03:00:29 <banix> #startmeeting kuryr
03:00:30 <openstack> Meeting started Tue Jan 26 03:00:29 2016 UTC and is due to finish in 60 minutes.  The chair is banix. Information about MeetBot at http://wiki.debian.org/MeetBot.
03:00:31 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
03:00:33 <openstack> The meeting name has been set to 'kuryr'
03:00:39 <baohua> :)
03:00:43 <banix> hi everybody
03:00:59 <baohua> good everning banix~
03:01:03 <fawadkhaliq> hi
03:01:08 <tfukushima> Hi guys.
03:01:14 <banix> hi fawadkhaliq baohua tfukushima
03:01:48 <banix> Agenda #link https://wiki.openstack.org/wiki/Meetings/Kuryr#Agenda
03:02:15 <banix> #topic IPAM
03:02:33 <banix> I don’t see vikasc around
03:02:55 <banix> I think we are in a pretty good shape after most of atches got merged and some bugs got fixed
03:03:41 <tfukushima> Nice.
03:03:47 <baohua> +1
03:04:05 <tfukushima> Has anyone tried it actually?
03:04:26 <banix> We will revisit if vikas shows up; otherwise let’s move on unless there are questions/comments
03:04:41 <banix> tfukushima: Yes I have been testing it
03:04:47 <baohua> i am trying it yesterday, works roughly
03:05:00 <tfukushima> Ok.
03:05:04 <banix> I think we still need to work on cleanup
03:05:15 <baohua> some little privilege problem, need discussed later
03:05:19 <banix> baohua: if you see any problems, please open a bug on launchpad
03:05:24 <baohua> sure :)
03:05:53 <banix> Let us make sure if there are problems we all know about it and track it through filing bugs
03:06:00 <banix> #topic testing
03:06:19 <banix> baohua: do you want to comment
03:06:26 <banix> tell us where we are on that front
03:06:41 <baohua> yes, i'm adding the fullstack testing case, please help review https://review.openstack.org/#/c/265105
03:06:52 <baohua> and, seems gate server cannot pull down docker images.
03:07:03 <baohua> we need that image for the fullstack testing
03:07:21 <banix> what does the fullstack test do?
03:07:27 <baohua> also, i am looking at the rally part.
03:07:39 <banix> Fullstack test #link https://review.openstack.org/#/c/265105
03:07:50 <baohua> the fullstack testing is actually calling docker api to start container
03:08:09 <baohua> and running real-case operations like net-create/delete, port-bind/unbind
03:08:26 <baohua> this can make sure the kuryr works well in real deployment
03:08:26 <banix> you mean join/leave?
03:08:35 <baohua> yes, instance join/leave
03:08:59 <banix> and the issue is you dont have access to docker binary?
03:09:23 <baohua> from http://logs.openstack.org/05/265105/3/check/gate-kuryr-dsvm-fullstack-nv/25e9ab6/console.html
03:09:45 <baohua> seems the issue is pulling images failure
03:10:01 <baohua> docker.errors.NotFound: 404 Client Error: Not Found ("No such image: busybox:latest")
03:10:48 <banix> so the issue is not being able to access the image
03:10:58 <baohua> yes, we might ask the gate maintaine team for help.
03:11:14 <banix> yes pls ask on infra channel
03:11:28 <baohua> sure, and will trigger recheck for sure
03:11:48 <baohua> recently the gate testing is not that stable
03:12:04 <banix> is it possible to use an image we have locally?
03:12:19 <banix> I mean as an alternative, if you have the image can you use it at the gate?
03:12:20 <baohua> u mean adding a local image inside the code repo?
03:12:51 <baohua> i guess the gate server should have external network access.
03:13:27 <banix> I see… wouldn’t be ideal but you can make progress for testing…. but first lets ask infra
03:13:43 <baohua> sure, will also consider your suggestion, thanks banix
03:14:06 <banix> baohua: np. Can you please tell us about Rally tests?
