15:02:36 <apuimedo> #startmeeting kuryr
15:02:37 <openstack> Meeting started Mon May  9 15:02:36 2016 UTC and is due to finish in 60 minutes.  The chair is apuimedo. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:02:38 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:02:41 <openstack> The meeting name has been set to 'kuryr'
15:02:59 <apuimedo> Hello! And welcome to yet another kuryr weekly meeting!
15:03:06 <apuimedo> Is there anybody here for it?
15:03:15 <banix> O/
15:03:46 <mspreitz> o/
15:03:51 <banix> That's me with my big head.
15:04:12 <apuimedo> so then, O-/ that's me with the big nose
15:04:23 <apuimedo> thanks mspreitz and banix for joining :-)
15:04:36 <apuimedo> I think Irena will join us in half an hour or so
15:05:05 <apuimedo> #topic repository splitting
15:05:35 <apuimedo> Gal is on holidays but he has submitted patches for creating the subprojects with openstack-infra
15:05:50 <apuimedo> we'll leave openstack/kuryr for the common code
15:07:08 <apuimedo> we'll move libnetwork to openstack/kuryr-libnetwork
15:07:28 <apuimedo> and k8s to openstack/kuryr-kubernetes
15:07:34 <apuimedo> #link https://review.openstack.org/#/c/313981/6/gerrit/projects.yaml
15:07:59 <apuimedo> #info patches submitted for creating kuryr-libnetwork and kuryr-kubernetes repositories as Kuryr subprojects
15:08:15 <apuimedo> Kuryr cores will be the same
15:08:30 <devvesa> o/
15:08:32 <apuimedo> but I expect that we may introduce cores for a single sub project in the future
15:08:40 <apuimedo> devvesa: nice to see you here
15:08:45 <devvesa> thx :)
15:09:12 <apuimedo> any questions about the split?
15:10:04 <apuimedo> ok, moving on
15:10:14 <apuimedo> #topic libnetwork updates
15:10:50 <apuimedo> #info The ovs libnetwork container has been merged into the repo
15:10:55 <apuimedo> #link https://review.openstack.org/#/c/279320/
15:11:06 <apuimedo> Today it will be added for automated builds
15:11:13 <apuimedo> the image is already available in docker hub
15:11:26 <apuimedo> #info docker pull kuryr/libnetwork:ovs
15:12:11 <apuimedo> I hope other vendors join and ask for publishing in kuryr/libnetwork:vendor_name or publish their own in their own repositories
15:12:30 <apuimedo> #info: ping apuimedo if you want to add your vendor automated build
15:12:35 <huikang> hello
15:12:46 <apuimedo> huikang: Hey! Nice to see you too!
15:13:10 <apuimedo> I was just mentioning that our dockerfile has been merged
15:13:16 <apuimedo> and has been tested alongside devstack
15:13:22 <huikang> nice
15:13:46 <apuimedo> that should allow us to move forward testing tests where instead of devstack we use containers from kolla for the networking
15:14:16 <huikang> ok
15:15:20 <huikang> so I will write the gate job to setup kolla and test kuryr
15:15:43 <huikang> apuimedo, how about the keystone v3 issue?
15:16:31 <apuimedo> huikang: ah yes. That's the remaining action :P
15:16:51 <apuimedo> #action apuimedo to check if there is a bug for it and to try to get it sorted out by next week
15:17:21 <apuimedo> we also got the calico binding script merged
15:17:35 <apuimedo> #info project Calico binding script merged
15:17:42 <huikang> For gate job, we can ask kolla keystone generate a temporary v3 token for kuryr.
15:17:57 <apuimedo> #info the container uses uwsgi to provide libnetwork with multiple cores and threads
15:18:18 <apuimedo> #action all packaging to include the uwsgi usage
15:18:55 <banix> Was posting on wrong channel :)
15:18:57 <apuimedo> huikang: well, we should do that as part of kuryr, or you mean as part of of the jenkins shellscript?
15:19:06 <apuimedo> banix: where were you posting?
