21:01:12 #startmeeting Networking 21:01:13 Meeting started Mon May 6 21:01:12 2013 UTC. The chair is markmcclain. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:01:14 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:01:16 The meeting name has been set to 'networking' 21:01:19 Welcome back gongysh 21:01:21 me too (back from vacation) 21:01:27 you too nati_ueno :) 21:01:41 i want to go on vacation 21:01:42 mestery: thx 21:01:57 me too. 21:01:59 here's our agenda 21:02:02 #link https://wiki.openstack.org/wiki/Network/Meetings 21:02:25 danwent: I forgot to add time for docs 21:02:32 so we''ll insert it 21:02:48 #topic Announcements/Reminders 21:03:05 #info Havana-1 is three weeks away 21:03:22 We've got a full list of items to be completed 21:03:28 markmcclain: tis ok, can do it at end 21:03:33 #link https://launchpad.net/quantum/+milestone/havana-1 21:03:42 danwent: ok 21:04:35 We currently have 27 blueprints targeted which seems high 21:05:11 seems too much. 21:05:35 Agree, there is a lot there for the next 3 weeks. 21:05:49 for the blueprints from my team: we have made a provisional allocation. Will rearrange during this week. 21:06:32 Thanks salv-orlando. You were reading my mind 21:07:24 Which is to let me or the sub-team know if there is blueprint that is not realistically going to make it in H1. We can move it back to H2 to make the plan more realistic 21:08:21 I do want to note the one essential blueprint for H1. Renaming will be a bit disruptive, but once we have the final plan I share with everyone so that we can work to minimize the disruptions. 21:08:43 #link https://blueprints.launchpad.net/quantum/+spec/remove-use-of-quantum 21:09:04 does this imply getting rid of all "quantum" in the code? 21:09:06 When we decide new name? 21:09:11 we have the blue print targeting multivendor driver for LBaaS which I do not see as part of H1 21:10:03 enikanorov_: It will involve some changes to comply with agreement we made to stop using Quantum. I'm still working out the final list of changes required. 21:10:38 nati_ueno: We have not settled on a final name. I'm still accepting alternatives as the legal team will vet a few before we as a team select one. 21:10:44 has a name been determined to replace the old one? 21:10:49 just wondering. getting rid of the name in the source tree is quite disruptive... 21:11:04 markmcclain: Are there candidates? 21:11:13 affects imports, deployment 21:11:34 current candidate is Quasar 21:11:38 The name change will require some amount of coordination with distributions as well I suspect. 21:11:42 markmcclain: Thanks! 21:11:59 yeah.. name changing will involve multiple changes 21:12:00 In regards to the name, is there a guidline to use a descriptive name such as os_network or to use an opaque name? 21:12:05 Let the legal team know Quasar is the name of a Marvel Comics superhero. :) 21:12:36 damn you mestery 21:12:37 :) 21:12:43 =) 21:12:45 danwent: I know my comic books. 21:13:01 And I would hate to have to rename twice. 21:13:03 Ugh. 21:13:13 so if any one else has suggestions, send them to me offline 21:13:25 vetting by legal is to ensure we don't have to rename twice 21:13:55 there is company in australia http://www.reverseaustralia.com/lookup/0412013398/ 21:14:20 What about "flux" 21:14:28 ? 21:14:33 interesting 21:14:51 but its stable :) 21:14:54 my insurance company is called flux :) 21:15:04 OpenStack Flux seems oddly appropriate. :) 21:15:15 may be we should use mkpasswd to generate safe project name 21:15:24 haha 21:15:26 nati_ueno. Heh, I like it. 21:15:26 uuid.uuidgen() 21:15:28 lol 21:15:44 there's always the dark horse: mutnauq 21:16:00 asdfasdf is easy to type 21:16:13 mutnauq #ftw 21:16:18 qnetwork if you still want the q 21:16:20 i am the mutnauq… hear me roar 21:16:27 how about quantim 21:16:37 mestery: don't forget MTV's Aeon Flux ;-) 21:16:38 garyk: easy to migrate! 