| alexpilotti | rkukura: we have an Hyper-V patch that got abandoned after approval due to rebasing issues | 00:00 |
|---|---|---|
| alexpilotti | rkukura: as the commit that merged in the meantime solved the issue that the failing patch was meant to solve, there was no need for it anymore | 00:01 |
| alexpilotti | rkukura: https://review.openstack.org/#/c/63840/3 | 00:01 |
| alexpilotti | rkukura: but we still need it in Havana | 00:01 |
| alexpilotti | rkukura: how can we do that, as normally when we do a backport, we alway refer to a patch in master! | 00:01 |
| *** spandhe has quit IRC | 00:02 | |
| alexpilotti | rkukura: as a side note, the reason why the patch is not needed in master is that the security groups blueprint code replaced the code where the issue was happening | 00:02 |
| *** Trozz has joined #openstack-neutron | 00:03 | |
| alexpilotti | rkukura: but we cannot backport the security groups blueprint, so we still need to backport the failed patch :-) | 00:03 |
| *** spandhe has joined #openstack-neutron | 00:03 | |
| *** banix has joined #openstack-neutron | 00:05 | |
| rkukura | alexpilotti: I think a stable-branch-specific patch is acceptible with this sort of justification. Just make sure thats clear in the commit message, and reference the patch fixing the issue on master. | 00:12 |
| alexpilotti | rkukura: ok tx! | 00:12 |
| *** gdubreui has joined #openstack-neutron | 00:17 | |
| *** krtaylor has joined #openstack-neutron | 00:20 | |
| *** mestery has quit IRC | 00:23 | |
| *** alagalah has quit IRC | 00:25 | |
| *** matsuhashi has joined #openstack-neutron | 00:27 | |
| *** rkukura has quit IRC | 00:28 | |
| *** singhs has quit IRC | 00:28 | |
| *** thuc has joined #openstack-neutron | 00:29 | |
| *** banix has quit IRC | 00:31 | |
| *** manishg has quit IRC | 00:32 | |
| *** morganfainberg is now known as morganfainberg_Z | 00:33 | |
| *** thuc has quit IRC | 00:33 | |
| *** yamahata has quit IRC | 00:34 | |
| *** SumitNaiksatam has quit IRC | 00:36 | |
| *** beagles has quit IRC | 00:40 | |
| *** banix has joined #openstack-neutron | 00:56 | |
| *** Jianyong has joined #openstack-neutron | 00:58 | |
| openstackgerrit | A change was merged to openstack/python-neutronclient: Show the unknown auth stratey in neutron client https://review.openstack.org/82834 | 00:59 |
| *** armitage81 has quit IRC | 01:03 | |
| *** tomoe_ has joined #openstack-neutron | 01:09 | |
| *** otherwiseguy has joined #openstack-neutron | 01:09 | |
| *** changbl has joined #openstack-neutron | 01:12 | |
| *** matsuhashi has quit IRC | 01:12 | |
| *** matsuhashi has joined #openstack-neutron | 01:14 | |
| *** tomoe_ has quit IRC | 01:19 | |
| *** spandhe has quit IRC | 01:30 | |
| *** ramishra has joined #openstack-neutron | 01:30 | |
| *** dave_tucker_zzz is now known as dave_tucker | 01:32 | |
| *** krtaylor has quit IRC | 01:34 | |
| *** devlaps1 has quit IRC | 01:36 | |
| *** _cjones_ has quit IRC | 01:37 | |
| *** catohornet__ has joined #openstack-neutron | 01:43 | |
| *** singhs has joined #openstack-neutron | 01:45 | |
| *** alexpilotti has quit IRC | 01:47 | |
| *** xuhanp has joined #openstack-neutron | 01:53 | |
| *** otherwiseguy has quit IRC | 01:55 | |
| *** dfarrell07 has joined #openstack-neutron | 01:58 | |
| *** yamahata has joined #openstack-neutron | 01:58 | |
