21:01:53 #startmeeting Networking 21:01:54 Meeting started Mon May 20 21:01:53 2013 UTC. The chair is markmcclain. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:01:55 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:01:58 The meeting name has been set to 'networking' 21:02:00 hi 21:02:04 hi 21:02:09 hi there 21:02:10 #link https://wiki.openstack.org/wiki/Network/Meetings#Bugs 21:02:24 #links https://wiki.openstack.org/wiki/Network/Meetings 21:02:45 I've moved up ML2 so we don't cut rkukura off this week 21:03:11 Any changes before we get started? 21:03:41 #topic Announcements 21:03:51 Hi 21:03:54 Havana-1 will be cut next week 21:04:04 #link https://launchpad.net/quantum/+milestone/havana-1 21:04:38 I think we're doing ok on our blueprints. 21:04:59 The one that is dragging the most is the name change, but I have not heard from legal 21:05:14 markmcclain: i had hoped to get the oslo db support in. thi is stuck on backward compatibility issues 21:06:03 garyk: right. want to leave it in for now? 21:06:27 markmcclain: at the moment we are waiting for oslo updates 21:07:07 sorry.. I forgot about those 21:09:13 I'll be reaching out to blueprint owners later this week for any that are not in review by Thursday. 21:09:36 The other announcement is that we will be meeting next week. 21:10:03 I know it is a holiday in the US, so update your portion of the agenda if you can't make the meeting 21:10:19 #topic bugs 21:10:38 maybe we should do a bug fix day(s) 21:10:45 +1 for bug fix day 21:10:48 markmcclain: I think next week you'll see how nice are midnight meetings ;) 21:11:00 :) 21:11:18 haha…with the virus I've been trying to shake..I kind of feel like I'm in one now 21:12:08 Currently we're not tracking any high or critical bugs that have been filed in the last week. 21:12:12 what virus? h1n9? 21:12:24 sorry for that. 21:12:58 And hello all. I'm new around here, hoping to dive deeper into quantum land. Just lurking for now. 21:13:00 gongysh: one of those random viruses you get from kids.. always worse on adults 21:13:19 roaet: welcome 21:13:45 Any bugs the team needs to know about? 21:14:00 Let's dive into sub team reports 21:14:04 #topic API 21:14:05 hello 21:14:11 blueprints: 21:14:38 https://blueprints.launchpad.net/quantum/+spec/l3-ext-gw-modes on track 21:14:49 might use another core reviewer (beyond amotoki) 21:15:04 I can review it 21:15:10 thanks 21:15:12 today is my review day. 21:15:13 https://blueprints.launchpad.net/quantum/+spec/quantum-l3-routing-plugin 21:15:39 patch resumed and updated. I will review it. Not sure if we have another core. 21:16:05 https://blueprints.launchpad.net/quantum/+spec/l3-router-port-relationship 21:16:20 that patch is too big https://review.openstack.org/#/c/20693/ 21:16:27 is the patch for this on, going to be restored soon? 21:16:38 we should definitely divide it. 21:16:45 gongysh: agreed, point that on review 21:16:52 gongysh: I will let Bob know. 21:16:57 ok 21:16:59 He couldn't make the meeting 21:17:06 There is a devstack patch for that as well, FYI. 21:17:17 markmcclain: any comment on l3-router-port-relationship 21:17:29 asking just because you're the owner 21:18:01 I handed testing off to others on our dev staff and was out on Friday and today… let me follow with them 21:18:13 ok, thanks 21:18:33 finally, the last patch expected for H-1 for bp/make-authz-orthogonal merged 21:18:36 rest will come in H-2 21:18:43 I also have a question about https://blueprints.launchpad.net/quantum/+spec/configurable-ip-allocation 21:18:56 This is not API, but has implications, and it's not in IPAM agenda 21:19:02 are we still targeting H-1? 21:19:55 my guess is it will slide to early H2 21:20:02 it would a bit difficult to complete the review for H-1, I'm afraid. 21:20:12 danwent still has -2 on it 21:20:32 thanks. So perhaps we can untarget it. 21:20:42 salv-orlando: I'll remove it 21:20:47 And with this our weekly report from the API world is done. 21:20:50 Thanks everybody, 21:20:59 PS: Questions? 