03:02:31 <banix> #startmeeting kuryr
03:02:31 <openstack> Meeting started Tue Jan 12 03:02:31 2016 UTC and is due to finish in 60 minutes.  The chair is banix. Information about MeetBot at http://wiki.debian.org/MeetBot.
03:02:32 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
03:02:34 <openstack> The meeting name has been set to 'kuryr'
03:02:49 <kexiaodong> hi
03:03:10 <banix> baohua: fawadkhaliq vikasc kexiaodong hi
03:03:29 <banix> lets’ wait for a minute and then get started
03:03:40 <baohua> sure, what's the topic today? not see the agenda
03:04:12 <tfukushima> #link Today's agenda https://wiki.openstack.org/wiki/Meetings/Kuryr#Meeting_January_11.2C_2015
03:04:23 <banix> thanks Taku
03:04:24 <baohua> thanks, taku
03:04:53 <banix> So I have put a few topics on teh Agenda but we can discuss others as well
03:05:02 <banix> #topic IPAM
03:05:15 <banix> I think we got several IPAM patches merged
03:05:26 <vikasc> yes banix ,,
03:05:36 <banix> thanks to vikasc
03:05:48 <banix> I ran into a problem testing today and opens a bug for it
03:05:53 <vikasc> i have pushed some more minor fixes in ipam
03:06:19 <vikasc> i pushed a fix for that too..but saw your mail now.. will try to optimize it
03:06:21 <banix> #link https://bugs.launchpad.net/kuryr/+bug/1532982
03:06:23 <openstack> Launchpad bug 1532982 in kuryr "list ports with specific fixed ip" [High,In progress] - Assigned to vikas choudhary (choudharyvikas16)
03:06:39 <vikasc> banix, thanks for suggesting logic
03:06:59 <banix> vikasc: ok thanks I will check your patches;
03:07:12 <banix> we should increase the test coverage so we can catch these
03:07:29 <vikasc> in coming days, i will target that
03:07:52 <banix> Another thing I am noticing and I may be doing something wrong is that I dont see ipam being called for allocating IP address
03:08:15 <banix> vikasc: is there something wrt configuration that needs to be done?
03:08:32 <vikasc> are u using --ipam-driver option?
03:08:41 <baohua> vikasc, we may add to doc to call --ipam-driver to issue docker command.
03:08:53 <vikasc> yesterday only i tested
03:09:07 <vikasc> baohua, its there in devref
03:09:12 <banix> vikasc: can you paste a sample call here please
03:09:22 <vikasc> one moment
03:09:23 <baohua> yes, but too few info now in the doc, i'm thinking adding more steps there.
03:09:23 <banix> or a link to the devref
03:09:36 <banix> vikasc: thank you
03:10:17 <vikasc> banix, https://review.openstack.org/#/c/248010/
03:10:47 <banix> vikasc: others, anything else ypu want discussed regarding IPAM here?
03:10:57 <vikasc> baohua,  please provide your review comments on https://review.openstack.org/#/c/248010/
03:11:14 <baohua> sure, reading it these days :)
03:11:24 <vikasc> baohua, thanks :)
03:11:34 <banix> all please review the IPAM patches
03:12:08 <baohua> +1
03:12:16 <banix> ok moving on
03:12:36 <banix> #topic testing
03:13:04 <banix> i dont think gal is around. baohua, how are things on that front?
03:13:27 <baohua> i'm adding ci testing on container operations, see https://review.openstack.org/#/c/265105/
03:13:45 <vikasc> baohua, may be we can discus scope thing here
03:13:53 <baohua> gal just update his part on the netwroks.
03:14:06 <baohua> see https://review.openstack.org/#/c/259751/
03:14:14 <baohua> sure, vikasc
03:14:45 <banix> vikasc: one second please
03:14:46 <vikasc> https://review.openstack.org/#/c/264653/
03:15:05 <vikasc> banix, sure
03:15:07 <banix> baohua: can you tell us more about what the plan is
03:15:22 <banix> wrt testing that is
03:15:36 <banix> what are the next steps, etc
03:15:55 <baohua> we start at basic network operations like create/delete, and will write the container operations part in next step.
