14:00:12 #startmeeting networking 14:00:13 \o 14:00:13 Meeting started Tue Oct 20 14:00:12 2015 UTC and is due to finish in 60 minutes. The chair is armax. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:14 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:14 hi 14:00:14 hi 14:00:16 The meeting name has been set to 'networking' 14:00:19 o/ 14:00:21 mestery: diet dr pepper 14:00:24 morning/afternoon/evening folks 14:00:25 o/ 14:00:25 hello all! 14:00:31 regXboi: Makes more sense ;) 14:00:31 hi 14:00:35 hi 14:00:38 I second Sam-I-Am: moo 14:00:39 thanks for the ping, I may need to run around 16:45, depending on how my "backup" for a task works.. 14:00:52 xx:45 :) 14:00:54 already had coffee... ready to go 14:01:05 emagana: and you're an hour earlier than me 14:01:10 impressive 14:01:18 hi 14:01:19 #link https://wiki.openstack.org/wiki/Network/Meetings Agenda 14:01:21 thanks for those early "wake-ups" 14:01:21 hi 14:01:30 #topic Announcements 14:01:32 Sam-I-Am: yeah.. I can't believe that I am already at work 14:01:33 aloha 14:01:38 we all know this is the week before the summit 14:01:40 hi 14:01:40 \o 14:01:44 * gongysh step in 14:01:47 hi 14:01:48 * jroll pokes dtantsur 14:01:56 o/ 14:01:56 hi 14:01:56 as a reminder, we’ll cancel the next two meetings 14:02:04 hi all 14:02:09 yeah.. Tokyo! 14:02:11 hi 14:02:18 hi 14:02:30 * mestery expects some karaoke in Tokyo 14:02:35 enjoy tokio, I'll miss hanging out with you guys&gals 14:02:47 ajo: You're not gonna be there? #sadpanda 14:02:49 ajo: i'm not going either 14:02:54 * regXboi wonders if mestery can sing... 14:03:01 I can't sing, so be warned regXboi 14:03:04 * dtantsur bites jroll 14:03:05 talking about the summit 14:03:07 regXboi: depends on booze levels? 14:03:15 * regXboi isn't worried - he won't be there either 14:03:18 there’s a social event being organized 14:03:22 #link http://lists.openstack.org/pipermail/openstack-dev/2015-October/077108.html 14:03:24 regXboi: we can have our own denver summit 14:03:24 Sam-I-Am: that's very likely :) 14:03:29 mestery, sadly I can't this year, but you will have to bear with me next one ;D 14:03:33 please sign up if you intend to join 14:03:41 Sam-I-Am & ajo: What? We will miss you guys 14:03:41 dtantsur: well that escalated quickly 14:03:59 emagana: i'll be there in spirit 14:04:04 * rossella_s already signed uo 14:04:11 * regXboi updates calendar to skip next two meetings 14:04:12 * mestery registered too 14:04:17 it will be thursday 19:00-. amotoki made a reservation. further details will be announced on ML. 14:04:22 Sam-I-Am: spirits dont drink with us 14:04:29 * ihrachys registered long time ago 14:04:30 thanks for organizing yamamoto and amotoki! 14:04:30 emagana: but you drink spirits 14:04:31 * armax reminds himself to register 14:04:35 thanks yamamoto and amotoki 14:04:37 emagana: technically we drink spirits 14:04:41 hi 14:04:52 the summit is also a week away 14:04:54 #link http://mitakadesignsummit.sched.org/overview/type/neutron 14:04:57 salv-orlando: good point.. 14:05:05 we have a nearly final draft of the schedule 14:05:16 I am working with the session chairs to polish it up 14:05:43 great :) 14:05:54 thanks to mestery, dougwig, carl_baldwin, kevinbenton, ihrachys, sc68cal, xgerman, cathy, gsagie, et al 14:06:21 oh and by the way... 14:06:31 #link https://launchpad.net/neutron/+milestone/7.0.0 14:06:35 Liberty is OUT! 14:06:38 yay 14:06:39 sc68cal: you are there already! 14:06:39 yay 14:06:42 meh. 14:06:43 :) 14:06:47 so is the install guide, with linux bridge :) 14:06:47 yay! 14:06:56 all the fun is about master. 