14:02:27 <mestery> #startmeeting networking 14:02:27 <openstack> Meeting started Tue Jan 27 14:02:27 2015 UTC and is due to finish in 60 minutes. The chair is mestery. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:02:28 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:02:30 <openstack> The meeting name has been set to 'networking' 14:02:38 <mestery> #link https://wiki.openstack.org/wiki/Network/Meetings Agenda 14:03:09 <mestery> This may be a light meeting today, the main focus I'd like to have is on Kilo-2 work and some time on plugin decomposition 14:03:12 <mestery> #topic Announcements 14:03:24 <mestery> Kilo-2 is in 2 weeks on February 5 14:03:28 <mestery> #link https://wiki.openstack.org/wiki/Kilo_Release_Schedule 14:03:37 <mestery> #link https://launchpad.net/neutron/+milestone/kilo-2 14:04:02 <amotoki> hi 14:04:04 <mestery> We'll spend some time on specific BPs there later in the meeting. 14:04:28 <mestery> That's all I had for announcements on the agenda this week. Does anyone have anything else? 14:05:33 <mestery> #topic Bugs 14:05:43 <mestery> enikanorov_ enikanorov__: Hi there 14:05:54 <enikanorov_> mestery: hi 14:06:18 <enikanorov_> mestery: may we move bugs section to the end of the meeting? 14:06:23 <enikanorov_> sorry about that 14:06:27 <mestery> enikanorov_: Sure! We'll come back to it, no worries. 14:06:36 <mestery> #topic Docs 14:06:41 <mestery> emagana: Hi there! 14:06:55 <emagana> mestery: hello! 14:08:03 <emagana> Networking guide is going well. We have a legacy implementation 14:08:05 <emagana> #link https://github.com/ionosphere80/openstack-networking-guide/blob/master/scenario-legacy-ovs/scenario-legacy-ovs.md 14:08:32 <mestery> emagana: neat! 14:08:32 <emagana> The idea is to cover what nova-network provides but with overlays network besides VLANs 14:09:03 <mestery> emagana: That's a pretty good guide, just scrolling through it now. 14:09:18 <emagana> I encourage everybody to provide feedback now on this github account or once it is move to gerrit, which I will provide the link but it is not there yet 14:09:55 <mestery> #info Encourage everyone to provide feedback on the networking guide in github for now until it moves to gerrit. 14:10:05 <mestery> emagana: Is there a deadline for this feedback? 14:10:25 <emagana> Just a reminder for everybody, in order to facilitate the documentation for the people who are not familiar with the Docs tools, this .md format is much more easier to update ;-) 14:10:56 <emagana> mestery: No yet! But our target is to complete the guide by mids February 14:11:25 <mestery> #info Target is to complete networking guide by mid February 14:11:26 <mestery> thanks emagana! 14:11:28 <amotoki> emagana: how to feedback? IRC channel? docs ML? 14:11:30 <emagana> mestery: but I do not think we have a hard deadline. The DVR side has been moving along nicely 14:11:49 <ihrachyshka> pull requests ;) 14:11:55 <mestery> ihrachyshka: lol 14:12:09 <emagana> amotoki: it depends, if you want to provide changes to the document, just follow the typical git process and you name will be included in the credits of the guide 14:12:36 <emagana> if you have few comments, just added directly in git or wait for the official gerrit commit 14:12:47 <mestery> amotoki: Do you have plans to present this guide at the Ops Summit in Philly? May be a good idea to get feedback there as well. 14:12:55 <mestery> emagana: 14:12:55 <mestery> Sorry. not amotoki :) 14:13:35 <emagana> mestery: I am planing to attend it but I did not think about present it, good idea! I should! 14:14:05 <mestery> emagana: Yes, would be good to get feedback. Even better if you can get the ops there to collaborate on it ;) 14:14:17 <emagana> basicallym you will see all the scenarios that we are covering here: 14:14:19 <emagana> #link https://github.com/ionosphere80/openstack-networking-guide 14:15:04 <mestery> emagana: Looks good 14:15:20 <emagana> sc68cal is working on a provider network scenario as well ;-) 14:15:30 <emagana> mestery: no more from my side! 14:16:07 <mestery> emagana: One thing I wanted to bring up was that the LBaaS folks found some issues with docs around LBaaS (dougwig and blogan). 