21:02:16 #startmeeting Networking 21:02:17 Meeting started Mon Jul 8 21:02:16 2013 UTC. The chair is markmcclain. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:02:18 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:02:20 The meeting name has been set to 'networking' 21:02:33 #link https://wiki.openstack.org/wiki/Network/Meetings 21:02:47 #topic Announcements 21:03:32 I think we mostly survived the rename in one piece. Thanks to the infra team transitioning the repositories and CI infrastructure on Saturday. 21:03:43 hi 21:03:47 hi 21:03:48 hi! 21:04:01 markmcclain: sorry late 21:04:05 hi 21:04:16 hello all 21:04:47 If you haven't already make sure to update your git configs to point at the new repo locations 21:05:22 hi 21:05:25 #link https://launchpad.net/neutron/+milestone/havana-2 21:06:21 H2 is right around the corner. All blueprints for H2 need to be proposed by Thursday, so that we we have time to review them before the branch is cut. 21:07:05 We've only merged 25% of the targeted blueprints, so I expect the next week to busy 21:07:54 If you have code out for review, you will need to rebase it to pick up the name changes otherwise it will fail on merging 21:08:11 #topic renaming 21:08:40 We removed most references to quantum with the code itself over the weekend. 21:09:23 markmcclain: terrific job!!! 21:09:25 The code passes unit tests, but we are unable to validate the vendor plugins. I've reached out to the plugin contacts to ask they test their plugins and report back. 21:09:32 hi all 21:10:28 emagana: thanks 21:10:51 I want to ask if someone is experiencing this devstack problem after renaming: https://bugs.launchpad.net/devstack/+bug/1199126 21:10:52 Launchpad bug 1199126 in devstack "starting neutron agent fails" [Undecided,New] 21:11:15 I ran a quick check on the Cisco plugin on devstack and didn't notice any problems. 21:11:31 HenryG: good to know 21:12:01 emagana: looks like a devstack bug 21:12:04 when we publish a new neutron client 21:12:07 ? 21:12:48 gongysh: the NeutronClient (2.2.4) that was published last week was equivalent to quantumclient 2.2.3 21:13:05 ok 21:13:14 I'd like to get the rest of the integrated projects switched 21:13:18 script is not able to find this function: neutron_plugin_configure_plugin_agent 21:13:29 and then we can push 2.2.5 with newer features 21:14:32 Any questions about renaming? 21:14:48 when we update doc? 21:15:05 hope they were done too. 21:15:08 gongysh: emagana started a bit of that last week 21:15:15 gongysh: I am working on them! 21:15:22 ok 21:15:34 submitter should do (1) Rebase the patch to the neutron (2) replace s/Quantum/Neutron/ s/quantum/neutron/ in the newer added part ? 21:15:57 emagana: oh the rumor get truth :) 21:16:12 nati_ueno: that's one part of it. In general look for conflict and solve them just like with any other patch 21:16:22 nati_ueno: ;-) 21:16:35 salv-orlando: I got it. IMO, we should have wiki for this instructions 21:16:39 nati_ueno: text substitution works for most cases, but there a few plugins were quantum_* is actually exchanged with the backend 21:16:51 nati_ueno: It's really quite easy, any errors will be caught in flake8 or UT. 21:17:05 markmcclain: thanks for fixing in the plugins as well 21:17:08 mestery: I see! good point! 21:17:28 yep, the only issues I had so far are with imports. But it's easy to pick out which ones you added/changed/removed with git diff 21:17:28 make sure you use "git mv" when you rename files/directories 21:17:42 else it will show up as new files in existing patches 21:18:00 SumitNaiksatam: most of that should have been taken care of by the big rename this weekend 21:18:05 ok I would like to write wiki page for this. Or Should I add this information for existing page? 21:18:19 nati_ueno: you can add instructions on the existing 21:18:23 I vote for adding to the existing page. 21:18:33 markmcclain: I got it 21:18:47 markmcclain: i was referring to patches in review which introduce new files 21:19:51 right.. I've found that rebasing generally does the right thing 21:20:11 SumitNaiksatam: reagrding patches, rebase will address renaming issues. 