21:16:41 kuantom 21:16:45 lol.. so send me your suggestions for names and clear them before voting 21:16:49 i dont know what a mutnauq is but im scared to google it at work 21:17:05 Any other questions before moving on? 21:17:29 Samuel_: I'll add the blueprint to H1 21:17:30 sthakkar: i think that either danwent or salv-orlando or markmcclain were in favor of that at some stage 21:18:04 you need a mirror to get mutnauq 21:18:19 #topic Bugs 21:18:50 This was the high importance bug from last week 21:18:50 salv-orlando: a-ha 21:19:09 Thanks to garyk for looking into the issue 21:19:17 markmcclain: np. 21:19:32 Any other bugs that the team should be aware of? 21:20:00 #topic API 21:20:01 Bug 1174293 21:20:03 Launchpad bug 1174293 in quantum "Network creation does not notify DHCP agent" [High,In progress] https://launchpad.net/bugs/1174293 21:20:20 Is that one bug you were mentioning? it did not show up on my client 21:20:35 salv-orlando: you're right.. I forgot to paste it 21:20:38 This is behavior by design 21:20:39 that is the bug 21:20:45 i am currently looking into the issue of the gre and 3 nodes - https://bugs.launchpad.net/quantum/+bug/1154383 21:20:47 Launchpad bug 1154383 in quantum "Invalid entries in OVS flow table on a 3 node install using GRE Tunnels" [Undecided,Confirmed] 21:20:48 so perhaps you might want to discuss it a little 21:21:09 gongysh think it's not a bug (and I remember I did in this way on purpose) 21:21:14 I -> he 21:21:54 salv-orlando: yes (bug 1174293) 21:21:55 Launchpad bug 1174293 in quantum "Network creation does not notify DHCP agent" [High,In progress] https://launchpad.net/bugs/1174293 21:22:04 If I understand correctly, there is a chance that the agent never gets selected if the conditions are right 21:23:06 makrmcclain: what do u mean by 'never get selected'? 21:23:14 garyk: can you elaborate? 21:23:28 when port is created, we will schedule the network auto. 21:24:03 markmcclain: maybe i missed the part of the port creation. i'll go back and check again (maybe it was a misunderstanding on my side) 21:24:27 ok.. gongysh can you add that to the bug report so that others have that info too? 21:24:38 sure 21:24:41 thanks 21:24:41 i'l try and reproduce. 21:24:54 sounds good 21:24:57 #topic API 21:25:00 salv-orlando: hi 21:25:02 hi 21:25:20 I wanted to spend some more time on API today, but things are getting tight 21:25:37 so let's be quick. Blueprints targeted for H-1 are still on target 21:25:39 salv-orlando: you can take the stables time (we are in freeze and i have updated the wiki with the status) :) 21:26:09 however, no significant progress happened over the last week 21:26:23 markmcclain: what's your estimate for the router port API? 21:26:37 https://blueprints.launchpad.net/quantum/+spec/l3-router-port-relationship 21:27:27 on the other blueprints, 21:27:29 we've been testing internally and need to fix a bug before pushing up for review 21:27:38 ok, thanks. 21:28:02 I have asked for feedback about how to handle a peculiar authZ problem in configurable-ext-gw-modes 21:28:23 it seems the best thing to do is to match policies against sub attributes. Should be easy, 21:28:34 Then https://blueprints.launchpad.net/quantum/+spec/api-core-for-services 21:28:58 i believe i have a patch for the last one 21:29:46 I have not yet completed my work, but I am happy to throw away what I have done if you're complete with it 21:30:08 please if you feel you want to own a blueprint, tell people in advance so there won't be duplicate work :) 21:30:34 This is for blueprints 21:30:45 yeah… we should avoid duplicate work whenever possible 21:31:02 there are no high importance bugs, but the ones in the agenda are worth some discussion 21:31:13 sure. api-core-for-services is 3 days old so i hope there hasn't been duplicate work yet 21:31:16 especially bug 1165002 21:31:17 Launchpad bug 1165002 in quantum "tenant lists the external network of other tenants" [Medium,Confirmed] https://launchpad.net/bugs/1165002 21:31:52 This is because we are treating external networks as shared networks from an authZ perspective. They can be seen by anyone, but nobody can operate on them 21:32:12 this means we don't have a concept of "private external network" 21:32:50 does it lead to problem for nova to create a instance on it? 21:32:55 imho we should not invent a new concept, but leverage existing facilities. Hence I think the new logic should be that an external network should be shared too to be seen by everybody, but this will break bw compatibility 21:33:11 gongysh: no problem for nova AFAIK 21:33:34 so to keep a long story short, I will post a patch at some point during this week. You are welcome to express your opinion and/or ideas on the bug report 21:33:46 Before finishing 21:33:58 ok.. makes sense to discuss there so everyone can follow along 21:34:08 one more pledge: please help with API doc reviews: https://review.openstack.org/#/q/status:open+project:openstack/netconn-api,n,z 21:34:16 that is all from the API side 21:35:07 thanks for the update 21:35:20 #topic VPNaaS 21:35:26 nati_ueno: 21:35:38 Ok 21:35:49 we agreed the first step is IPsec + Amazon VPC model 21:35:59 now we are discussion models in IPSec 21:36:11 Note first step must be pretty simple 21:36:23 we will have meeting today 5pm PST on conf call 21:36:47 See the mailer for details on the call if you're interested 21:37:02 This is Meeting Agenda and Note https://docs.google.com/presentation/d/1J7k1eI13-3pQVwp5XgZDWPfzUvuSqczRdK0lEZKQOKk/edit#slide=id.p 21:37:14 This is draft bp https://docs.google.com/a/ntti3.com/document/d/1Jphcvnn7PKxqFEFFZQ1_PYkEx5J4aO5J5Q74R_PwgV8/edit 21:37:24 thats all from me 21:37:38 ok.. thanks for the update 21:37:53 #topic Nova Integeration 21:38:31 garyk: any update or still investigating? 