14:16:11 * mestery is looking for the email link 14:16:25 <mestery> #link http://lists.openstack.org/pipermail/openstack-dev/2015-January/054827.html 14:16:34 <mestery> emagana: Just noting it here to keep you aware in case they need help 14:16:40 <dougwig> the v2 api docs somehow got released early. i think a bug got filed, and anne was looking into it. 14:16:55 <emagana> mestery: we took care of that.. well annegentle did! 14:17:00 <mestery> excellent! 14:17:25 <mestery> OK, anything else on Docs from anyone? 14:18:18 <mestery> OK moving on 14:18:26 <mestery> #topic Plugin Decomposition Status Update 14:18:42 <mestery> I'm happy to see a lot of work going on here! Cheers to those who have already started and are making progress! 14:19:09 <mestery> Next week, I'm going to have armax present some stats on where we're at with regards to overall stats of who's started, etc. 14:19:25 <mestery> For now, if anyone has any questions on this, this is a good time to ask for help. 14:19:55 <mestery> The ODL ML2 driver is now fully decomposed, as well as the midokura driver. 14:19:59 <mestery> Sukhdev: How is the Arista driver coming along? 14:20:09 <Sukhdev> mestery: I am very close 14:20:20 <mestery> Sukhdev: Excellent! 14:20:36 <Sukhdev> armax has been helping with UT issue 14:20:45 <mestery> Kudos to armax for working long hours in helping people over the past few weeks. 14:20:51 <ihrachyshka> speaking of arista, we should consider bug fixes that were discussed recently for review on master to be able to backport 14:21:19 <Sukhdev> ihrachyshka, mestery : I added an item to the agenda to discuss this 14:21:32 <mestery> Sukhdev: Lets discuss that here maybe? Seems to fit right in. 14:21:46 <Sukhdev> mestery: sounds good 14:21:53 <sean-k-mooney> are there currently any example of how to import the ml2 mechisium driver into neutron? ie the in tree part 14:22:00 <Sukhdev> so, here is the issue - 14:22:13 <mestery> sean-k-mooney: We'll come back to that, hold on for a minute. 14:22:21 <Sukhdev> While we are working on decomp work, we have patches that need backports 14:22:51 <Sukhdev> ihrachyshka: wrote a very good summary describing the issue and solutions 14:22:54 <ihrachyshka> hasn't https://review.openstack.org/148745 clarified the process for backports and doesn't it say that we should allow bug fixes to go into master even during spin-off? 14:23:17 <mestery> ihrachyshka: ++, that's the link I was looking for 14:23:21 <Sukhdev> ihrachyshka: correct - but, I still see the patches blocked 14:23:22 <mestery> #link https://review.openstack.org/148745 14:23:33 <mestery> Sukhdev: We should unblock them then. Do you have the links? 14:23:43 <ihrachyshka> Sukhdev, that's why I encourage people to unblock reviews and proceed with them :) 14:23:44 <mestery> ihrachyshka: When reviewing these on stable, lets make sure to include https://review.openstack.org/148745 in the review. 14:23:45 <Sukhdev> yes - give me a sec 14:24:01 <Sukhdev> https://review.openstack.org/#/c/147976/ 14:24:45 <Sukhdev> https://review.openstack.org/#/c/148369/ 14:24:54 <Sukhdev> those are the two 14:25:14 <mestery> Thanks Sukhdev 14:25:46 <Sukhdev> juno stable deadline is in few days - if cores can get these approved - then ihrachyshka can help me get them backported 14:25:48 <mestery> Sukhdev: I'll work with armax and ihrachyshka to unblock these for you. 14:26:02 <Sukhdev> mestery: Thank you so much 14:27:08 <mestery> sean-k-mooney: You had a question as well now around importing? 14:27:16 <emagana> Sukhdev: I was reviewing on of them, let me know if you need one more reviewer ;-) 14:27:27 <dougwig> mestery: ping me if you need an extra reviewer on those. 14:27:38 <Sukhdev> emagana: thanks - I will ping you 14:27:44 <mestery> dougwig: Please go ahead now if you want, we need armax on the second one but the first one is open. 14:27:46 <sean-k-mooney> yes when the dirver is split out. is there an example of the in tree part that import an ml2 mechisium dirver 14:28:18 <mestery> sean-k-mooney: Yes, let me grab a link. 14:28:49 <mestery> #link https://github.com/openstack/neutron/blob/master/neutron/plugins/ml2/drivers/mechanism_odl.py 14:28:51 <dougwig> sean-k-mooney: neutron/plugins/ml2/drivers/mechanism_odl.py 14:28:52 <sean-k-mooney> mestery: thanks 14:28:55 <mestery> sean-k-mooney: ^^^ Like that? 14:29:43 <mestery> OK, anything else on plugin decomposition? 