14:06:57 armax: great job! as well as the rest of the team 14:07:14 ihrachys: lol :) 14:07:18 emagana: the job was mestery's 14:07:18 :) 14:07:21 Well 14:07:27 It was everyone's, I merely herded some cats ;) 14:07:32 or a team effort 14:07:33 its all a mestery to me how it works 14:07:35 armax knows this well by now I suspect :) 14:07:35 let’s put it that way 14:07:38 * salv-orlando thinks ihrachys is a wise man 14:07:42 #neutron: great team people! 14:07:56 before we dive in to the next section 14:07:57 \o/ :D 14:08:27 * regXboi looks for the confetti 14:08:35 I want to remind that we have revised some of our processes 14:08:43 #link http://docs.openstack.org/developer/neutron/policies/blueprints.html 14:08:51 #link http://docs.openstack.org/developer/neutron/policies/bugs.html 14:09:48 please familiarize yourself with them and evangelize the gospel to your proselytes 14:10:10 heh 14:10:15 aye My Lord 14:10:39 ack :) 14:10:39 armax: wise words! 14:10:43 * mestery kneels before the processes 14:10:54 lol 14:11:01 Seriously, nice work on cleaning all this up armax. 14:11:13 mestery: hopefully it’ll make a difference 14:11:17 armax: ++ 14:11:22 in fact, this brings me to the next section 14:11:24 of the meeting 14:11:29 * regXboi wakes up 14:11:29 #topic Blueprints 14:11:39 regXboi: not yet ;) 14:11:47 but please do 14:12:02 #link https://launchpad.net/neutron/+milestone/mitaka-1 14:12:18 we have some blueprints optmistically targeted for M1 14:12:26 which is roughly a month and half away 14:12:52 yikes 14:12:59 nice to see linuxbridge in the list :) 14:13:01 one of the process changes we identified was to identify an approver per blueprint 14:13:22 regXboi: where's your dvr-fix-all-the-things? :) 14:13:36 the approver will pair up with the assignee of the blueprint to work closely together to get the blueprint throught the motions 14:13:38 Sam-I-Am: you think I'm crazy enough to add a BP? :) 14:13:47 regXboi: you're crazy enough to dvr 14:13:53 Sam-I-Am: that’ll happen 14:14:00 Sam-I-Am: no, that's just lunacy :) 14:14:02 Sam-I-Am: we only need to tackle regXboi into a corner 14:14:14 challenge accepted 14:14:24 armax: how do we assign ourselves? 14:14:32 now…most of the M1 blueprints have no approver 14:14:47 and some may even need the assignee to be revised 14:14:52 ihrachys: as an approver? 14:15:05 armax: yes. I don't have the option to assign myself in LP 14:15:10 * regXboi has seen evidence of assignee revision already 14:15:11 I would take that one https://blueprints.launchpad.net/neutron/+spec/ml2-lb-ratelimit-support 14:15:23 ihrachys: done 14:15:51 ihrachys: only members of the neutron-drivers team can 14:15:52 #link https://launchpad.net/~neutron-drivers 14:16:05 ihrachys: hey there :) 14:16:10 armax: aye. I believe I need to get into gerrit group for that. :) 14:16:20 ihrachys: these things will need some docs 14:16:32 armax https://blueprints.launchpad.net/neutron/+spec/restructure-l2-agent should also be targeted for M1, right? 14:16:41 Sam-I-Am: define 'these things' 14:16:41 Sam-I-Am: +2 14:16:44 rossella_s: let me make a note for it 14:16:49 lb+qos 14:16:55 rossella_s: +1 to that 14:17:08 Sam-I-Am: ok, sure. let's touch bases later out of meeting context. 14:17:16 armax: only members of the neutron-drivers team can assign people to blueprints? 14:17:20 rossella_s: who do you think could be a good approver for the work? 14:17:30 salv-orlando: it’s my understanding, yes 14:17:49 armax, I think yamamoto could be a good approver, he reviewed most of the patches already 14:18:00 rossella_s: ok, yamamoto you around? 