03:00:19 <banix> #startmeeting kuryr
03:00:20 <openstack> Meeting started Tue Dec 15 03:00:19 2015 UTC and is due to finish in 60 minutes.  The chair is banix. Information about MeetBot at http://wiki.debian.org/MeetBot.
03:00:21 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
03:00:23 <openstack> The meeting name has been set to 'kuryr'
03:00:42 <banix> Hi everybody. Who do we have present for the Kuryr meeting?
03:00:51 <baohua> baohua is here :)
03:01:02 <tfukushima> o/
03:01:20 <banix> hi baohua
03:01:25 <banix> and tfukushima
03:01:42 <banix> let’s wait a minute or so to see if Vikas and others join in
03:01:54 <baohua> sure
03:03:02 <banix> agenda at #link https://wiki.openstack.org/wiki/Meetings/Kuryr#Agenda
03:03:03 <tfukushima> #link today's agenda https://wiki.openstack.org/wiki/Meetings/Kuryr#Meeting_December_15.2C_2015
03:03:19 <tfukushima> Ooops, sorry.
03:03:26 <banix> no problem. thanks.
03:03:31 <banix> lets get started.
03:03:37 <banix> #topic IPAM
03:04:10 <banix> I have looked at the patches and they are coming along well. There are 3 that I believe we need to get the solution working:
03:04:44 <banix> #link https://wiki.openstack.org/wiki/Meetings/Kuryr#Agenda
03:05:03 <banix> and #link https://wiki.openstack.org/wiki/Meetings/Kuryr#Agenda
03:05:21 <banix> sorry
03:05:30 <vikasc> banix, https://review.openstack.org/#/q/owner:vikaschoudhary16+status:open,n,z
03:05:45 <banix> thanks vikasc
03:05:56 <banix> #link  https://review.openstack.org/#/q/owner:vikaschoudhary16+status:open,n,z
03:06:31 <banix> so for testing end to end we need these 4 patches vikasc?
03:06:39 <vikasc> yes banix
03:07:29 <banix> Let us review these as soon as possible
03:07:40 <baohua> +1
03:07:43 <banix> #action review patches for IPAM #link  https://review.openstack.org/#/q/owner:vikaschoudhary16+status:open,n,z
03:07:44 <vikasc> yes.. i will appreciate that.
03:07:56 <banix> vikasc: anything you want to discuss?
03:08:04 <vikasc> nothing from my side
03:08:19 <vikasc> we will discuss more after review comments
03:08:27 <banix> #topic binding
03:08:33 <banix> thanks vikasc
03:08:47 <banix> a few minor (and some not so minor) issues here
03:09:27 <banix> I am going to updat this patch: #link https://review.openstack.org/#/c/254417/
03:09:46 <banix> I will follow Irena’s suggestion and this should be straightforward
03:10:17 <banix> the 2nd item in this topic is unbind for OVS
03:10:56 <banix> I have to look into it; I thought there is nothing to be done there but I will take a second look
03:11:21 <banix> I will be submitting the binding for hybrid OVS shortly
03:11:42 <vikasc> thanks banix
03:12:42 <baohua> #idea #agreed Merging existing binary implementation into separate binding lib is a good idea
03:12:48 <banix> doing that piece and looking into what we need for Linux Bridge binding, If we need to spend the time and see if we can have a library for doing this and we can load the bindings as needed
03:13:29 <baohua> +1
03:13:35 <vikasc> +1
03:13:59 <banix> #action banix to add ovs_hybrid binding
03:14:09 <banix> #action banix to look into LB binding
03:14:28 <banix> #action banix to look into more efficient way of doing binding/unbinding
03:14:47 <banix> anything else to discuss wrt binding here?
03:15:15 <vikasc> i think we covered most of it
03:15:34 <banix> #topic Capabilities
03:16:12 <banix> Anybody knows what this refers to?
03:16:37 <baohua> no idea...
03:17:00 <vikasc> this is related to scope of network driver, global or local
03:17:11 <tfukushima> I think Toni added it and it's about the privilege thing related to the binding/unbinding.