15:19:10 <huikang> jenkins shellscript
15:19:21 <apuimedo> huikang: we can start with that, yes :-)
15:19:29 <huikang> great, apuimedo
15:19:31 <apuimedo> banix: I see it now
15:20:06 <apuimedo> banix: to the question about testing, the answer is
15:20:23 <apuimedo> I want the post merge test using the automatically built container
15:20:25 <banix> Regarding keystone v3, Hui did you asign the bug to yourself?
15:20:41 <banix> Or is it on Fawad's list?
15:20:58 <huikang> it is still on Fawad's list
15:21:03 <huikang> the same thread
15:21:33 <banix> Ok
15:22:01 <apuimedo> very well
15:22:11 <banix> apuimedo I think huikang is looking into next steps
15:22:15 <apuimedo> banix: any update on the routing/gateway specification?
15:22:44 <banix> I noticed a couple of new api in libnrtwork
15:22:58 <banix> That deal with external connectivity
15:23:24 <banix> Haven't looked closely but I thought we may use that for router stuff
15:23:37 <banix> Anyway will get that sorted out this week
15:23:51 <apuimedo> mmm
15:24:04 <apuimedo> so probably we should use those instead of making our own -o API
15:24:20 <apuimedo> #action banix to look into the new 1.11 libnetwork external connectivity API
15:24:26 <banix> Yeah if possible
15:24:59 <apuimedo> banix: you can do even the impossible :-)
15:25:17 <banix> One more thing
15:25:27 <banix> Regarding libnetwork :)
15:25:57 <apuimedo> banix: what else?
15:26:14 <banix> I'm going to have a patch to allow for user to wait for a join to complete only when port is operational
15:26:57 <huikang> sounds like an opposite to "--no-wait" in OVS
15:27:01 <apuimedo> banix: how often do you see delays due to it?
15:27:09 <banix> Put the bug reference in the wiki
15:27:28 <apuimedo> huikang: that in ovs is for the ovs-vsctl operation?
15:27:43 <banix> Based on past experience under heavy load that can be a problem
15:27:51 <apuimedo> interesting
15:27:52 <apuimedo> :-)
15:27:52 <huikang> yes, apuimedo. what is default behavior?
15:28:22 <apuimedo> huikang: no idea. I'm not very experienced with ovs. I only used it recently with the libnetwork container :P
15:28:55 <huikang> apuimedo, sorry. I was asking the default behavior in kuryr or libnetwork
15:29:06 <banix> Huikang nova has similar options
15:29:08 <apuimedo> #action banix to publish a patch so that network join waits for the port to be operational
15:29:18 <apuimedo> huikang: we use the ovs-vsctl default
15:29:21 <apuimedo> just like nova does
15:29:37 <banix> Pls sea the bug report. On my phone so don't have the link. Sorry.
15:29:38 <apuimedo> anything else about libnetwork?
15:29:48 <apuimedo> I'll check it then
15:29:50 <apuimedo> after the meeting
15:29:52 <apuimedo> thanks banix
15:29:55 <banix> No from me
15:29:59 <huikang> banix, apuimedo, thanks for explaining this
15:30:12 <apuimedo> thanks to you for the help
15:30:17 <apuimedo> #topic kubernetes
15:30:36 <apuimedo> As soon as the repository is created we'll start submitting the patches
15:30:41 <huikang> so I will start creating the shell script to pulling kolla on the gate
15:31:04 <apuimedo> #info kubernetes specs will still leave in the main repository, openstack/kuryr
15:31:17 <mspreitz> BTW, I have something that is for kubernetes but is directly something about libnetwork.
15:31:18 <apuimedo> thanks a lot huikang!
15:31:34 <apuimedo> mspreitz: you got the adapter working?
15:31:38 <huikang> apuimedo, my pleasure
15:32:35 <mspreitz> I have a simple "translate" approach CNI plugin in the Kubernetes contrib repo.  But I wanted to raise the issue of `docker network connect --ip=XXX`.  IBM is going to need that to work, by latching onto a pre-existing but unused Neutron Port.
15:33:03 <apuimedo> aha
15:33:15 <apuimedo> mspreitz: I take it, it does not work right now, does it?