21:21:14 Thanks for the update 21:21:30 #topic VPNaaS 21:21:38 First WIP code in the review ( we will continue discussion on the code) 21:21:42 Quantum Advanced Service Plugin for VPNaas https://review.openstack.org/#/c/29812/ 21:21:47 Quantum Advanced Service Plugin for VPNaas https://review.openstack.org/#/c/29812/ 21:21:58 sorry typo 21:21:59 Quantum Client for VPNaaS https://review.openstack.org/#/c/29811/ 21:22:05 discussed with driver archtecutre https://docs.google.com/presentation/d/1uoYMl2fAEHTpogAe27xtGpPcbhm7Y3tlHIw_G1Dy5aQ/edit 21:22:27 Still minor discussion on the API parameters. 21:22:44 also investigating how to use strongswan. (openswan is new candidate now) 21:22:48 That's all from me 21:23:10 VPN is running ahead of schedule 21:23:34 since most of these are marked H2 21:24:09 Yes code will be targeted for early H2. The review is for discussion :) 21:24:28 right.. just nice to see things running early :) 21:24:50 yeah :) Thanks for team 21:25:05 no meeting just week any discussions will be on the mailing list correct? 21:25:21 *VPNaaS Meeting 21:25:22 yes. mailing list or on the patch. 21:25:39 #topic Nova Integration 21:25:40 if it is needed, we will plan 21:25:47 nati_ueno: good to know 21:26:29 garyk: hi 21:26:44 sorry, was not paying attention 21:27:05 have not made any progress here. sorry 21:27:41 have been trying to invetigate the bugs with the multiple ip addresses being assigned. no reproduction at the moment 21:27:50 ok 21:28:07 I think others have tried to replicate too 21:28:21 regarding the migrations nova => quantum have run into some issues with the live migrations. need to work a little more on this too 21:28:38 thats about all. 21:28:49 garyk: what is the problem? 21:29:30 gongysh: we need to to doa migration from a linux bridge to a ovs. when i tried it failed. 21:30:13 hopefully next week i can provide some more details 21:30:14 garyk: we will talk after u wake up in the morning. thanks 21:30:18 ok 21:30:43 I noticed this update was revived: https://bugs.launchpad.net/nova/+bug/1112912 21:30:45 Launchpad bug 1112912 in quantum "get_firewall_required should use VIF parameter from quantum" [High,In progress] 21:31:05 yes, nachi aded in a patch for this which is in review 21:31:24 yes i rebased it 21:31:35 quantum - https://review.openstack.org/#/c/21946/ 21:31:52 i think that there is also a nova patch required too (not sure if this has been addressed) 21:31:55 This needs a counterpart patch n nova. It is good to go to H-2. 21:32:22 yes nova side is needed. OK let's fix both of that on the early H2 21:32:49 ok.. I'll retarget to H2 21:33:23 Any other Nova related items? 21:33:35 not that i am aware of 21:33:51 mine one 21:34:09 https://review.openstack.org/#/c/29767/ 21:34:20 passing of the host to quantum 21:34:28 thanks for it, I am look for the url. 21:34:49 i a tired but can still copy/paste :) 21:35:12 you are faster than me even you are in half sleeping state. 21:35:12 garyk: isn't there another patch for the same thing? 21:35:34 rkukura: i think that the other patch was abandoned. this one is more elegant 21:35:44 https://review.openstack.org/#/c/21141/ was it - OK 21:36:30 rkukura: looks like gongysh supersedes it 21:36:57 #topic Security Groups/Firewally 21:37:12 #topic Security Groups/Firewalling 21:37:23 Hi, nothing really new to report on the security profile side. 21:37:43 The one bug we had on it merged early last week 21:37:53 ok 21:37:55 SumitNaiksatam: How 21:38:01 hi 21:38:02 's FWaaS progressing? 21:38:09 yeah making progress: https://review.openstack.org/#/c/29004/3 21:38:28 have unit tests going through, but still need to add more 21:38:36 this is marked as WIP 21:38:47 how comfortable do you feel about making next Tuesday 21:38:52 yeah, i will remove that once i add a few more tests 21:39:15 for now we can leave this targeted the way it is 21:39:38 how far off are we on a ref implementation? 21:40:11 we don't have that targeted for H1, but discussion is progressing (with some code experiments) 21:40:20 I think RajeshMohan is here 21:40:22 ? 21:40:27 Yes, i am here today 21:41:00 RajeshMohan: you want to give a short updated on the discussion around the iptables implementation? 