21:38:34 markmcclain: sorry no updates on my side 21:38:38 still investigating 21:38:41 no problem 21:38:58 #topic Security/Firewalling 21:39:01 arosen: hi 21:39:01 hi, only update is this one bug (https://bugs.launchpad.net/quantum/+bug/1171997) around the creation of a default security group when doing list-security groups. 21:39:03 Launchpad bug 1171997 in quantum "For networks shared beetwen multiple tenants (net-create --shared) there's no 'default' security group" [Undecided,In progress] 21:39:06 There has been a good amount of discussion on the bug report. I think we are close to reaching consensus though. 21:39:21 * arosen on the review i mean. 21:39:25 That's it from me though. 21:39:32 https://review.openstack.org/#/c/27792/ 21:40:13 yup that's it. This should hopefully land shortly. 21:40:21 ok 21:40:42 SumitNaiksatam: anything on fwaas? 21:40:52 #topic FWaaS 21:40:57 yeah, nothing major 21:41:08 proceeding along the lines of what we laid out in the plan 21:41:24 couple of BPs targeted for H1 21:41:46 any questions/thoughts? 21:42:23 I thought the blueprints were focused fairly well 21:42:33 and the dependencies between them made sense 21:42:34 markmcclain: ok 21:42:52 background discussion going on the driver implementation 21:43:03 we are starting on that right away in parallel 21:43:09 iptables and ipsets for now 21:43:32 great 21:43:56 Any Security Groups/Firewall discussions? 21:43:58 what is ipsets? 21:44:20 gongysh: helps to define named objects 21:44:30 ok, thanks 21:44:31 so that you can reference them in iptables rules 21:44:37 and then late bind 21:45:16 there is also an abstraction called shorewall, which we can use on top of ipsets and iptables 21:45:30 rather than using iptables/ipsets directly 21:46:10 still investigating/exploring (rajesh/dell working on these) 21:46:26 ok.. I like that you've started working on it early 21:46:37 Thanks arosen and SumitNaiksatam for the update 21:46:45 #topic LBaaS 21:46:48 i'll update 21:46:58 Thanks.. sorry I missed the meeting today 21:47:02 there is several key items in progress on lbaas 21:47:18 1) ServiceType framework 'rework' 21:47:36 redefined some terms and agreed on iplementation steps 21:48:16 2) Some renaming/moving of lbaas files. 21:48:38 i'm going to create wiki page on final directory structure for the service and propose it for other services 21:48:59 question to discuss: whether to move services directory one lvel up ffrom plugins 21:49:13 ok.. I know you started discussion, but I'm sure we have consensus yet across the full team 21:49:19 *I'm not sure* 21:49:51 i know. I hope there will be more clarity once wikipage is created 21:50:16 yeah.. the wiki will help a bunch 21:50:24 enikanorov-w: seems like moving "services" one level up is a good idea 21:50:54 SumitNaiksatam: the sooner the better (if we ever decide to) 21:50:54 IMO, "Services" discussion should be managed in this meeting 21:50:57 sorry, don't mean to spark the debate here :-) 21:51:02 it affects many sub projects 21:51:05 time is tight 21:51:08 ok 21:51:10 so 21:51:18 3) two patches on "multivendor support" 21:51:41 how about enikanorov_ post the wiki and we can discuss on mailing list and then I'll make it a topic for next week 21:52:08 markmcclain: sure 21:52:12 that way everyone can be informed for we dive into the issue 21:52:23 regarding (3) - I hope some core will take a look at those patches 21:52:47 at least for some brief design review 21:53:03 and 4) LBaaS subteam meeting 21:53:10 I do not understand why we have two patches 21:53:11 I feel we might need one. 21:53:47 enikanorov_: I know last cycle the sub-team held ad-hoc meetings early on to organize 21:53:59 and I think makes sense again this cycle too 21:54:14 I think enikanorov_ is proposing regular ones this time 21:55:01 yeah.. that's certainly workable 21:55:34 some sub-teams of nova meet regularly to facilitate communication 21:55:40 should we do it 1 H before this meeting? 