14:29:59 <Sukhdev> mestery: I have one on ODL- decomp 14:30:13 <mestery> Sukhdev: Please go ahead 14:30:45 <Sukhdev> I notice networking_old is not registered for import - is it by design? 14:31:01 <rprakash> Hi prakash here can any one tell me if I want to develop anew plugin for ONOS, which one shall I pick to start for a BP 14:31:09 <Sukhdev> In other words, I can not do sudo python install networking_odl 14:31:29 <mestery> Sukhdev: Yes, you have to manually pull it and install it now, we haven't pushed a release to pypi yet 14:31:37 <rprakash> or can I add ONOS to ODL like OZC2ODL does? 14:31:48 <mestery> rprakash: For a new plugin/driver, you need to file a wishlist bug and keep a thin shim in neutron, the rest needs to be somewhere else. 14:31:56 <Sukhdev> but, that is a requirement, right? we all have to do it? 14:32:00 <dougwig> well, you could in theory do "pip install -e git+{url}" straight from github, right? 14:32:05 <mestery> rprakash: http://specs.openstack.org/openstack/neutron-specs/specs/kilo/core-vendor-decomposition.html 14:32:09 <mestery> rprakash: That has instructions 14:32:15 <mestery> dougwig: Yes 14:32:25 <mestery> Sukhdev: Right 14:32:36 <rkukura> dane_leblanc: Does the Cisco MD decomposition depend on https://blueprints.launchpad.net/neutron/+spec/ml2-hierarchical-port-binding? 14:32:38 <rprakash> thanks will follow the link thanks 14:33:04 <mestery> rprakash: Once you've looked at that one, please find myself or armax in #openstack-neutron with questions. 14:33:24 <dane_leblanc> rkukura: No, the first pass of decomp should be on the current upstream Nexus plugin, that doesn't have hierarchical port binding 14:33:42 <rkukura> dane_leblanc: Thanks. 14:34:08 <rprakash> Thanks will connect back to team 14:34:09 <dane_leblanc> rkukura: But Arvind's work will depend on hierarchical port binding. 14:34:16 <mestery> rprakash: Great! 14:34:42 <sean-k-mooney> mestery: may i ask one more question regarding the import. is there a requirement file that specifies which verion of the networking_odl libary your are importing. 14:35:25 <mestery> sean-k-mooney: There is not, we haven't released networking-odl yet, so it's just using current master from git. 14:35:52 <sean-k-mooney> ok thanks that makes sense 14:36:00 <mestery> sean-k-mooney: Excellent! 14:37:32 <Sukhdev> mestery: I am confused - what do you mean by you OLD has not been released? 14:37:45 <Sukhdev> Aren't people using it? 14:37:47 <mestery> Sukhdev: We haven't released it on pypi yet. 14:37:55 <mestery> Meaning, a released version. 14:38:23 <mestery> Sukhdev: It's "released" in that it's public, but there is no version 0.1 released yet :) 14:38:37 <Sukhdev> mestery: But, if someone wants to install from (say devstack) without requirements.txt file, how does devstack know what to pull? 14:39:13 <mestery> Sukhdev: The ODL devstack work will pull networking-odl itself. 14:39:30 <mestery> Sukhdev: Due to the pluggable devstack work, we have to pull that out of devstack and make it pluggable, which we're doing now. 14:39:41 <dougwig> you could also put FOO=$(pip install git+blah) into your local.conf in the right section. 14:39:45 <Sukhdev> mestery: Oh so, for Arista MD, do I have to modify the devstack as well? 14:40:04 <Sukhdev> mestery: I thought if I added requiremnts.txt file, that will do the trick, no? 14:40:05 <mestery> Sukhdev: There are a few ways around this, dougwig proposed another one right there :) 14:40:25 <mestery> Sukhdev: That would do it too, but we didn't want to add that into requirements.txt for all these new external sources. 14:40:40 <mestery> armax documented that in the spec and I believe there was a discussion on that earlier as well. 14:40:51 <mestery> Sukhdev: Lets circle back with armax when he's online for that one. 14:41:04 <Sukhdev> mestery: I am using what dougwig proposed - but, I do not believe it is a scalable and long term solution, no? 14:41:23 <Sukhdev> mestery: I agree - I will bring it up with armax 14:41:26 <mestery> Sukhdev: Correct 14:42:25 <mestery> Great discussions here! 14:42:26 <dougwig> it scales fine, it's just not terribly user friendly or allowing of self-discovery, which i'm guessing is what you meant. 14:42:50 <Sukhdev> dougwig: that is what I meant - sorry :-) 14:42:55 <mestery> We'll leave this spot in the agenda going forward so each week we can work with people who are decomposing their plugins and have questions. 