03:17:30 <banix> vikasc: and tfukushima please ellaborate
03:18:08 <tfukushima> He mentioned he knew how to deal with the permission problem we're experiencing on creating and deleting veth pairs.
03:18:52 <banix> tfukushima: ok thanks, let me create an action item for Toni :)
03:18:52 <tfukushima> I mean the problem pyroute2 requires root for manipulating interfaces through rtnetlink.
03:18:59 <banix> understood
03:19:11 <vikasc> thanks tfukushima
03:20:00 <banix> #action apuimedo to propose solution for avoiding the need to run pyroute2 as root
03:20:16 <banix> #topic Testing
03:20:42 <banix> So this is a really big topic and we want to get in the best possible shape in this cycle
03:21:08 <banix> Gal has a series of patches for improving Kuryr gate jobs
03:21:52 <banix> HE also sent out an email to openstack-dev wrt to use of Rally for testing at the gate
03:22:09 <banix> I would like to emphasize the need for more unit and functional testing
03:22:35 <banix> tfukushima: has done most of the work. Do you want to add anything here Taku?
03:23:43 <banix> I believe we have baohua who has colunteered to work in this area (and Vikas too but he may be busy until IPAM work i done)
03:23:56 <tfukushima> Uh, not really I'm looking forward to seeing the nice infrastructure for testing is coming.
03:24:34 <baohua> Yes, would like to help fix the testing work
03:24:39 <banix> tfukushima: on ths unit test from, does it may sense to analyze our current coverage and try to address any existing gaps?
03:24:53 <banix> unit test front
03:25:04 <banix> make sense
03:25:05 <banix> the
03:25:17 <banix> boy, who is typing here? :)
03:26:10 <tfukushima> The current unit tests are covering limited cases. It'd be nice to cover all cases but that'd require some dull works.
03:26:44 <banix> baohua: should I assign a work item to you? :)
03:26:50 <tfukushima> And they're going to be changed by Vikas's IPAM patches. So let's merge his patches first and see how we can go on.
03:27:31 <banix> tfukushima: sure but I think this is a very important area
03:27:32 <vikasc> tfukushima, +1
03:27:41 <baohua> banix, sure, first i would like to see what framework we need for unit test and functional test.
03:28:11 <banix> baohua: for unit tests we already have a framework with some tests
03:28:32 <baohua> great then, i think this can be more efficient if we choose the right tools.
03:28:43 <banix> #action to look into extending the coverage of unit testing after IPAM patches merge
03:29:15 <banix> with respect to functional testing where do we we stand?
03:29:59 <banix> it is perhaps worth looking into in tree functional tests in Neutron
03:30:10 <baohua> i guess it's some functional capability, like create/delete, high-level operations etc.
03:30:39 <banix> #action to look into a framework for functional testing for Kuryr
03:30:59 <banix> baohua: added another action item to you :)
03:31:10 <baohua> sure :)
03:31:23 <banix> ok let’s move on
03:31:38 <banix> #topic Docker Swarm
03:32:30 <banix> I have swarm working with Kuryr as the docker network plugin.
03:32:42 <banix> The integration is simple and straightforward
03:33:14 <tfukushima> Oh, that's nice. With the IPAM patches applied?
03:33:56 <banix> I hopefully will record a demo and the steps taken but the basic integration is painless. No IPAM patches. Relying on the existing subnet pools, etc.
03:34:18 <tfukushima> Ok, good to know.
03:34:31 <vikasc> banix,  nice!!
03:34:52 <baohua> nice work!
03:35:01 <tfukushima> I don't know how scalable Kuryr is at this point though. :-)
03:35:46 <banix> tfukushima: yes indeed; I hope as we extend our testing, that leads to performance and scalability testsing so we can address any issues
03:36:07 <banix> We don’t have much info about libnetwork scalability either. Do we?
03:37:03 <tfukushima> The overhead of Kuryr is not small actually comparing to the regular bridge plugin. I didn't compare to the overlay plugin legitimately.
03:38:13 <tfukushima> Well, see more and we'll have some concrete figures.