| openstackgerrit | Henry Gessau proposed a change to openstack/neutron: Remove auto-generation of db schema from models at startup https://review.openstack.org/40296 | 01:59 |
| openstackgerrit | Li Ma proposed a change to openstack/neutron: Race condition of L3-agent to add/remove routers https://review.openstack.org/73234 | 02:00 |
| *** flwang has quit IRC | 02:00 | |
| *** tomoe_ has joined #openstack-neutron | 02:00 | |
| *** matsuhashi has quit IRC | 02:00 | |
| xuhanp | amotoki, ping | 02:04 |
| *** singhs has quit IRC | 02:07 | |
| *** matsuhashi has joined #openstack-neutron | 02:08 | |
| *** tomoe_ has quit IRC | 02:10 | |
| *** dave_tucker is now known as dave_tucker_zzz | 02:10 | |
| *** thuc has joined #openstack-neutron | 02:11 | |
| *** otherwiseguy has joined #openstack-neutron | 02:12 | |
| *** xianghui has joined #openstack-neutron | 02:22 | |
| *** harlowja is now known as harlowja_away | 02:27 | |
| *** thuc has quit IRC | 02:27 | |
| *** thuc has joined #openstack-neutron | 02:28 | |
| *** thuc has quit IRC | 02:32 | |
| *** catohornet___ has joined #openstack-neutron | 02:33 | |
| *** catohornet__ has quit IRC | 02:35 | |
| *** singhs has joined #openstack-neutron | 02:37 | |
| *** krtaylor has joined #openstack-neutron | 02:38 | |
| *** singhs_ has joined #openstack-neutron | 02:40 | |
| *** flwang has joined #openstack-neutron | 02:41 | |
| *** singhs has quit IRC | 02:41 | |
| *** singhs_ is now known as singhs | 02:41 | |
| *** nati_ueno has quit IRC | 02:42 | |
| *** catohornet___ has quit IRC | 02:44 | |
| *** tomoe_ has joined #openstack-neutron | 02:45 | |
| *** ramishra has quit IRC | 02:45 | |
| *** ramishra has joined #openstack-neutron | 02:45 | |
| openstackgerrit | enikanorov proposed a change to openstack/neutron: Fix namespace exist() method https://review.openstack.org/81537 | 02:48 |
| *** baoli has quit IRC | 02:50 | |
| *** ramishra has quit IRC | 02:50 | |
| *** manishg has joined #openstack-neutron | 02:54 | |
| *** ramishra has joined #openstack-neutron | 02:54 | |
| *** carl_baldwin has joined #openstack-neutron | 02:55 | |
| *** tomoe_ has quit IRC | 02:56 | |
| *** manishg_ has joined #openstack-neutron | 02:57 | |
| *** manishg has quit IRC | 02:58 | |
| *** manishg has joined #openstack-neutron | 02:59 | |
| *** manishg_ has quit IRC | 03:01 | |
| *** ramishra has quit IRC | 03:02 | |
| *** manishg_ has joined #openstack-neutron | 03:02 | |
| *** ramishra has joined #openstack-neutron | 03:02 | |
| *** devlaps has joined #openstack-neutron | 03:03 | |
| *** manishg has quit IRC | 03:04 | |
| *** manishg has joined #openstack-neutron | 03:05 | |
| *** banix has quit IRC | 03:07 | |
| *** manishg_ has quit IRC | 03:07 | |
| *** ramishra has quit IRC | 03:07 | |
| *** manishg_ has joined #openstack-neutron | 03:08 | |
| *** tomoe_ has joined #openstack-neutron | 03:09 | |
| *** manishg has quit IRC | 03:10 | |
| *** manishg has joined #openstack-neutron | 03:11 | |
| *** BuSerD has quit IRC | 03:12 | |
| *** manishg_ has quit IRC | 03:13 | |
| *** emagana has joined #openstack-neutron | 03:13 | |
| *** manishg_ has joined #openstack-neutron | 03:14 | |
| *** matsuhashi has quit IRC | 03:15 | |
| *** manishg has quit IRC | 03:16 | |
| *** manishg has joined #openstack-neutron | 03:17 | |