03:16:30 <banix> these are functional test?
03:16:54 <baohua> no, this ci testing will call local docker api to do real actions.
03:17:26 <banix> these are for testing at the gate?
03:17:52 <baohua> currently it can be used at the gate, as fullstack testing.
03:18:10 <banix> are Rally tests in the plan?
03:18:47 <baohua> the rally tests is not started yet, we can discuss it more with gal
03:19:23 <banix> The question as to where they should reside are still under discussion?
03:20:00 <baohua> actually there already exist some code under rally-jobs
03:20:04 <vikasc> banix, can you please elaborate "where should they reside"
03:20:23 <baohua> but i do not think the scope is clear enough
03:20:45 <banix> last week it was questioned whether we should have these tests in the Rally repo or in the Kuryr repo
03:20:59 <vikasc> banix, okie
03:21:10 <fawadkhaliq> gsagie was going to start a discussion on ML for this
03:21:11 <baohua> not sure, but currently it's under kuryr
03:21:27 <banix> ok. baohua how can we speed up the test coverage from the full stack testing to more?
03:21:29 <fawadkhaliq> there are pros and cons for both. but the final call needs to be made.
03:21:54 <banix> Do you need help with reviews? Anything else?
03:22:25 <banix> fawadkhaliq: I agree. Gal mentioned about sending an email to the dev list. I am not sure if he got to it or not
03:22:27 <baohua> oh, on the full testing, still working on updating the code, will send invitation later
03:22:43 <baohua> gal says he is on trip this week
03:23:26 <banix> baohua: ok thanks. Since we want to get to a reasonable state by the end of this cycle, we should make sure we increase or test coverage.
03:23:35 <banix> Anything else regarding testing?
03:24:58 <banix> vikasc: please go ahead wrt capabilities
03:25:33 <vikasc> baohua pushed a patch for changing scope to local
03:25:34 <vikasc> https://review.openstack.org/#/c/264653/
03:25:37 <vikasc> default scope
03:26:05 <vikasc> because if scope is global, etcd/db configuration is required
03:26:15 <vikasc> with devstack it works
03:26:30 <banix> i see
03:26:47 <banix> so dont we do the testing at the gate with devstack?
03:26:49 <vikasc> should we just update manual installation steps or go ahead with this patch
03:27:04 <vikasc> baohua, can better update
03:27:47 <baohua> sure, the problem is clear here. should we enable db backend as default requirement
03:28:10 <baohua> if so, we may need to update the doc to add those steps
03:28:29 <baohua> please help add your comments on that patch :)
03:28:45 <banix> ok sounds good
03:28:47 <vikasc> need to go off for couple of minures
03:28:49 <banix> moving on
03:28:50 <vikasc> *minutes
03:28:55 <banix> #topic Kubernetes Integration
03:29:09 <banix> We need irenab for this topic
03:29:26 <banix> irenab: are you around by any chance?
03:30:02 <tfukushima> banix: I heard irenab got sick.
03:30:20 <banix> tfukushima: ahh. sorry to hear that
03:30:54 <baohua> sorry for that too
03:31:01 <banix> k8s and docker communities seem to have different views of how networking is to be done… we can discuss further next week
03:31:15 <tfukushima> I worked on this with Neutron actually as the PoC.
03:31:26 <banix> tfukushima: please go ahead
03:31:52 <tfukushima> I confirmed pods can communicate over VXLAN with that test network plugin.
03:32:33 <tfukushima> On each node K8s delegate the private IP allocation to Docker and the plugin read the info through the Docker client library, py-docker.
03:32:57 <banix> tfukushima: which plugin you are referring to?
03:33:04 <tfukushima> It creates a network, subnets and ports as Kuryr does.
03:33:43 <tfukushima> banix: The network plugin. But at this point I'm not leveraging Kuryr legitimately as the libaray.
03:33:52 <tfukushima> I used some code though.