21:55:48 i think everyone involved in writing code/reviewing should attend 21:56:06 of if most of the team is not in US, we might be able to do it earlier ;-) 21:56:51 ok.. look at the main openstack meetings wiki 21:57:08 to find an open time and then suggest options on the mailing list 21:57:21 markmcclain: a question 21:57:34 sure 21:57:40 (1) and (2) is not bound to any blueprint right now, i believe 21:57:57 but they are necesssary to impl before moving to (3) 21:58:07 ok.. we need to make sure that work is tracked by a blueprint or bug 21:58:17 we missed the deadline as we had the meeting today 21:58:32 we can coordinate offline to fix those up 21:58:32 I disagree. we can complete 3 very quick and then do the reordering 21:58:47 markmcclain: ok 21:58:53 got to run folks, see you next week! 21:59:04 danwent: thanks 21:59:32 ok.. we're running short on time so I want to highlight one CLI review 21:59:43 #topic CLI 22:00:02 https://launchpad.net/python-quantumclient/+milestone/2.2.1 22:00:29 ssl is in the progress ( if I am right) 22:00:35 the next release will actually be 2.2.2 22:00:47 #action fix CLI milestone 22:01:29 some deployers have run into problems with adminURL, publicURL in the catalogs 22:01:30 I am confused too. :) 22:01:37 https://review.openstack.org/#/c/28057/ 22:02:00 seeks address those issues.. I don't want to discuss here, but I would like everyone to look at this review 22:02:14 #topic Stable 22:02:28 markmcclain: i updated wiki.3release is 9 may 22:02:46 we are in freeze at the moment 22:03:08 Right.. we did get an exception for the fix Windows metadata support 22:03:16 correct. 22:03:30 i am backporting a few fixes at the moment - for the next release cycle 22:04:00 That's great you're getting a jump on the next cycle 22:04:25 #topic Docs 22:04:35 https://bugs.launchpad.net/openstack-manuals/+bugs?field.tag=quantum 22:05:05 Dan had to leave early but we've got three review in progress and three more we need to complete. 22:05:47 Works in progress will be posted for migration and db and then we can fix those update as a community vs funneling the changes through one person. 22:06:37 As salv-orlando ask earlier, please take a look at the doc reviews so we can close these out. 22:06:49 #topic Other Reports 22:06:56 https://review.openstack.org/#/q/status:open+project:openstack/netconn-api,n,z 22:07:12 salv-orlando: Thanks for the link 22:07:42 i am going to crash. good night guys. sorry 22:07:44 maybe we need to give some bonus for reviewing. 22:08:01 garyk: Night.. sorry we ran over on time 22:08:14 will review some docs with bonus of course! 22:08:16 :-) 22:08:23 gongysh: You offering to pay :) 22:08:32 bonuses awake ppl 22:08:36 I think you have kind of a duty to do that, don't you :) 22:08:36 ^_^ 22:08:37 enikanorov_: Error: "_^" is not a valid command. 22:08:42 ops 22:09:35 #Open Discussion 22:09:44 #topic Open Discussion 22:09:45 See meeting agenda for open source plugin items (ml2, vxlan, hardware drivers, etc.) 22:10:04 rkukura: Thanks for updating the agenda 22:10:10 where is the agent for ml2? 22:10:27 agent -> agenda 22:10:38 ml2 uses the linuxbridge, openvswitch, and hyperv agents 22:10:51 gongysh: https://wiki.openstack.org/wiki/Network/Meetings#Other_team_reports_.28e.g._Open_Source_Plugins.29 22:11:16 Here's the associated review: https://wiki.openstack.org/wiki/Network/Meetings#Other_team_reports_.28e.g._Open_Source_Plugins.29 22:11:35 https://review.openstack.org/#/c/20105/ 22:11:35 do we have a patch for ml2 now? 22:11:41 posted update WIP patch set for ml2 22:12:02 this works with the linuxbridge agent, adding support for other agents this week 22:12:26 We're over time, but everyone can discuss and provide feedback via the review and mailing list 22:12:39 I will join to review it today. 22:12:40 Anything else before we wrap up? 22:12:50 gongysh: this 1st ml2 phase overlaps https://blueprints.launchpad.net/quantum/+spec/merge-plugin-codes-into-one - lets not duplicate effort 22:13:03 yes, 22:13:29 I will not do it. I like to reivew instead of code myself. :) 22:13:47 cool. Then make that blueprint superseded and untarget it. 22:13:52 gongysh: either one beats meetings! 22:14:10 This way markmcclain will be slapped one time less by ttx 22:14:23 rkukura, can u give me the chance to help with other plugin porting? 22:14:29 ok, bye folks 22:14:53 gongysh: thanks! 22:14:59 salv-orlando: haha.. thanks for the reminder to tidy up the old BPs 22:15:30 ok.. I've got to run. Thanks everyone. Talk you all on the mailing list or next week back here. 22:15:35 #endmeeting