03:14:31 <baohua> oh, i have not much update on rally test, please ask gal about that
03:15:24 <banix> baohua: ok.
03:15:36 <banix> Any questions regarding testing?
03:16:03 <banix> #topic Magnum Integration
03:16:35 <banix> fawadkhaliq: has put out a spec #link https://review.openstack.org/#/c/269039/
03:16:56 <banix> Haven’t been able to review yet unfortunately.
03:17:21 <banix> fawadkhaliq: can you give us an update? What do we need on the Kuryr side? What’s going on on the Neutron side?
03:17:34 <fawadkhaliq> banix: sure
03:18:12 <fawadkhaliq> the idea is to leverage the VLAN aware VMs from Neutron
03:18:28 <banix> is there progress on that front?
03:18:29 <fawadkhaliq> and introduce the notion of port in the container API
03:18:38 <fawadkhaliq> banix: not really :(
03:19:05 <banix> so VLAN aware VMs is not something that is bening worked on actively?
03:19:05 <fawadkhaliq> Nova instances managed by Magnum would launch a Kuryr agent inside the Nova instances
03:19:23 <banix> sorry for interruption pls go ahead
03:19:37 <fawadkhaliq> the agent will be responsible for veth creation as per the COE model, ip assignment and VLAN tagging etc
03:20:09 <fawadkhaliq> Kuryr would have two roles, the agent part and the VLAN allocation engine
03:20:38 <fawadkhaliq> Magnum will utilize these modules in to plug Neutron networks to container it would launch.
03:21:13 <fawadkhaliq> Based on the VLAN aware VMs implementation, each Neutron backend can have a logic to handle it.
03:21:32 <fawadkhaliq> So overall the solution would be vendor neutral works for all kind of design
03:21:36 <fawadkhaliq> more details in the spec :)
03:21:48 <fawadkhaliq> And coming to the progress part of VLAN aware VMs
03:22:04 <banix> thanks fawadkhaliq. All please review
03:22:09 <baohua> fawadkhaliq, before reviewing the spec, a quick question. are both the vm and the inside container using neutron for connectivity? both are endpoints managed for neutron?
03:22:45 <fawadkhaliq> its not being actively looked at, unfortunately. We shall have some discussions there. Goal for Mitaka is to complete the design for this in Kuryr.
03:23:06 <fawadkhaliq> baohua: that's correct, everything on Neutron
03:23:13 <banix> #action review c #link https://review.openstack.org/#/c/269039/ for the Magnum integration spec
03:23:19 <baohua> ok, thanks, will review it
03:23:30 <fawadkhaliq> banix: baohua: thank you
03:23:56 <banix> @fawadkhaliq so is there anyone working on VMLAN aware VMs?
03:24:22 <banix> I mean those who put out the spec, are they working on it?
03:24:48 <fawadkhaliq> banix: last I checked recently, no one is looking at it. I will drive some discussions around importance of that spec and maybe we can get some help
03:25:09 <banix> fawadkhaliq: great. thanks.
03:25:39 <banix> #topic Kubernetes Integration
03:26:17 <banix> We plan to have a meeting for discussing the CNI tomorrow (or today if it is alreay tomorrow for you :) )
03:26:37 <banix> it will be on the openstack-kuryr channel at 15:00 UTC
03:27:10 <banix> The main issue is that CNI as of now don’t care for supporting/utilizing libnetwork
03:27:49 <banix> The question becomes how we support CNI within the Kuryr project
03:28:35 <banix> The time of the meeting is not great for the folks from the West coast of the US
03:28:49 <banix> tfukushima: and company have been investigating this
03:29:09 <banix> irenab: and tony have also given this some thought
03:29:23 <banix> tfukushima: will you be able to attend the meeting?
03:29:54 <tfukushima> Unfortunately, it's 0am for me so I unlikely attend.
03:30:20 <banix> unless there are questions we can leave this topic for the meeting coming up in 12 hours or so
03:30:40 <banix> tfukushima: i understand
03:30:45 <tfukushima> But I shared some knowledge and experiences with Irena and Tony, so they know what I know basically.