03:34:03 <banix> tfukushima: k8s plugin?
03:34:15 <banix> libnetwork plugin?
03:34:20 <tfukushima> A network plugin for K8s.
03:34:34 <tfukushima> #link network plugin spec for K8s https://godoc.org/github.com/kubernetes/kubernetes/pkg/kubelet/network/exec
03:34:38 <banix> i see
03:34:43 <vikasc> thanks tfukushima
03:34:53 <tfukushima> It's not compatible with CNI at this point.
03:35:17 <banix> tfukushima: thanks for the link
03:35:35 <tfukushima> I needed to add subnets on different hosts to a router. And then the pods on the different hosts could communicate each other.
03:36:18 <banix> tfukushima: so is this utilizing kuryr? or independent of it?
03:36:58 <tfukushima> It's independent because Kuryr's code is strongly assuming the libnetwrok model and we need to make its logic more modular.
03:37:37 <banix> tfukushima: yeah that’s the main point of divergence
03:38:07 <tfukushima> For instance, libnetwork has several steps for connecting the container to the network but K8s has two notion, setup and teeardown.
03:38:19 <banix> yup
03:38:55 <vikasc> tfukushima, is not CNI same to k8s what libnetwork is for docker?
03:39:30 <tfukushima> It's going to be at the end but it'd be introduced in 1.2, which is not released yet.
03:39:55 <vikasc> so current k8s dont use CNI?
03:40:23 <tfukushima> It's very similar in my opinion but they're different technically.
03:40:56 <tfukushima> #link CNI discussion on K8s https://github.com/kubernetes/kubernetes/pull/14525
03:41:04 <vikasc> tfukushima, similar in the way that provides way to use different plugins and different in apis?
03:41:15 <vikasc> tfukushima, thanks for the link tfukushima
03:42:01 <vikasc> similar in the way that both provides way to use different plugins and different in apis?
03:42:37 <banix> still evolving wrt plugin architecture as I understand it
03:43:05 <banix> #topic Open Discussion
03:43:15 <banix> Anything else we want to discuss now?
03:43:28 <baohua> y
03:43:37 <baohua> anyone can help review on this? https://review.openstack.org/#/c/264531/
03:43:51 <vikasc> baohua, sure
03:43:53 <baohua> we'd better make the doc part more formated and detailed.
03:43:57 <baohua> thanks vikasc
03:44:01 <banix> sure
03:44:07 <baohua> thanks banix
03:44:15 <banix> fawadkhaliq: Fawad I recall you having an action item from last weelk?
03:44:35 <baohua> recently seems the review process is not that agile :)
03:44:44 <fawadkhaliq> banix: yes, its around magnum integration and nested containers
03:44:50 <fawadkhaliq> Its in progress
03:44:57 <fawadkhaliq> I have started working on the design
03:45:13 <banix> ok, may be update us next week?
03:45:27 <fawadkhaliq> absolutely
03:45:35 <banix> fawadkhaliq: thanks
03:46:08 <banix> so let’s cleanup and finish the IPAM patches :) vikasc please ping me and i try to be responsive
03:46:18 <vikasc> sure banix
03:46:26 <banix> vikasc: what time zone are you at? what time is it right now?
03:46:35 <vikasc> 9:16 AM
03:46:52 <vikasc> india
03:47:08 <banix> vikasc: ok and I am at 10:46pm
03:47:22 <banix> will try to be available in the morning and at night
03:47:38 <vikasc> banix, appreciate that.
03:47:47 <banix> vikasc: thanks you
03:47:50 <banix> thanks everybody
03:47:57 <vikasc> banix, thanks
03:48:00 <banix> we managed to finish the meeting early!
03:48:05 <baohua> thanks everyone
03:48:06 <vikasc> :)
03:48:11 <fawadkhaliq> bye!
03:48:11 <vikasc> bye evryone
03:48:14 <banix> see you next wek and online in the mean time
03:48:17 <tfukushima> Thanks, guys.
03:48:17 <banix> bye
03:48:26 <banix> #endmeeting