21:20:39 at least, it did not happen in my case 21:20:56 when you add new files, "git mv" will help you. 21:21:01 SumitNaiksatam: I had to manually use "git mv" to move files in one of my patchsets which was new in my patchset. 21:21:52 Last thing on renaming.. if you find a bug make sure to file it in launchpad so that we can track them and prioritize them 21:21:56 amotoki, mestery: yeah, you need to do a "git mv" for new files (i noticed lots of new files in patches that were in review, probably not doing a git mv) 21:21:57 #topic bugs 21:22:25 We've this bug still to close: https://bugs.launchpad.net/neutron/+bug/1184484 21:22:28 Launchpad bug 1184484 in tripleo "Quantum default settings will cause deadlocks due to overflow of sqlalchemy_pool" [Critical,Triaged] 21:22:43 lifeless: is the bug you wanted to discuss? 21:22:45 markmcclain - does the oslo db merge take care of that? 21:22:46 I am working on it. Expecting to have a fix or an headache tomorrow 21:22:55 updated https://wiki.openstack.org/wiki/Network/neutron-renaming#How_to_rename_existing_patches 21:22:59 markmcclain: it is one of two. 21:23:04 salv-orlando: thanks for looking into it 21:23:11 lifeless: what is the other one? 21:23:12 garyk: I am trying with the new modelbase 21:23:15 markmcclain: bug 1189385 21:23:16 Launchpad bug 1189385 in tripleo "quantum-server hung up it's listening port" [Critical,Triaged] https://launchpad.net/bugs/1189385 21:23:24 markmcclain: we've supplied all the info asked for 21:23:30 markmcclain: (AFAICT) 21:23:31 salv-orlando ok thanks 21:23:31 hopefully that will be solution otherwise, we might have a bad problem with sqlalchemy leaking sessions 21:24:02 for bug 1189385, all that I can see is that is seems unrelated to but 1184484 21:24:40 lifeless: thanks for adding the tracebacks… I'll dig into 1189385 21:24:46 markmcclain we also should discuss the sporadic failures of tempest 21:24:54 https://bugs.launchpad.net/nova/+bug/1186161 21:24:55 Launchpad bug 1186161 in nova "Tempest tests sometimes get: [Fail] Couldn't ping server" [Undecided,New] 21:25:00 garyk: yes can do that here 21:25:12 this check fails in the gate randomly 21:25:22 and running the gate again it is likely to pass 21:25:32 markmcclain: thank you! 21:25:57 garyk: I know you were looking into 1186161 have found anything of interest? 21:27:16 markmcclain - i ahve yet been able to reproduce this when i run tempest 21:27:42 ok 21:27:51 markmcclain - i will continue to look into it 21:28:02 thanks 21:28:15 Any other high priority bugs to discuss? 21:28:46 markmcclain - https://bugs.launchpad.net/nova/+bug/1192287 21:28:47 Launchpad bug 1192287 in nova "Creating server did not fail when exceeded Quantum quota limit" [Undecided,New] 21:29:12 i think that this is more nova related than quantum. the quotas check is done on the compute node instead of at the api level 21:29:28 right 21:29:35 arosen1: we're you working on this too? 21:30:06 markmcclain, arosen1: i can look into this if you guys want 21:30:18 I think this is similar bug 1172808 21:30:19 Launchpad bug 1172808 in nova "Nova fails on Quantum port quota too late" [Wishlist,Triaged] https://launchpad.net/bugs/1172808 21:30:25 yea ^ 21:30:33 I think this goes back to the 'fail fast' discussions we had early in the release cycle 21:30:35 we need to move the port creation to nova-api to fix that. 21:30:39 year that thing 21:31:19 any objections if i take it? 21:31:22 arosen1 has a blueprint for it; but community had mixed feelings about it. 21:31:51 https://blueprints.launchpad.net/nova/+spec/nova-api-quantum-create-port 21:33:15 cool.. we probably link the bugs and blueprint to prevent more duplicate entires 21:33:16 garyk: it's a big change. I was planing on bitting this off next but if you want it, it's all yours :) 21:33:52 arosen1: i'll take a look at the bp and let you know 21:34:00 garyk: k 21:34:20 Ok let's run through the sub teams real quick 21:34:23 #topic API 21:34:27 hi 21:34:34 salv-orlando: look like we're in good shape for H2 21:34:54 very quickly: we are risking to miss the H-2 deadline for service type refactoring 21:35:04 because of new concerns around design 21:35:17 enikanorov is going to push this to the mailing list soon. 21:35:26 please note that we already have patches dependent on it 21:36:00 and that with 'service types' providers or whatever we want to call them there's a very thin line between a productive discussion and speculation 21:36:24 so let's get that to what matters for our goals, and ensure its merged by H-2 deadline! 