15:33:33 <mspreitz> I think banix said something that implies it does not work right now.
15:33:41 <banix> Should be easy to add
15:33:52 <apuimedo> it's part of the Docker libnetwork API and we should consider it a bug, if it does not work
15:34:10 <apuimedo> I will try it out
15:34:24 <banix> It should work
15:34:30 <apuimedo> #action apuimedo to test "docker network connect --ip="xxx"
15:34:41 <banix> Specifying the ip that is
15:34:43 <apuimedo> banix: I could imagine us not honoring the IP parameter though
15:34:47 <mspreitz> The possibly unexpected part is that what I want is `--ip=W.X.Y.Z` to latch onto a pre-existing Port with that Ip address
15:34:55 <banix> Using pre existing port may not
15:35:09 <banix> Yes exactly
15:35:39 <banix> We honor the up request
15:35:44 <banix> Up
15:35:46 <apuimedo> banix: mspreitz: We'll only do that if the port has not been updated with binding information
15:35:56 <apuimedo> otherwise it opens the door for nasty hijacking, I believe
15:36:03 <apuimedo> (correct me if I'm wrong)
15:36:18 <banix> Correct
15:36:20 <mspreitz> Exactly.  I want `--ip=W.X.Y.Z` to latch onto a pre-existing Port that has that IP address and is not currently in use.
15:36:52 <apuimedo> mspreitz: very well
15:36:54 <apuimedo> we can add that
15:36:59 <banix> We will work through the use case Mike has
15:37:07 <mspreitz> great
15:37:10 <apuimedo> banix: perfect!
15:37:37 <apuimedo> #action mspreitz banix to report a bug for reusing unused ports (finding them by IP)
15:37:56 <apuimedo> mspreitz: any other news from the kubernetes-net-sig camp?
15:38:13 <apuimedo> I have to admit that I've not kept up to date recently with the summit and all
15:38:17 <mspreitz> The spec for ingress filtering is firming up.
15:38:26 <mspreitz> that is what that group has been working on.
15:38:40 <mspreitz> They are now starting to turn some attention to multi-tenancy in the network
15:39:17 <mspreitz> I listed a lot of ways to approach it.  So far it is not clear that there is much that should go into k8s for this, except probably a little generalization to allow some of the techniques.
15:39:26 <mspreitz> More pluggability.
15:39:35 <mspreitz> and parameterization.
15:39:52 <apuimedo> mspreitz: fkautz and I discussed some Docker (and by extension libnetwork) multitenancy. fkautz is trying to push it into docker
15:40:05 <mspreitz> Where was this discussed?
15:40:25 <apuimedo> mspreitz: in the summit work session and last week in #openstack-kuryr
15:40:41 <apuimedo> mspreitz: we'd like to see you in the irc more often :P
15:41:10 <mspreitz> I work odd hours, seems that whenever I am on IRC, nobody else is.
15:41:29 <mspreitz> Anyway, is that proposal written down anywhere?
15:41:31 <apuimedo> mspreitz: you should set up an IRC bouncer like I did last week :-)
15:41:46 <apuimedo> mspreitz: I guess fkautz will update us when it is
15:41:50 <mspreitz> Or we should switch to Slack?
15:42:06 <apuimedo> mspreitz: not possible for openstack projects :P
15:42:25 <mspreitz> then someone should tell me, offline, what an IRC bouncer is.
15:42:54 <apuimedo> #info: devvesa is preparing CI jobs for coverage of the kuryr-kubernetes code that we'll be able to set up when we push the code in the new repo
15:43:07 <apuimedo> mspreitz: sure, I'll send you an email ;-)
15:43:17 <mspreitz> thanks
15:43:35 <apuimedo> mspreitz: please, paste here the link to your cni->libnetwork translator in the k8s repo
15:43:42 <apuimedo> for the meeting logs
15:43:55 <mspreitz> https://github.com/kubernetes/contrib/tree/master/cni-plugins/to_docker
15:44:07 <apuimedo> thanks!