14:18:06 hi 14:18:17 armax: it must be one of those policies that I have not read ;) Sounds a bit too restrictive to me, but I also understand why it's being doner 14:18:18 done 14:18:23 i'm not sure what's approver's job 14:18:34 yamamoto: can you pair up with rossella_s blueprint on restructure l2 agent ? 14:18:43 yamamoto: please review the policy change above 14:18:55 * ajo needs to review those too 14:18:56 yamamoto: if you are still in doubt come and talk to me 14:19:11 salv-orlando: feedback is welcome 14:19:19 salv-orlando: if you keep it to yourself it’s hardly useful 14:19:47 salv-orlando: besides if you haven’t read it, how can you claim that it is restrictive :)? 14:20:45 there are a few other blueprints that need a custiodian angel 14:20:48 armax: i got it. please assign me. 14:20:52 another word for approver :) 14:21:05 today I feel very religious for no obvious reason 14:21:17 * Sam-I-Am backs away slowly 14:21:21 :) 14:21:28 Sam-I-Am: don’t be shy 14:21:30 this explains a lot of neutron things 14:21:34 * regXboi wonders if armax is taking the crusade 14:22:06 I’ll go over the current backlog and preaching people to join the mass 14:22:12 how about taht? 14:22:18 * Sam-I-Am preaches docs 14:22:33 armax, I can take https://blueprints.launchpad.net/neutron/+spec/vlan-aware-vms 14:22:46 you want to take over as approver, correct? 14:23:02 armax, yes, you were looking for approvers right? :P 14:23:07 yes 14:23:18 I need to see whether the code submitters are still around 14:24:29 I can be code submitter. :) 14:24:36 this BP: https://blueprints.launchpad.net/neutron/+spec/get-me-a-network 14:24:40 does not have an approver 14:24:45 armax: re your last message referencing me... you know I have a long history of talking about stuff I know nothing about 14:24:45 we need to pair up with haleyb 14:24:45 an important one imo 14:25:04 at least from a usability and docs perspective :) 14:25:05 * haleyb is listening 14:25:06 and this work has already languished for too long 14:25:40 haleyb: pls, scroll back a little 14:26:02 any takers for helping haleyb move this along? 14:26:04 re get-me-a-network, do we have patches up? why is it Slow progress? 14:26:11 ihrachys: yes we do 14:26:24 ihrachys: https://review.openstack.org/#/c/233306/ 14:26:35 ihrachys: yes, initial patch, but needs work 14:26:48 that will be a single patch? 14:26:56 most likely it won’t 14:27:08 anyone interested? 14:27:11 russellb: ? hint hint 14:27:35 should assign sc68cal :) 14:27:35 armax: I can assist, maybe with russellb? 14:27:40 armax: I can volunteer to help move those along as well. 14:27:41 not sure I should opt-in while I already have one, but if everyone is so shy or busy, I can consider it. 14:27:49 ok cool, I step back 14:27:54 lol :) 14:27:59 armax: what roles do you need for this blueprint? reviewers, developers, nit-pickers? 14:28:05 O.O 14:28:13 sure! 14:28:16 I think we should be ok so long as we keep the number of bps assigned to a few 14:28:17 what did i agree to 14:28:24 russellb: all the things 14:28:28 cool, that works 14:28:30 salv-orlando: all of the above? 14:28:31 buying first round at summit 14:28:36 ok! beers on me 14:28:39 salv-orlando: mostly code-review 14:28:55 salv-orlando: and make sure that any road blocker is flagged promptly 14:29:00 blueprint assignments sound like fantasty football drafts 14:29:01 russellb: thanks russellb 14:29:10 obviously anyone should feel free to help 14:29:11 yeah, i can help with that post-summit probably 14:29:15 this week is kind of nuts already 14:29:21 armax: make sense. I can assist with reviews - I hope I still remember how this stuff works 14:29:34 salv-orlando: any other blueprint you fell like taking? 