21:36:27 that's all 21:37:17 enikanorov said he will support full db based implementation in hanava, so I'm going to +2 for the patch after checking it again 21:37:47 ok.. looks like we have enough cores on it 21:38:01 Anything else besides this blocker? 21:38:02 nati_ueno I am not sure about you mean with 'full db based', but I think we need to give room to next topic 21:38:07 nope, that is all 21:38:08 so salv-orlando and akihiro looks already +1 or +2 for that patch, I'm expecting that patch will land soon 21:38:18 #topic VPN 21:38:27 nati_ueno: hi 21:38:48 Regarding to service type discussion, we decided to remove vpn_type 21:39:02 We will add support for service type framework in H3 21:39:16 ok 21:39:26 One news is how to test page have horizon ui instruction also 21:39:45 https://wiki.openstack.org/wiki/Quantum/VPNaaS/HowToInstall 21:40:03 great.. minus pep8 errors.. like folks should review VPN 21:40:17 Yes since h2 end is close, 21:40:41 so it is great if we could have cores on REST API and Driver patch 21:41:04 I'm reviewing REST API so one additional for REST API and 2 core for driver patch 21:41:22 I'll be one of the cores on this 21:41:33 markmcclain: Thanks! 21:41:43 markmcclain: if you cannot take both patches, I can do the api one 21:42:09 salv-orlando: cool.. I'll let you have one and I'll review the driver 21:42:10 salv-orlando: Thanks! 21:42:27 I need one more for drivers 21:42:44 i will look the driver one. 21:42:49 amotoki: Thanks! 21:42:51 amotoki: thanks 21:43:05 Anything else on VPN? 21:43:19 markmcclain: salv-orlando: amotoki: OK We are renaming code. After we check it working with new name, we will ping you guys 21:43:29 markmcclain: That's all from us 21:43:29 k 21:43:39 nati_ueno: sounds good 21:43:40 Thank you for your help 21:43:42 #topic firewalling 21:44:09 SumitNaiksatam: hi 21:44:12 hi 21:44:17 API, Agent and Driver patches have been rebased with the name change. CLI patch still needs to be rebased. 21:44:29 API and Driver patches are out of WIP. Unit tests are in the works on the agent patch, hence it's still WIP. 21:44:39 thanks for getting them rebased so quickly 21:44:40 RajeshMohan, who is the author of the driver patch, is traveling to India today, so expect some delays on review responses at his end. 21:44:53 ok 21:44:55 I need to add the db migration script to the API patch. 21:45:07 SridarK do you have anything more to add on the agent work? 21:45:28 Nothing more - will get UT done by Wed 21:45:46 in general, there have been cores on the API patch (thanks Salvatore, amotoki, arisen, others) 21:45:59 we need more on the driver and agent patch (when its ready) 21:46:10 arisen -> arosen 21:46:11 I'll review them :) 21:46:21 thanks nati_ueno !! :-) 21:46:38 great 21:46:41 thats our update 21:46:47 SumitNaiksatam: Thank you for your sharing test instruction. It is great to if we could have wiki page for that 21:46:53 thanks for updating 21:46:58 #topic ml2 21:47:02 nati_ueno: its on the wiki page 21:47:07 mestery: hi 21:47:11 SumitNaiksatam: Ah sorry. Thanks 21:47:12 markmcclain: hi 21:47:28 ML2 changes are progressing nicely, though it's unclear if we'll get much more merged before H2 ends. 21:47:39 I think the GRE and VXLAN type drivers have a good chance if we can get some core reviewers. 21:47:46 And the MechanismDriver change as well. 21:48:04 I took a look at the mech driver this afternoon 21:48:05 Also, I'm hoping to get gongysh to review this one again https://review.openstack.org/#/c/33107/ (I added a link to a design document there). 21:48:23 sorry for the delay to look at it. 21:48:27 I will do it today 21:48:32 Other than that, we have a sub-team meeting Wednesday to discuss in detail ML2. 21:48:38 gongysh: will you be able to look an unlblock? 21:48:39 gongysh: Thanks! I plan to work on that tomorrow. 21:48:47 gongysh: thanks 21:48:56 That's it for ML2. 21:49:16 so this https://blueprints.launchpad.net/neutron/+spec/ml2-portbinding 21:49:19 will miss H2? 21:49:37 I think so, I haven't seen any review for it yet. 21:49:45 rkukura was going to try to get a review out early this week though. 