21:41:04 I mapped the APIs to IPTables rules and was looking for a good place to apply them 21:41:36 We plan to apply it on internal interface of quantum router 21:42:10 seems like a natural place 21:42:15 We are experimenting with IPSet but that is a strech goal at this stage 21:42:22 ok 21:42:43 stretch goal for the first iteration, but should be there eventually 21:42:45 I am also experimenting with Shorewall but most probably we will not do that 21:43:00 We will go there if we expose zones 21:43:13 ok 21:43:19 The reason I was asking about the ref implementation is that we had talked at the summit about having a ref implementation before merging the API 21:43:30 markmcclain: thats correct 21:43:39 I am working closely with Sumit 21:44:01 we structured the blueprint dates such that the ref impl will be available after the API work is complete 21:44:12 no easy way to do the ref impl without the API 21:44:34 correct.. I'm just going to move the release target since the code itself won't merge until H2 21:44:37 markmcclain: if it helps we can target the API to H2, so that the ref impl is also available along with it 21:44:49 markmcclain: you read my mind :-) 21:44:57 This is Gary from vArmour 21:45:04 the review is already there, so we can keep discussing 21:45:14 We just created a blueprint 21:45:32 gduan: thanks for joining and welcome to the quantum team 21:45:42 try to integrate our firewall with fwaas 21:45:42 good to see your blueprint for the fw driver 21:45:48 SumitNaiksatam: yep we definitely should review 21:45:53 gduan: welcome 21:45:55 also working with Sumit on the API 21:46:17 Any questions before we move onto LBaaS? 21:46:24 #topic LBaaS 21:46:29 hi 21:46:51 there's couple of things going on here. 21:46:53 I was out at the end of last week, so I was not able to follow up on my two action items from the LBaaS meeting 21:47:56 ok. would you have time to look at them this week? 21:48:00 yes 21:48:12 ok. so, it's https://review.openstack.org/#/c/28245/ 21:48:39 I'm working closely with Avishay on this one 21:48:54 Good 21:49:04 markmcclain: that one needs your attention https://review.openstack.org/#/c/28289/ 21:49:17 Right.. I have both flagged for review 21:49:45 Avishay suggested to remove driver selection part. Actually I don't see much value in removing what works, but I can add driver selection later 21:49:50 ok, thanks. 21:49:56 We can coordinate via email 21:50:03 #topic ML2 21:50:17 rkukura: hi 21:50:32 I've been getting some comments on the WIP patches, nothing too major 21:50:48 Plan to post updated quantum and devstack patches this evening or tomorrow 21:51:03 this will work with linuxbridge and openvswitch agents 21:51:07 will you be removing the WIP tag too? 21:51:15 this should be the last WIP version 21:51:19 ok 21:51:28 still need DB migrations, more unit tests, and maybe GRE support 21:52:04 after that, can we make the original plugins deprecated? 21:52:08 considering deferring the GRE support to H-2 due to the changes in the partial-mesh patch 21:52:29 migrations are a must for the patch to land, but I think delaying GRE to H2 is ok 21:52:33 I think we need full parity before we can deprecate 21:52:55 Note that this isn't migration from ovs or lb plugins to ml2 21:52:58 seems like something we can do in a patch and then make OVS and LB plugins deprecated 21:53:10 right.. it's the migration to create the underlying tables 21:53:25 main goal is to get new development onto ml2 21:53:32 right 21:53:43 rkukura: quick question, do you plan to include any other plugin into ML2? which one? 21:54:25 initial version focuses on RPC agent support - linuxbrige, openvswitch, and hyperv should work 21:54:40 rkukura: We are looking to port our Nexus plugin into ML2, would like to have this done for H2 if possible. Will syncup over email. 21:54:42 getting mellonox to work shouldn't be too much work either 21:55:08 ok, that should be one of the first uses for MechanismDrivers 21:55:15 I think all plugin should move to Ml2. 21:55:22 which should get in early H-2 21:55:26 +1 21:55:37 that is our big picture, isn't it? 21:55:38 rkukura: Thanks! 21:56:20 I'm not sure how far we'll get during havana, and monolithic plugins are still possible 21:56:43 yeah.. as a first step consolidating the open source versions is a win 21:56:55 Any thing else on ML2? 21:57:03 on track for H-1 for first step 21:57:09 cool 21:57:14 #topic python client 21:57:24 https://review.openstack.org/#/q/status:open+project:openstack/python-quantumclient,n,z 21:57:33 Wanted to ask about this: https://review.openstack.org/#/c/29157/ 21:57:47 I will help with it today. 