14:29:48 salv-orlando: there are a few which have no approver 14:29:50 armax: should edgar and/or i somehow tag ourselves to track these bps from a docs perspective? 14:30:07 they shouldnt get completed without docs, imo 14:30:26 * ihrachys notes that would help if LP would show assigned approvers on the milestone page 14:30:40 ihrachys: I think there’s a way 14:30:44 it’s not great 14:30:57 ihrachys: https://blueprints.launchpad.net/neutron/+assignments 14:30:59 #link https://blueprints.launchpad.net/neutron/+assignments 14:31:24 I see this one has no approver: https://blueprints.launchpad.net/neutron/+spec/vm-without-l3-address 14:31:27 I could watch over it 14:31:34 Sam-I-Am: definitely, I would expect the approver to work with you to see where the doc gaps are 14:31:39 Any way to search for blueprints by approver (e.g. myself)? 14:31:40 ajo: nice 14:31:40 armax: cool 14:31:55 armax: I think I can assist with https://blueprints.launchpad.net/neutron/+spec/address-scopes 14:32:01 * carl_baldwin seeing any advanced search option for Bps 14:32:06 *net seeing 14:32:13 carl_baldwin: f.e. https://blueprints.launchpad.net/~brian-haley/+specs?role=approver 14:32:16 armax: I found hard to believe that the plan of working closely by magic will work 14:32:29 ihrachys: Thanks! 14:32:30 emagana: we don’t know if we don’t try 14:32:37 armax: we need to have a plan to really enforce it.. otherwise Sam-I-Am and I will be just chasing people 14:33:04 we can talk about that in more detail offline 14:33:08 armax: We have been under this model for several releases.. Do not know really what is different this time.. 14:33:28 armax: sounds good. 14:33:39 emagana: did you read the new details in the blueprint process? 14:33:40 keep me in the loop on that one 14:33:49 carl_baldwin: btw the link https://blueprints.launchpad.net/neutron/+assignments 14:33:53 Sam-I-Am: yes! 14:33:58 shows a huge list of bp's 14:34:01 emagana: oh, for docs. I see 14:34:04 kevinbenton: yes! 14:34:08 armax: I still have not read those revised blueprint policies, but were you asking if I could be approver for some blueprint? If non-cores can be trusted in that role, I am ok for taking care of the address scopes bp 14:34:11 kevinbenton: and yes again! ;-) 14:34:11 That link has all the BPs I believe, right armax? 14:34:30 when we manage to cut it down to the bps only for the most recent series we should be able to have a better view of the workload 14:34:57 salv-orlando: surely, you’re a trusted resource 14:35:12 currently the address-scope one is assigned to carl_baldwin 14:35:23 as approver I mean 14:35:34 salv-orlando: core emeritus 14:35:49 carl_baldwin: indeed! 14:35:56 armax: it seems to me there's no approver there 14:36:01 carl_baldwin is the assignee 14:36:02 salv-orlando: I must have removed it 14:36:06 salv-orlando: ok we’re good 14:36:08 I 14:36:23 I’ll go over the list after the meeting and see if it’s sane 14:36:30 and reach out to people for help 14:36:35 reach read as plea 14:36:39 I see this one needs a nova spec too: https://blueprints.launchpad.net/neutron/+spec/vm-without-l3-address 14:36:55 I have pointed it out so the authors make sure it's submitted before nova deadlines 14:37:06 yamahata__ ^ 14:37:13 ajo: good point 14:37:19 ok, let’s move on for now 14:37:27 we’ll come back to this 14:37:31 #topic Bugs 14:37:41 * regXboi now wakes up 14:38:18 regXboi: how did it go? 14:38:24 I handed over the swatter to markmcclain yesterday 14:38:31 regXboi: ok 14:38:35 bugs are piling up 14:38:39 and that’s no surprise 14:38:43 armax: pretty well - the bug process docs *seem* to