03:30:52 <tfukushima> The models of libnetwork and CNI are different and we would need to think about to refactor the current code.
03:31:27 <banix> so looking at the model, do you see any chance CNI could be implemented through libnetwork?
03:31:57 <banix> I mean the add and delete could be implemented through join and leave with some out of band create networks, etc
03:32:25 <banix> i have a tough time seeing it done that way but we dont want to overlook that possibility
03:32:28 <tfukushima> In my opinion, it's not reasonable. They are totally different in terms of the API interface and it's better to think them separately in practice.
03:33:01 <banix> tfukushima: yeah, that leads us to somewhat different implementations
03:33:08 <fawadkhaliq> tfukushima: +1
03:33:13 <banix> ok lets talk more in that meeting
03:33:25 <banix> fawadkhaliq: you can make the meeting at 7am your time?
03:33:38 <fawadkhaliq> I will be there
03:33:43 <banix> great
03:34:04 <banix> hope to get the rest of the team as well
03:34:18 <banix> baohua and others
03:34:21 <banix> #topic Packaging and Kolla
03:34:30 <baohua> 11pm for me, and will try to be there :)
03:34:52 <banix> Have to check with Toni wrt packaging; does anyone know about the packaging work?
03:35:42 <banix> I know Hui Kang has been working on Kola for Kuryr. I will ask him to participate in the next meeting and give us an update. I think we should have that working pretty soon.
03:35:57 <banix> #topic Bugs
03:37:02 <banix> I have started looking at our bug list and triage them. We as a group haven’t been using launchpad for tracking bugs and following up on the fixes
03:37:24 <banix> So please file bugs when you encounter them and pick any you think you can resolve
03:37:50 <banix> There is one bug in particular on teh agenda: #link https://bugs.launchpad.net/kuryr/+bug/1516539
03:37:52 <openstack> Launchpad bug 1516539 in kuryr "failed to start Docker container with network created by kuryr" [High,In progress] - Assigned to Baohua Yang (yangbaohua)
03:38:22 <baohua> yes, the bug is reported last year, and i can repeat it with latest code. i dig out and find the cause is the privilege.
03:38:31 <banix> baohua: you have mentioned that this has to do with Kuryr requiring ro be run as root.
03:38:44 <banix> So this is a somewhat seriou problem we have right now
03:38:53 <tfukushima> I found "1. create network: sudo docker network create --driver=kuryr foo" was missing --ipam-driver=kuryr as well.
03:38:59 <baohua> if we run with root, then it's ok. if without root, then bug happens
03:39:09 <baohua> this can repeat with --ipam-driver=kuryr
03:39:34 <baohua> the cause is when using pyroute to create interface, we need high privilege
03:39:47 <banix> baohua: can you check with toni, tfukushima mentioned in an earlier meeting that toni had some ideas about resolving this requirement
03:40:14 <baohua> ok, his irc name is toni?
03:40:37 <banix> #action baohua to coordinate with apuimedo  to address the requirement to run Kuryr as root
03:40:53 <tfukushima> baohua: His nickname is apuimedo.
03:40:57 <baohua> ok....
03:41:13 <baohua> got this toni, sorry
03:41:43 <banix> please keep us posted baohua
03:41:47 <banix> moving on
03:41:59 <banix> #topic Moving to the big tent
03:42:31 <banix> gsagie will be working on introducing patches needed to move under the big tent
03:42:54 <banix> I think that’s the next step for us as we get our house in shape and cleaned up
03:43:04 <fawadkhaliq> banix: very nice!
03:43:19 <banix> #topic Open Discussion
03:43:43 <banix> Any topic to discuss here?
03:44:40 <banix> alright everybody; thanks for joining and see you all next time and on IRC
03:44:48 <baohua> thanks :)
03:44:52 <fawadkhaliq> thanks
03:44:55 <fawadkhaliq> bye
03:45:00 <banix> #endmeeting