15:44:13 <apuimedo> #link https://github.com/kubernetes/contrib/tree/master/cni-plugins/to_docker
15:44:13 <mspreitz> There is a PR in progress on that, https://github.com/kubernetes/contrib/pull/845
15:44:21 <apuimedo> #link https://github.com/kubernetes/contrib/pull/845
15:44:30 <apuimedo> anything more on kubernetes from anybody?
15:44:53 <apuimedo> mspreitz: I trust you saw my answer on the mailing about the service load balancer usage
15:45:23 <mspreitz> Yes, thank you.
15:45:31 <apuimedo> :-)
15:45:31 <irenab> apuimedo: sorry I have network issues today, will have to catch up from the log
15:45:44 <apuimedo> irenab: me too. I'm tethering right now
15:45:56 <apuimedo> (and that after having gone to my parents for better wifi)
15:46:09 <apuimedo> #topic open discussion
15:46:19 <apuimedo> Anybody else has any other topic to bring up?
15:46:43 <apuimedo> #info We should consider a mid-cycle virtual sprint at some point
15:46:50 <apuimedo> We'll update on the mailing list
15:46:57 <apuimedo> about that
15:47:07 <banix> Meeting time !
15:47:23 <banix> Do we need changing this time?
15:47:43 <apuimedo> also, banix, what do you think about doing the first release after we support the external connectivity (if possible) and have successfully moved to the new repo?
15:47:47 <mspreitz> There was nobody around for last week's meeting.
15:48:12 <mspreitz> Does anybody know where Weave and Calico stand wrt libnetwork and external connectivity?
15:48:16 <banix> apuimedo sounds good
15:49:24 <apuimedo> mspreitz: the alternate time is far too late for me to join
15:49:39 <apuimedo> mspreitz: I think that they provide external connectivity always by default
15:49:43 <apuimedo> but I may be wrong
15:49:53 <mspreitz> I am not criticizing anyone, just suggesting the time may not be good.
15:50:15 <mspreitz> I think Weave's libnetwork plugin does not provide external connectivity
15:50:28 <mspreitz> At least, not in the normal way.  They have special ways of their own.
15:50:50 <apuimedo> mspreitz: well, it's up for the people in those time zones to propose a new time
15:51:09 <apuimedo> I hope they do
15:51:27 <apuimedo> and participate more, otherwise I don't see a lot of good from having the alternate meeting
15:51:59 <apuimedo> mspreitz: being that it's all routed I expect that they just propagate some bgp to the edge
15:52:05 <apuimedo> but I've never really tried it
15:52:17 <apuimedo> mspreitz: you should ask neiljerram in #openstack-kuryr channel
15:52:25 <apuimedo> he's from kuryr
15:52:30 <apuimedo> sorry, calico
15:52:33 <apuimedo> any other topic?
15:53:12 <banix> I think we can leave storage for when gal comes back?
15:53:31 <banix> Do we have anyone from fushi here?
15:53:45 <banix> Or magnum storage?
15:54:12 <irenab> Can you share if there were any agreements regaridng the Newton features at the summit? I mean except for k8s
15:54:38 <apuimedo> banix: I think we'll have to take to the mailing list about that
15:54:41 <apuimedo> :P
15:54:55 <banix> Sure
15:55:04 <irenab> in case someone has etherpads of the Thursday/Friday sessions, it can be great
15:55:39 <apuimedo> banix should have the links
15:55:44 <apuimedo> I was lazy to open the laptop
15:55:47 <apuimedo> :P
15:55:58 <irenab> I wanted to ask to move the Monday meeting one hour before or after if it works for other people
15:56:20 <banix> Irenab please let me find and share by email
15:56:28 <irenab> banix: thanks
15:56:56 <apuimedo> irenab: sounds good to me
15:57:03 <banix> Moving either way is ok with me
15:57:06 <apuimedo> irenab: please, start an ml thread about it
15:57:12 <apuimedo> thanks all for joining
15:57:24 <banix> Thx
15:57:24 <apuimedo> have a nice and productive week!
15:57:30 <apuimedo> #endmeeting