| *** thuc has joined #openstack-neutron | 03:17 | |
| *** manishg_ has quit IRC | 03:19 | |
| *** thuc has quit IRC | 03:19 | |
| *** thuc has joined #openstack-neutron | 03:20 | |
| *** manishg_ has joined #openstack-neutron | 03:20 | |
| *** baoli has joined #openstack-neutron | 03:20 | |
| *** Matt1 has quit IRC | 03:21 | |
| *** manishg has quit IRC | 03:22 | |
| *** manishg has joined #openstack-neutron | 03:23 | |
| *** emagana has quit IRC | 03:24 | |
| *** manishg_ has quit IRC | 03:25 | |
| *** manishg_ has joined #openstack-neutron | 03:26 | |
| *** sbalukoff has quit IRC | 03:26 | |
| *** sbalukoff has joined #openstack-neutron | 03:27 | |
| *** thuc has quit IRC | 03:27 | |
| *** tomoe_ has quit IRC | 03:27 | |
| *** manishg has quit IRC | 03:28 | |
| *** manishg has joined #openstack-neutron | 03:28 | |
| *** baoli has quit IRC | 03:29 | |
| *** manishg_ has quit IRC | 03:30 | |
| *** manishg_ has joined #openstack-neutron | 03:31 | |
| *** manishg has quit IRC | 03:33 | |
| *** manishg has joined #openstack-neutron | 03:34 | |
| *** tomoe_ has joined #openstack-neutron | 03:35 | |
| *** manishg_ has quit IRC | 03:36 | |
| *** manishg_ has joined #openstack-neutron | 03:37 | |
| *** manishg has quit IRC | 03:38 | |
| *** emagana has joined #openstack-neutron | 03:39 | |
| *** manishg has joined #openstack-neutron | 03:40 | |
| *** manishg_ has quit IRC | 03:42 | |
| *** manishg_ has joined #openstack-neutron | 03:43 | |
| amotoki | xuhanp: pong | 03:43 |
| xuhanp | amotoki, Can I add you as the reviewer of my patch related to IPv6 security group? we (ipv6 sub team) hope this can be the next patch of IPv6 changes | 03:44 |
| *** manishg has quit IRC | 03:44 | |
| xuhanp | amotoki, here is the link https://review.openstack.org/#/c/72252/ | 03:45 |
| *** tomoe_ has quit IRC | 03:45 | |
| amotoki | xuhanp: sure | 03:45 |
| xuhanp | amotoki, thanks a lot! | 03:46 |
| *** ramishra has joined #openstack-neutron | 03:46 | |
| *** manishg has joined #openstack-neutron | 03:46 | |
| *** manishg_ has quit IRC | 03:47 | |
| amotoki | xuhanp: Do we need to target this bug to rc1? | 03:48 |
| *** thuc has joined #openstack-neutron | 03:48 | |
| xuhanp | amotoki, we hope so. But depends on how many comments we get | 03:48 |
| *** manishg_ has joined #openstack-neutron | 03:49 | |
| *** spandhe has joined #openstack-neutron | 03:49 | |
| amotoki | xuhanp: okay. i added it to my review list. | 03:51 |
| *** mestery has joined #openstack-neutron | 03:51 | |
| *** emagana has quit IRC | 03:51 | |
| xuhanp | amotoki, perfect. Thanks again | 03:51 |
| *** manishg has quit IRC | 03:51 | |
| *** armax has joined #openstack-neutron | 03:52 | |
| *** manishg has joined #openstack-neutron | 03:52 | |
| *** carl_baldwin has quit IRC | 03:53 | |
| *** spandhe_ has joined #openstack-neutron | 03:54 | |
| *** spandhe has quit IRC | 03:54 | |
| *** spandhe_ is now known as spandhe | 03:54 | |
| *** manishg_ has quit IRC | 03:54 | |
| *** carl_baldwin has joined #openstack-neutron | 03:55 | |
| *** manishg_ has joined #openstack-neutron | 03:55 | |
| *** manishg has quit IRC | 03:56 | |
| *** manishg has joined #openstack-neutron | 03:58 | |
| *** manishg_ has quit IRC | 03:59 | |
| *** manishg_ has joined #openstack-neutron | 04:00 | |