work 14:38:48 were we systematic about triaging this week? 14:38:55 we have some stats on the meeting wiki 14:38:59 #link https://wiki.openstack.org/wiki/Network/Meetings#Bugs 14:39:13 ihrachys: there were a couple I had issues with and so left incomplete 14:39:19 are we an effective team? 14:39:26 ajo: that remains to be seen 14:39:43 regXboi: beware of the drone trying to kill you otherwise (oblivion) 14:39:46 we managed to see and acknowledge most of the bugs 14:40:02 but we don’t fix fast enough 14:40:20 alternatively, we don't review fast enough 14:40:44 I believe we are acknowledging - I think the next step is for folks to be looking at tags 14:40:45 ihrachys: or we’re simply not good enough :) 14:41:02 and trying to push their tags forward 14:41:04 it's also worth excluding rfe bugs from the stats since they are kinda special 14:41:07 ihrachys: fixes need too many rounds to get in the right shape 14:41:11 ihrachys: agreed 14:41:18 ihrachys: +1 14:41:18 especially this time of year 14:41:23 yup 14:41:35 maybe next sprint/virtual sprint can be about bug fixing 14:41:41 exactly. and they don't get fixes in a week. 14:41:51 I have captureed a couple of interesting links on the wiki page 14:42:02 #link https://bugs.launchpad.net/neutron/+bugs?field.status%3Alist=NEW&field.status%3Alist=CONFIRMED&field.tag=gate-failure 14:42:09 i.e. gate failures 14:42:18 confirmed gate failures 14:42:27 #link https://bugs.launchpad.net/neutron/+bugs?field.tag=needs-attention 14:42:32 bugs that would need a second look 14:43:14 a sheet of data pulled out of Launchpad: 14:43:15 https://docs.google.com/spreadsheets/d/1evlDp4tU6aRzBM8VdLjA68MnHBQ_g6lBQ0zWAWX1Gt0/edit#gid=305713688 14:43:43 the latter should help doing better filtering in lieu of Launchpad capabilities 14:44:29 we have deputies for the next couple of weeks 14:44:46 thanks to markmclain and mestery for stepping up 14:45:03 anything else? 14:45:11 I want to raise a bug 14:45:18 we broke flat networking in L 14:45:23 the fix is https://review.openstack.org/#/c/237586/ 14:45:25 ihrachys: ouch 14:45:36 #sadpanda 14:45:47 among other things, it breaks default RDO installer undercloud that uses flat 14:45:54 I'm also having trouble with VLAN networking in L. 14:46:00 but anyway. there is one thing about it 14:46:01 armax, add me after mestery :) 14:46:04 we have two fixes on the bug 14:46:05 ihrachys: I'll target that bug to Liberty as well 14:46:09 ok, I can add myself, sorry, doing 14:46:10 * carl_baldwin ready to step up and find a week, armax 14:46:11 the install guide uses flat, but its linuxbridge 14:46:14 I have seen this in the past 14:46:16 this is only ovs? 14:46:20 emagana looked into it too 14:46:25 the one I posted is created today. I wonder whether it should have continued the original one. 14:46:34 armax: sorry alert didn't fire this morning 14:46:38 but we didn’t enough to go by 14:46:51 Sam-I-Am: ovs only 14:47:08 ok, that explains why i didnt run into it 14:47:24 happy to test if needed though 14:47:25 ihrachys: I gather now we have a good understanding of the root cause? 14:47:33 markmcclain: now worries 14:47:39 armax: yes. I have it in my brain I think. :) 14:47:45 let me know if you need more testing as well 14:47:47 ihrachys: ok cool 14:47:49 armax: I also posted some comments on the original patch 14:48:15 ihrachys: s/flat/vlan ? 14:48:17 so please review, and advice what we do with duplicates like that 14:48:20 ihrachys: quick link? 14:48:35 orgiinal patch: https://review.openstack.org/#/c/225001/ 14:49:24 * ajo needs to run 14:49:26 ihrachys: thanks! 