21:49:54 ok.. I'll hold it in for now 21:50:01 Thanks! 21:50:05 mestery: thanks for updating 21:50:12 #topic lbaas 21:50:24 hi 21:50:40 the agent scheduling patch isstill waiting for cores 21:51:04 nothing has changed from past week except now the patch depends on another one 21:51:14 which divides agent scheduling to dhcp & l3 21:51:25 (and lbaas, in lbaas-related patch) 21:51:47 gongysh: can you take a look at this one: https://review.openstack.org/#/c/35266/ 21:51:50 so those two reviews require core's attention 21:52:16 I'll review it also 21:52:26 sure 21:52:32 nati_ueno, gongysh: thanks 21:52:39 enikanorov: anything else? 21:52:42 that's all from lbaas for h2 21:52:57 enikanorov: thanks for updating 21:53:08 #topic nova 21:53:19 we have spoken about bugs above 21:53:20 garyk: I think we covered most of the pressing items earlier 21:53:44 we had same thought :) 21:53:47 #topic stable 21:53:50 markmcclain agreed. there is one issue though that it would be nice if quantum guys could look at (1 sec) 21:53:56 https://review.openstack.org/#/c/34968/ 21:53:59 nova side 21:54:11 https://review.openstack.org/#/c/33054/ 21:54:57 gongysh - i think that this was reverte in nova as they felt it was the cause f the tempest problems 21:55:23 no, not this one. 21:55:23 monkeypatching eventlet causes some problems deep down with httplib2 21:55:42 it is the one with shared quantum client. 21:55:56 ok 21:56:25 I'll look at the token issue because we've similar stuff internally 21:56:40 #topic stable 21:56:49 no updates here 21:57:01 Now that we've renamed the code the backports are going to get much harder 21:57:12 a few people have been dong backports (sorry i do not recall the names - but thanks!) 21:57:24 we're also farther into the dev cycle, so might need to raise the bar for what we allow as a backport 21:58:01 hopefully neutron will have fewer problems than quantum 21:58:10 :) 21:58:18 garyk: Heh :) 21:58:50 #topic Horizon 21:58:57 hi 21:58:58 amotoki: anything to share? 21:59:13 markmcclain: is renaming patch from you ready for reivew? 21:59:18 yes 21:59:32 thank. will check it. 21:59:40 great! 21:59:55 for H2, security group support is the only one related to neutron. 22:00:25 cool.. so most of the services will land in H3? 22:00:35 i think so. 22:00:42 vpnass has a good progress. 22:01:13 i have no concerns for fwaas too. 22:01:17 good to hear 22:01:30 KC is working on the CLI, but should be on FWaaS shortly 22:01:50 SumitNaiksatam: good to hear 22:01:59 all from me. 22:02:08 Any questions on Horizon<>Neutron integration? 22:02:29 #topic Open Discussion 22:02:54 I will begin the old work on multihost feature this week. 22:03:03 i am going to crash. good night. have a good one 22:03:09 garyk: night 22:03:11 garyk: night! 22:03:13 garyk: night! 22:03:13 since the host-id is in. 22:03:17 gongysh: this will be h3? 22:03:22 garky: dream. 22:03:26 yes 22:04:31 Anything other items to cover? 22:04:36 and about concurrent lock 22:04:42 this is a big problem 22:05:53 overlapping cidr is just a case of them. 22:05:58 gongysh: that's a db issue 22:06:08 gongysh: a concurrent lock is not required 22:06:14 no, that is a business logic problem. 22:06:29 gongysh: yes, one that needs to be enforced at the db level 22:06:31 not all problem can be fixed by a constraint. 22:06:32 gongysh: not in code 22:07:07 gongysh: this isn't a hard problem. it's been done by db's since the 1970s 22:07:35 haha, 22:08:07 gongysh: if you are not aware of this, I respectfully suggest that you are not the best person to address the problem. There is a reason I assigned the bugs to markmcclain. 22:08:12 ok, what is your opinion of the overlapping cidr checking. 22:09:30 Which bug are you discussing? 22:09:46 We're out of time.. if gongysh, marun want to hop into #openstack-neutron we can talk about this more 22:10:01 ok 22:10:19 and those that are interested can join us there too 22:10:59 nati_ueno: Bug #1194206 22:11:02 Launchpad bug 1194206 in neutron "allow_overlapping_ips=False fails for Concurrent requests" [High,In progress] https://launchpad.net/bugs/1194206 22:11:04 gongysh: thanks 22:11:04 Thanks for everyone who stepped up to review important patches for H2 22:11:09 #endmeeting