| *** nati_ueno has joined #openstack-neutron | 04:02 | |
| *** manishg has quit IRC | 04:02 | |
| *** thuc_ has joined #openstack-neutron | 04:03 | |
| *** manishg has joined #openstack-neutron | 04:03 | |
| *** manishg_ has quit IRC | 04:05 | |
| *** thuc has quit IRC | 04:06 | |
| *** manishg has quit IRC | 04:08 | |
| *** tomoe_ has joined #openstack-neutron | 04:08 | |
| *** manishg has joined #openstack-neutron | 04:14 | |
| *** manishg_ has joined #openstack-neutron | 04:17 | |
| *** manishg has quit IRC | 04:18 | |
| *** SumitNaiksatam has joined #openstack-neutron | 04:18 | |
| *** armax has quit IRC | 04:21 | |
| *** crc32 has quit IRC | 04:24 | |
| *** manishg_ has quit IRC | 04:24 | |
| *** matsuhashi has joined #openstack-neutron | 04:25 | |
| *** SumitNaiksatam has quit IRC | 04:25 | |
| *** mestery has quit IRC | 04:25 | |
| *** baoli has joined #openstack-neutron | 04:25 | |
| *** baoli has quit IRC | 04:33 | |
| *** tomoe_ has quit IRC | 04:40 | |
| *** tomoe_ has joined #openstack-neutron | 04:41 | |
| *** tomoe_ has quit IRC | 04:41 | |
| *** tomoe_ has joined #openstack-neutron | 04:42 | |
| *** carl_baldwin has quit IRC | 04:48 | |
| *** yfried has quit IRC | 04:53 | |
| openstackgerrit | Akihiro Motoki proposed a change to openstack/neutron: nec plugin: allow to delete resource with ERROR status https://review.openstack.org/82143 | 04:56 |
| *** sacharya has joined #openstack-neutron | 04:57 | |
| *** WackoRobie has quit IRC | 04:58 | |
| *** carl_baldwin has joined #openstack-neutron | 04:58 | |
| *** WackoRobie has joined #openstack-neutron | 04:58 | |
| *** carl_baldwin has quit IRC | 05:01 | |
| *** WackoRobie has quit IRC | 05:03 | |
| *** ramishra_ has joined #openstack-neutron | 05:04 | |
| *** ramishra has quit IRC | 05:07 | |
| *** Longgeek has joined #openstack-neutron | 05:11 | |
| *** Longgeek has quit IRC | 05:12 | |
| *** Longgeek has joined #openstack-neutron | 05:12 | |
| openstackgerrit | Li Ma proposed a change to openstack/neutron: Race condition of L3-agent to add/remove routers https://review.openstack.org/73234 | 05:16 |
| *** dguitarbite has joined #openstack-neutron | 05:17 | |
| *** Longgeek_ has joined #openstack-neutron | 05:19 | |
| *** Longgee__ has joined #openstack-neutron | 05:20 | |
| *** tomoe_ has quit IRC | 05:21 | |
| *** skraynev_afk is now known as skraynev | 05:22 | |
| *** Longgeek has quit IRC | 05:23 | |
| *** Longgeek_ has quit IRC | 05:24 | |
| *** thuc_ has quit IRC | 05:26 | |
| *** thuc has joined #openstack-neutron | 05:27 | |
| *** flwang1 has joined #openstack-neutron | 05:27 | |
| *** tomoe_ has joined #openstack-neutron | 05:27 | |
| *** flwang has quit IRC | 05:29 | |
| *** baoli has joined #openstack-neutron | 05:30 | |
| *** thuc has quit IRC | 05:31 | |
| *** baoli has quit IRC | 05:34 | |
| *** armax has joined #openstack-neutron | 05:36 | |
| *** irenab has joined #openstack-neutron | 05:36 | |
| *** Longgee__ has quit IRC | 05:39 | |
| *** emagana has joined #openstack-neutron | 05:40 | |
| *** devlaps has quit IRC | 05:43 | |
| *** nati_ueno has quit IRC | 05:57 | |
| *** WackoRobie has joined #openstack-neutron | 05:59 | |
| *** Jabadia has joined #openstack-neutron | 06:00 | |
| *** WackoRobie has quit IRC | 06:00 | |
| *** WackoRobie has joined #openstack-neutron | 06:01 | |