03:38:36 <banix> tfukushima: it will be nice to have some figures to discuss at some point; Taku, do you have any mubers you want to share may be next week?
03:39:14 <kexiaodong> scalable means multi-node, so we should install Kuryr in multi-node, each compute node should have one kuryr.
03:39:54 <tfukushima> Hi kexiaodong, yes that's true.
03:40:04 <banix> kexiaodong: yes that is correct; we are talking about multi-node installations
03:41:01 <banix> shall we move on?
03:41:34 <banix> #topic packaging
03:41:40 <tfukushima> Yes, I don't have the numbers that I can give at this point.
03:42:06 <banix> tfukushima: sure
03:42:42 <banix> I don’t have anything on this topic; Does anyone present have something to discuss?
03:43:10 <banix> I think we will start thinking about packaging as we aim for releasing by the end of this cycle
03:43:25 <baohua> any one can share the current status?
03:43:33 <tfukushima> Me neither. Is it going to be .rpm and .dpkg?
03:44:25 <banix> I think we can postpone discussion to next week
03:44:37 <banix> #topic Kolla
03:44:58 <banix> We have Hui Kang on this task
03:45:29 <banix> He started working on this a while back but since our configuration piece was not in place, couldn’t get far.
03:46:03 <banix> I believe he is going to start working on this shortly and since he has done this for other projects I expect we will have good progress shortly
03:46:31 <banix> #action banix to follow up with Hui wrt the Kuryr Kolla integration
03:46:39 <tfukushima> Nice, this is going to be nice. But we need to think about the root privilege requirement for pyroute2, which could be solved by "capability" apuimedo mentioned.
03:46:57 <tfukushima> Otherwise the privilege mode of Docker would be used.
03:47:13 <banix> tfukushima: agreed
03:47:37 <banix> I hope to see “a” version working as we iron the details out :)
03:48:29 <banix> #topic Magnum
03:48:56 <banix> Do we have any updates wrt Neutron trunk ports?
03:49:34 <tfukushima> I haven't heard about it.
03:50:15 <baohua> neither from me
03:50:22 <banix> is fawadkhaliq around?
03:50:35 <fawadkhaliq> hi banix
03:50:40 <banix> Hi Fawad
03:50:50 <banix> I was about to assign an action item to you :)
03:50:59 <banix> now that you are here, let me ask you first
03:51:07 <fawadkhaliq> oh :-)
03:51:11 <fawadkhaliq> tell me
03:51:26 <banix> do you knw the state of trunk port effort, if not would you be willing to follow on and give us an update next week?
03:52:05 <fawadkhaliq> I am actually not aware of the latest update but definitely. let me find out and provide an update.
03:52:16 <banix> thanks!
03:52:23 <fawadkhaliq> so please go ahead and assign an action item to me and I will update you guys
03:52:57 <banix> #action fawadkhaliq to follow up on trunk ports (and possibly similar) effort in Neutron
03:53:15 <banix> #topic Open Discussion
03:53:43 <banix> Anyone wnats to bring up any issue to discuss?
03:54:22 <banix> To summarize, our work items:
03:54:39 <banix> 1)  IPAM remains the main priority. Please review Vikas’ patches
03:55:01 <banix> 2) Testing: Help get the gate jobs reviewed
03:55:24 <banix> 3) Binding: add new bindings
03:55:48 <banix> 4) Work out the root access requirement
03:56:07 <banix> 5) Get  the Kolla integration underway
03:57:06 <banix> All right folks. Thanks for participating. See you all on the Kuryr IRC channel till next week
03:57:09 <tfukushima> BTW, Magnum work might require the mixed environment with the VM instances and containers. Please review my spec.
03:57:15 <tfukushima> https://review.openstack.org/#/c/254005/
03:57:25 <tfukushima> #link existing neutron network spec https://review.openstack.org/#/c/254005/
03:57:27 <fawadkhaliq> tfukushima: will do
03:57:47 <tfukushima> I was assigned to the internal urgent task so my response will be slow this week.
03:57:59 <banix> thanks taku
03:58:01 <banix> sure
03:58:16 <banix> #endmeeting