But couldn't find any recommended approacj
16:39:43 <strigazi> to the doc
16:40:35 <vipulnayyar> http://docs.openstack.org/developer/tempest/configuration.html#credential-provider-mechanisms
16:41:03 <tonanhngo> Maybe we should ask someone from the Tempest team to help review.
16:41:53 <hongbin> vipulnayyar: In which case, you would need the users from the tenant (which dynamic credential is not statisfied)?
16:41:53 <vipulnayyar> Sure
16:42:12 <strigazi> So it's addint pre-provisioned credentials?
16:42:18 <strigazi> So it's adding pre-provisioned credentials?
16:42:19 <hongbin> vipulnayyar: FOr example, which test cases you want to add?
16:42:24 <vipulnayyar> I need it for bay deletion test by different users under same tenant
16:42:36 <hongbin> I see
16:43:08 <vipulnayyar> The pre-provisioned creds would only be used for tests we require. For rest, we can pass tenant isolation parameter to Tempest cred creation api
16:43:40 <vipulnayyar> which would use dynamic credentials as always
16:44:04 <hongbin> OK. For me, it sounds reasonable approach
16:44:12 <vipulnayyar> I'll ask about this on Tempest, also
16:44:34 <hongbin> Thanks vipulnayyar
16:44:38 <tonanhngo> +1, they would have more expertise than us here
16:44:51 <hongbin> 3. Add OpenSUSE to Magnum (Michal Jura)
16:44:51 <strigazi> update the ticket with more information, when you ask
16:45:01 <hongbin> #link http://lists.openstack.org/pipermail/openstack-dev/2016-August/100833.html
16:45:32 <hongbin> mjura: ^^
16:45:43 <mjura> we talked about this already with strigazi
16:45:56 <mjura> I'm going to move this change to contrib/drivers
16:46:14 <mjura> and adopt changes from comments which I've got already
16:46:24 <mjura> thank you for code-review
16:46:29 <tonanhngo> mjura:  do you plan to support this driver for the long term?
16:46:45 <mjura> tonanhngo: yes, this is our idea
16:46:54 <muralia> mjura: +1. once the driver becomes stable, we'll move it to /drivers.
16:47:01 <mjura> tomorrow I have also meeting with SUSE engineers
16:47:13 <mjura> from containers team
16:47:36 <mjura> we would like to support it for long term as other OpenStack components
16:47:45 <tonanhngo> +1
16:47:56 <mjura> we did already demo about openSUSE driver and Magnum
16:48:07 <mjura> everything works fine
16:48:10 <strigazi> I think it's good to have a contrib driver, it will help us develop the dynamic mechanism for bay drivers
16:48:33 <mjura> yes, I'm happy to help with it
16:48:43 <tonanhngo> muralia: is there a mechanism for users to pick up contrib drivers?
16:48:59 <tonanhngo> (to load the driver)
16:49:02 <muralia> yes, we'll need to just add an entry point
16:49:28 <mjura> muralia: if somebody would like to add new driver
16:49:37 <mjura> he has to add it to entry points
16:49:45 <tonanhngo> Maybe we should add that details to the user guide, if it's not there yet.
16:50:04 <mjura> and adding this to magnum.conf to enabled_definitions is not enough
16:50:14 <muralia> yes, i'll update the user guide. mjura, lets chat once I submit my patch. this will be a good opportunity to test it.
16:50:33 <strigazi> mjura: it didn't work?
16:50:51 <mjura> muralia: is this patch already submited
16:50:55 <muralia> not yet
16:51:02 <mjura> strigazi: entry_points has to be updated too
16:51:11 <mjura> magnum.conf is not enough
16:51:18 <strigazi> ok
16:51:24 <mjura> I will include this in README file
16:51:42 <mjura> about installation
16:51:51 <mjura> until tomorrow I will upload new patchset
16:51:57 <strigazi> cool
16:52:00 <mjura> and I will be gone for one week
16:52:10 <mjura> so, no worries I will be back :)
16:52:13 <strigazi> also cool
16:52:48 <hongbin> mjura:  Thanks for your interest to contribute the driver to Magnum. I am looking for the new driver.
16:53:03 <hongbin> #topic Open Discussion
16:53:03 <mjura> I thank you to everybody
16:53:11 <mjura> for support and comments
16:53:14 <strigazi> Make the test coverage job voting (Spyros Trigazis)
16:53:21 <strigazi> When a change decreases the test coverage we should not merge it. If we want to increase our test coverage we should stop adding code that decreases it.
16:53:24 <mjura> yes, I'm big fan of Magnum and k8s
16:53:36 <strigazi> thoughts?
16:53:53 <muralia> do other projects do this too?
16:53:58 <strigazi> yes
16:54:02 <strigazi> rally
16:54:11 <strigazi> gophercloud
16:54:40 <strigazi> haven't checked for others
16:54:45 <Drago> What about template changes?
16:54:46 <strigazi> but I think makes sense
16:54:58 <strigazi> I mean python code
16:55:23 <strigazi> I would like the same policy for templates too, but one step at a time
16:55:56 <tonanhngo> What if there is a major change and the patches are broken up into small pieces?
16:56:27 <strigazi> The tests should be broken in pieces as well
16:56:54 <strigazi> When you write a function you also write a test function
16:57:33 <muralia> are we ready to make the job voting right away? or do we need to refactor and add more tests before we enable voting?
16:57:54 <strigazi> I can modify the current job to report
16:58:04 <muralia> ok.
16:58:05 <strigazi> when the coverage decreases and then decide
16:58:14 <muralia> i like that
16:58:23 <tonanhngo> Does the coverage test point to the function that fails the coverage?
16:58:44 <strigazi> not sure what you mean, time is up
16:58:52 <hongbin> 2 minutes
16:59:03 <strigazi> Ton, can you elaborate
16:59:13 <tonanhngo> we can follow up on the other channel
16:59:17 <strigazi> ok
16:59:39 <tonanhngo> Thanks to the Rackspace team for hosting the midcycle, it was very productive in my opinion
16:59:47 <strigazi> +1
16:59:48 <hongbin> Thanks
16:59:51 <muralia> it was great meeting all of you.
17:00:14 <hongbin> All, thanking for joining the meeting.
17:00:17 <mjura> +1
17:00:17 <hongbin> #endmeeting