| *** sridhar has joined #openstack-neutron | 06:03 | |
| *** WackoRobie has quit IRC | 06:05 | |
| *** rkukura has joined #openstack-neutron | 06:06 | |
| *** Longgeek has joined #openstack-neutron | 06:10 | |
| *** bvandenh has quit IRC | 06:13 | |
| *** yfried has joined #openstack-neutron | 06:14 | |
| *** Longgeek has quit IRC | 06:15 | |
| *** saju_m has joined #openstack-neutron | 06:19 | |
| *** otherwiseguy has quit IRC | 06:25 | |
| openstackgerrit | Jenkins proposed a change to openstack/neutron: Imported Translations from Transifex https://review.openstack.org/82435 | 06:29 |
| *** matsuhashi has quit IRC | 06:33 | |
| *** gdubreui has quit IRC | 06:34 | |
| *** matsuhashi has joined #openstack-neutron | 06:35 | |
| *** thuc has joined #openstack-neutron | 06:37 | |
| *** spandhe has quit IRC | 06:41 | |
| *** thuc has quit IRC | 06:42 | |
| *** spandhe has joined #openstack-neutron | 06:43 | |
| *** armax has quit IRC | 06:47 | |
| *** jlibosva has joined #openstack-neutron | 06:49 | |
| *** kedarkul has joined #openstack-neutron | 06:56 | |
| *** WackoRobie has joined #openstack-neutron | 06:59 | |
| *** Longgeek has joined #openstack-neutron | 07:03 | |
| *** matsuhas_ has joined #openstack-neutron | 07:03 | |
| *** flwang1 has quit IRC | 07:03 | |
| *** WackoRobie has quit IRC | 07:03 | |
| *** matsuhas_ has quit IRC | 07:04 | |
| *** singhs has quit IRC | 07:05 | |
| *** flwang has joined #openstack-neutron | 07:06 | |
| *** matsuhas_ has joined #openstack-neutron | 07:06 | |
| *** matsuhashi has quit IRC | 07:07 | |
| *** saju_m has quit IRC | 07:13 | |
| *** tomoe_ has quit IRC | 07:13 | |
| *** afazekas has joined #openstack-neutron | 07:21 | |
| *** rotbeard has joined #openstack-neutron | 07:21 | |
| *** tomoe_ has joined #openstack-neutron | 07:22 | |
| *** sacharya has quit IRC | 07:23 | |
| *** afazekas has quit IRC | 07:25 | |
| *** Jianyong has quit IRC | 07:27 | |
| *** Jianyong has joined #openstack-neutron | 07:30 | |
| *** ramishra_ has quit IRC | 07:34 | |
| *** markmcclain has joined #openstack-neutron | 07:37 | |
| *** jprovazn has joined #openstack-neutron | 07:39 | |
| *** afazekas has joined #openstack-neutron | 07:40 | |
| *** Longgeek has quit IRC | 07:42 | |
| *** Longgeek has joined #openstack-neutron | 07:43 | |
| *** spandhe has quit IRC | 07:46 | |
| *** Longgeek has quit IRC | 07:47 | |
| enikanorov_ | markmcclain: hi. I've filed another bug for the change we were discussing. also refined the commit message: https://review.openstack.org/#/c/81537/ could you please remove -2? | 07:50 |
| markmcclain | I'm about to board a plane. I'll take a look when I land | 07:50 |
| *** yamamoto_ has joined #openstack-neutron | 07:52 | |
| yamamoto_ | any core reviewers? | 07:54 |
| yamamoto_ | we have some changes in gerrit waiting for core reviewers. | 07:56 |
| *** markmcclain has quit IRC | 07:57 | |
| *** matsuhas_ has quit IRC | 07:59 | |
| *** WackoRobie has joined #openstack-neutron | 07:59 | |
| *** banix has joined #openstack-neutron | 08:01 | |
| *** WackoRobie has quit IRC | 08:03 | |
| *** matsuhashi has joined #openstack-neutron | 08:04 | |
| *** luqas has joined #openstack-neutron | 08:04 | |
| *** luqas has quit IRC | 08:06 | |
| *** bvandenh has joined #openstack-neutron | 08:07 | |
| *** Jianyong has quit IRC | 08:10 | |