14:42:57 <mestery> But lets move on now. 14:43:02 <mestery> #topic nova-network to neutron migration 14:43:06 <mestery> anteaya: Hi! 14:43:08 <anteaya> o/ 14:43:16 <anteaya> so we are meeting 14:43:19 <anteaya> #link http://eavesdrop.openstack.org/meetings/nova_net_to_neutron_migration/ 14:43:25 <mestery> Cool! 14:43:31 <anteaya> and we have a spec which is a two parter 14:43:34 <anteaya> here is part one 14:43:42 <anteaya> #link https://review.openstack.org/#/c/147723/ 14:43:47 <anteaya> we are getting close 14:44:03 <mestery> Awesome! 14:44:04 <anteaya> basically this patch sets expectations of the users of the migration work 14:44:39 <anteaya> one thing that was clear yesterday at the nova mid-cycle is that we need deployers to know that if they are holding out for a push button solution, they will be disappointed 14:44:56 <anteaya> this was captured in the community impact section of the latest patchset 14:45:18 <anteaya> I really would like to push for getting this first patch merged by the end of this week 14:45:48 <anteaya> the child patch to this includes the migration path and nova folks want to see some proof of concept code before they sign off on the spce 14:45:51 <anteaya> which is fair 14:46:00 <mestery> anteaya: Makes sense 14:46:02 <anteaya> jlibosva has a db migration wip patch up 14:46:18 <anteaya> and obondarev has told me he will get a wip proxy patch up on gerrit soon 14:46:41 <anteaya> there is some resistance here, mostly I feel due to his concern about putting up something in wip 14:46:56 <anteaya> but others need to see the work to be able to offer opinons 14:47:06 <anteaya> so obondarev heard my concerns and will put something up 14:47:16 <anteaya> I assured him he would feel supported in doing so 14:47:28 <mestery> anteaya: I love the wording in the "Community Impact" section of gus's spec, nicely done. 14:47:30 <anteaya> emagana has said he will lead the docs effort 14:47:37 <anteaya> mestery: thanks 14:47:56 <anteaya> one of the issues is we still don't have anyone doing docs who can attend meetings 14:48:07 <anteaya> so would be great to have someone here to help emagana 14:48:22 <anteaya> someone named chris emailed me to say he wants to help with docs 14:48:29 <anteaya> speak up if you are on irc chris 14:48:44 <anteaya> also the cells issue while not in scope for the migration work 14:48:51 <anteaya> is still an issue for cern 14:49:01 <anteaya> I am talking with alaski to see what we can do here 14:49:20 <anteaya> and he has told me he would really like a point person in neutron to understand cells impact 14:49:42 <anteaya> the problem being that neutron doesn't talk cells so nova cells users can't use neutron 14:49:48 <mestery> cern uses cells? Talk about living on the edge ;) 14:49:49 <anteaya> so we need to address this gap 14:49:55 <anteaya> cern uses cells 14:49:59 <mestery> cool 14:50:00 <ajo> yes, they do 14:50:01 <emagana> anteaya: it will be great to have someone else because I can't attend the IRC meetings, are too early for me 14:50:18 <anteaya> cells do two things I understand, load balance the scheduler and isolate against failure 14:50:35 <anteaya> so we need to find a neutron person willing to work with alaski on this gap 14:50:44 <anteaya> emagana: yes Understood 14:50:56 <mestery> anteaya: Thanks for the update here. 14:50:58 <anteaya> emagana: obondarev said he would work with you this week on getting started on docs here 14:51:07 <mestery> There is one more item I'd like to hit on the agenda in a few minutes here yet. 14:51:16 * anteaya stops 14:51:18 <anteaya> thanks 14:51:30 <enikanorov_> mestery: bugs? ;) 14:51:33 <mestery> anteaya: Thanks for the great update here! I encourage folks to attend the migration meetings and work with anteaya. 14:51:40 <mestery> enikanorov_: Two items I guess :) 14:51:47 <mestery> enikanorov_: You did ask to go last today :P 14:51:48 <anteaya> thank you, and to all who particiapte 14:51:53 <enikanorov_> haha :) 14:51:54 <enikanorov_> yep 14:51:55 <mestery> #topic Client Extensibility 14:51:55 <mestery> roaet: You're up :) 14:51:58 <roaet> o/ 14:52:01 <roaet> I'll go fast. 14:52:10 <roaet> This is in regard to the client extensibility bp 14:52:15 <mestery> #link https://blueprints.launchpad.net/python-neutronclient/+spec/extensible-neutronclient 14:52:22 <roaet> We have a change here: https://review.openstack.org/#/c/148318/ 14:52:25 <mestery> #link https://review.openstack.org/#/c/148318/ 14:52:37 <roaet> It currently only supports shell extensions and models after the nova client extensions 14:52:43 <mestery> amotoki: I'd really like you to review this closely, as you're our most prolofic client reviewer 14:52:58 <roaet> This adds 'commands' and what not to the client. Work can be done to add 'commands' to the library if needed. 