14:49:49 anything else? 14:50:01 ihrachys: fwiw, I thought your comments on the new patch were very appropriate. +1 14:51:08 there's no gating of flat networks? 14:51:13 ihrachys: but is it okay to place an unknown vlan tag with 0? 14:51:19 ihrachys: that’s the part that puzzles me 14:51:26 patch https://review.openstack.org/#/c/237586/ doesn’t seem to do that 14:51:40 Sam-I-Am: there's no gating of any networks 14:51:42 armax: that's smth I need to think more. In theory, 0 is no vlan tag, but it may have differences in how it flows in ovs agent. 14:51:50 so the last one seems the better fix? 14:51:59 Sam-I-Am: without multinode at least 14:52:00 typically an untagged vlan has no vlan id 14:52:01 yes, 237586 seems more safe 14:52:26 Sam-I-Am: apparently no gating 14:52:28 ihrachys: so we should supersed 225001 14:52:32 hmmm 14:52:34 * gongysh feel bad Liberty breaks the flat mode. 14:52:47 kevinbenton ihrachys we should talk about these things 14:52:55 Sam-I-Am: or maybe it's because it's done only on agent start, and we don't restart it in gate 14:52:59 ihrachys: do we know the breaking commit? 14:53:06 ihrachys: oh, interesting 14:53:24 armax: yes, it's ryu support one I think 14:53:29 it was mentioned in the bug 14:53:45 ihrachys: ok, I haven’t gone through the comments yet 14:53:53 ihrachys: let’s circle back on this quickly I guess 14:54:00 the breaking one: https://review.openstack.org/#/c/153946/ 14:54:06 Liberty may have also messed up VLAN networks as per claudiub's bug 14:54:07 aye, out of meeting 14:54:18 wow. nice work folks 14:54:21 :D 14:54:30 kevinbenton: I don't think so, we have a fullstack connectivity test that uses VLANs 14:54:30 kevinbenton: link? 14:54:30 that was bound to break something! 14:54:33 that tends to pass 14:54:54 ihrachys: https://bugs.launchpad.net/neutron/+bug/1507776 14:54:54 Launchpad bug 1507776 in neutron "Wrong OVS flows created for new networks" [Undecided,New] 14:55:00 amuller: afaik claudiub is interested in ovs on hyper-v, maybe that's what we don't gate ;) 14:55:06 this must be it: https://review.openstack.org/#/c/153946/136/neutron/plugins/ml2/drivers/openvswitch/agent/ovs_neutron_agent.py 14:55:35 ihrachys: actually, this wasn't run with ovs on hyper-v 14:55:43 armax: heh, i think so 14:55:55 such a small change :) 14:55:55 ihrachys: classic ovs-agent + neutron-hyperv-agent 14:56:06 yamamoto: any input? 14:56:08 claudiub: which platform under ovs? 14:56:18 ubuntu 14.04 14:56:33 is this a bug in the OVS agent or not? Why would the nova hypervisor matter 14:56:37 ok, let’s continue bug conversation in channel 14:56:50 the hypervisor doesn't matter 14:57:08 I've stated in the bug message that, if the neutron-ovs-agent is restarted, the flows are recreated properly 14:57:10 we got 4 minutes left 14:57:16 yep :/ 14:57:18 and the traffic is ok 14:57:21 armax: need to leave.. another meeting waiting in the schedule.. 14:57:33 armax: i have 1 docs request 14:57:34 armax: I believe we'll need to handle open agenda in #neutron 14:57:44 well, many, but one for here :) 14:57:51 emagana: ok…let’s continue the conversation in channel for now and in person next week for the folks who are going to be in tokyo 14:58:00 Sam-I-Am: shoot 14:58:13 amuller: how does full stack simulate two connected agents via a physical interface? 14:58:20 right now, the networking guide scenarios with L3 only support attaching VMs to private networks 14:58:45 i've seen interest in hybrid implementations that allow connection to both public/provider and private networks 14:59:02 i'm wondering if i should add this capability to the existing scenarios, or make new hybrid scenarios 14:59:09