21:58:15 I know several folks have independently run into issues on adminURL vs publicURL 21:58:25 why? 21:58:41 isn't adminURL there? 21:58:44 gongysh: by default it's using adminURL and they want it to use publicURL 21:58:45 difference deployments will specify the values differently 21:58:56 nova uses publicURL by default which is my understanding. 21:59:09 yes. 21:59:11 https://review.openstack.org/#/c/29157/ 21:59:16 so we need that patch. 21:59:48 gongysh: looks like that is addressed here https://review.openstack.org/#/c/29831/ 22:01:20 there are patches dependency 22:01:46 need more review on python quantum client. 22:01:58 right 22:01:58 that's all from me. 22:02:05 gongysh: thanks 22:02:16 the last one changes the default value. Do we change the default? 22:02:43 amotoki: changing to publicURL would be consistent with Nova's behaviour 22:02:50 I think we can. 22:03:02 markmcclain: gongysh: sounds good. 22:03:19 #topic Horizon 22:03:34 I picked up the registered blueprints related to Networking. The list is available on Wiki. 22:03:48 Thanks for volunteers for key netwokring features (FWaaS and VPNaaS) :-) 22:03:49 Thanks for adding the list 22:04:12 i will access the owner of BPs after H-1. 22:04:23 yeah it's cool we've had folks setup to volunteer 22:04:36 do we still need a volunteer for the Quote BP? 22:04:49 cheer for new volunteers. 22:05:11 Quote BP? 22:05:20 quota bp 22:05:22 Sorry Quota 22:05:32 quota bp is already assigned, but i don't know the status. I will ask the assignee. 22:05:46 ok 22:06:05 I'm excited for the new networking features that are being added to Horizon 22:06:23 o/ 22:06:32 that's all from me. 22:06:38 Thanks for the update 22:06:39 so it is natural for quantum to replace nova-network since quantum is everywhere. 22:06:52 gongysh: right :) 22:07:02 #topic Open Discussion 22:07:10 Xen reviews have been stagnating 22:07:12 IPv6 meeting? 22:07:23 multihost? 22:07:34 gongysh: need to schedule a meeting for v6 22:08:02 I'll contact the v6 folks via email and ML 22:08:02 have we scheduled? 22:08:10 ok, thanks. 22:08:14 https://review.openstack.org/#/c/24771/ 22:08:39 markmcclain: -2 from u 22:09:27 right.. sorry that should have been removed after you added tests 22:09:34 i am sorry but i am going to have to call it a day (night). good night 22:09:39 garyk: night 22:09:44 garyk: night 22:10:01 Any other items for open discussion? 22:10:04 yep 22:10:09 a minor one 22:10:20 have this patch https://review.openstack.org/#/c/29155/ 22:10:38 i thought folks working on service plugins may consider this approach 22:10:48 Nachi, Sumit 22:10:58 enikanorov_: hi 22:11:19 hi! 22:11:28 enikanorov_: i saw that, my current implementation has the inheritance approach 22:11:42 enikanorov_: I'll comment this for VPN patch 22:11:46 but not set in stone 22:12:03 inheritance could be ok if the plugin is doing db-only 22:12:20 if it communicates via drivers/agents then you might want to mock db stuff anch test that 22:12:26 enikanorov_: could you also review vpn one https://review.openstack.org/#/c/29812/ ? 22:12:30 thet's where composition helps 22:12:46 nati_ueno: I will 22:12:57 enikanorov_: thats a good point 22:13:10 enikanorov_: will take a look 22:13:31 we're over time, so everyone should make sure to comment on the reviews 22:13:32 has anyone bumped into this yet http://lists.openstack.org/pipermail/openstack-dev/2013-May/009276.html, ran into it this morning https://review.openstack.org/#/c/29280/ 22:14:12 hi markmcclain: we'd like to get this in for H1. https://bugs.launchpad.net/quantum/+bug/1133215 22:14:13 Launchpad bug 1133215 in quantum "MidoNet test fails when new test is added in DBMixins " [Medium,In progress] 22:14:39 dkehn: yes, multiple times 22:15:10 HenryG, just keep submitting your patch unitl it works or?? 22:15:32 tomoe_: I am helping on the review 22:15:37 ping me averytime you have a new patch 22:15:47 thanks emagana 22:15:53 tomoe_: I'll take a look too 22:15:55 it's been floating around and the patch is ready. 22:15:59 dkehn: yup, but sdague helped me on #openstack-qa to root-cause 22:16:15 so it'd be nice if we could get that out. 22:16:20 thanks markmcclain ! 22:16:50 dkehn: I haven't seen that in awhile 22:17:32 Remember we will meet next week for those who can make it. 22:17:35 #endmeeting