14:53:04 <ajo> mestery, roaet , looking into the future, shouldn't that work be done in the openstack-sdk 14:53:05 <ajo> ? 14:53:25 <amotoki> roaet: will look at it 14:53:27 <mestery> ajo: That's a good point, although roaet's work solves their problem for the current :) 14:53:33 <roaet> ajo: ^ 14:54:01 <ajo> sure, but may be they want to implement extension mechanism there too, or will face the same problem sooner or later :) 14:54:18 <mestery> ajo: Valid point. roaet, I'd encourage you to look there too. 14:54:31 <roaet> ajo: absolutely. I'm guessing that'd be a part of the library aforementioned library portion. 14:54:32 <amotoki> perhaps we need to explore both ways (openstack-sdk/openstackclient and existing neutronclient) 14:54:34 <mestery> Also, I encourage folks to review what roaet has here in the client. amotoki, I'll sync with you on this as well. 14:54:40 * dougwig notes an eery similarity between neutronclient/openstackclient and nova-network/neutron 14:54:50 <amotoki> but I can't say more because I haven't look into the detail yet. 14:54:56 <mestery> OK, thanks roaet ! 14:54:59 <roaet> thanks 14:55:06 <mestery> Lets leave 5 minutes for bugs, there is a meeting right after this one so we have to end on time :) 14:55:08 <mestery> #topic Bugs 14:55:10 <mestery> enikanorov_: Hi! 14:55:13 <ajo> :D 14:55:14 <enikanorov_> hi 14:55:16 <enikanorov_> :) 14:55:24 <enikanorov_> here's a short list of bugs to discuss 14:55:25 <mestery> I guess we get to highlight only a few bugs today ;) 14:55:32 <enikanorov_> https://bugs.launchpad.net/neutron/+bug/1414812 14:55:34 <russellb> related to bugs, this patch fixes hyper-v (sorry) https://review.openstack.org/#/c/150373/ 14:55:46 <enikanorov_> hmm, does bot work? 14:55:56 <mestery> Maybe it's busted :) 14:56:10 <enikanorov_> ok, this one is related to dvr-job, armax is handling it 14:56:25 <mestery> enikanorov_: It's in the queue I think 14:56:29 <russellb> yep 14:56:38 <enikanorov_> second one "Delete subnet can fail for SLAAC/DHCP_STATELESS with 409" https://bugs.launchpad.net/neutron/+bug/1414199 14:56:46 <enikanorov_> with a patch on review from salv-orlando 14:57:14 <enikanorov_> back to https://bugs.launchpad.net/neutron/+bug/1382064 - "Failure to allocate tunnel id when creating networks concurrently" 14:57:24 <enikanorov_> we tried to fix that with changing transaction isolation level 14:57:33 <enikanorov_> which created more issues that solved 14:57:53 <enikanorov_> so now there's another patch on review, that employes retry approach similar to nova's 14:58:03 <enikanorov_> russellb: welcome to review the patch 14:58:24 <enikanorov_> russellb: btw, i've rebased your patch on default api_workers on that fix 14:58:30 <russellb> yeah saw that, thanks 14:58:34 <ihrachyshka> with my red hat on, I'd like folks to consider https://bugs.launchpad.net/neutron/+bug/1413042 since currently neutron/kilo is broken on centos7/rhel7 14:58:39 <enikanorov_> https://review.openstack.org/#/c/149261 14:59:31 <enikanorov_> finally, https://bugs.launchpad.net/neutron/+bug/1411816 - "DB failure during functional job run" 14:59:48 <enikanorov_> that one was fixed by a revert of one of previous commits (tx isolation level change) 14:59:58 <enikanorov_> but it revealed the issue with functional DB tests 15:00:09 <enikanorov_> which worth to be fixed 15:00:13 <mestery> OK, we're at time now I think. 15:00:18 <enikanorov_> aaand that's it 15:00:19 <mestery> enikanorov_: Thanks for the updates on bugs! 15:00:23 <markvoelker> One last thing: don't forget to wish mestery a happy birthday today everybody! https://www.youtube.com/watch?v=2ImgF-gesXE 15:00:25 <mestery> I encourage folks to jump into #openstack-neutron to continue 15:00:31 <mestery> markvoelker: lol :) 15:00:36 <mestery> We'll see you all next week! 15:00:41 <mestery> #endmeeting