| *** pnavarro has joined #openstack-neutron | 08:15 | |
| *** luqas has joined #openstack-neutron | 08:16 | |
| *** dfarrell07 has quit IRC | 08:18 | |
| *** Longgeek has joined #openstack-neutron | 08:20 | |
| *** tomoe_ has quit IRC | 08:21 | |
| *** Longgeek has quit IRC | 08:24 | |
| *** tomoe_ has joined #openstack-neutron | 08:29 | |
| *** jgallard has joined #openstack-neutron | 08:31 | |
| *** Matt2 has joined #openstack-neutron | 08:31 | |
| *** Jianyong has joined #openstack-neutron | 08:31 | |
| *** jgallard has quit IRC | 08:33 | |
| *** jgallard has joined #openstack-neutron | 08:33 | |
| *** emagana has quit IRC | 08:33 | |
| *** jistr has joined #openstack-neutron | 08:37 | |
| *** bjornar has joined #openstack-neutron | 08:38 | |
| *** banix_ has joined #openstack-neutron | 08:40 | |
| *** banix has quit IRC | 08:41 | |
| *** banix_ is now known as banix | 08:41 | |
| *** chandankumar_ has joined #openstack-neutron | 08:42 | |
| *** banix has quit IRC | 08:50 | |
| *** amotoki has quit IRC | 08:51 | |
| openstackgerrit | Jakub Libosvar proposed a change to openstack/neutron: Migrate data from cap_port_filter to vif_details https://review.openstack.org/83008 | 08:51 |
| *** banix has joined #openstack-neutron | 08:52 | |
| *** banix has quit IRC | 08:52 | |
| *** leseb has joined #openstack-neutron | 08:54 | |
| *** djoreilly has joined #openstack-neutron | 08:59 | |
| *** WackoRobie has joined #openstack-neutron | 08:59 | |
| *** jpich has joined #openstack-neutron | 09:03 | |
| *** WackoRobie has quit IRC | 09:04 | |
| *** ygbo has joined #openstack-neutron | 09:04 | |
| *** Longgeek has joined #openstack-neutron | 09:10 | |
| *** yfried1 has joined #openstack-neutron | 09:10 | |
| *** yfried has quit IRC | 09:12 | |
| *** Longgeek has quit IRC | 09:14 | |
| *** jianingy_afk has quit IRC | 09:14 | |
| *** tomoe_ has quit IRC | 09:22 | |
| *** Trozz has quit IRC | 09:28 | |
| *** Trozz has joined #openstack-neutron | 09:29 | |
| *** Trozz_ has joined #openstack-neutron | 09:30 | |
| *** tomoe_ has joined #openstack-neutron | 09:30 | |
| *** Trozz_ has quit IRC | 09:30 | |
| *** Trozz has quit IRC | 09:30 | |
| *** Trozz has joined #openstack-neutron | 09:31 | |
| openstackgerrit | Li Ma proposed a change to openstack/neutron: Race condition of L3-agent to add/remove routers https://review.openstack.org/73234 | 09:31 |
| *** Trozz_ has joined #openstack-neutron | 09:31 | |
| *** mich5225 has joined #openstack-neutron | 09:32 | |
| *** samuelbercovici has joined #openstack-neutron | 09:32 | |
| *** Trozz has quit IRC | 09:33 | |
| *** Trozz_ has quit IRC | 09:33 | |
| *** mich5225 has quit IRC | 09:33 | |
| *** Trozz has joined #openstack-neutron | 09:33 | |
| *** flwang has quit IRC | 09:35 | |
| *** samuelbercovici1 has joined #openstack-neutron | 09:36 | |
| *** sphoorti has joined #openstack-neutron | 09:36 | |
| *** tomoe_ has quit IRC | 09:36 | |
| *** Trozz has quit IRC | 09:39 | |
| *** samuelbercovici has quit IRC | 09:39 | |
| *** samuelbercovici1 is now known as samuelbercovici | 09:39 | |
| *** Trozz has joined #openstack-neutron | 09:40 | |
| *** Trozz has quit IRC | 09:40 | |
| *** Trozz has joined #openstack-neutron | 09:43 | |
| *** amuller has joined #openstack-neutron | 09:45 | |
| *** Trozz has quit IRC | 09:45 | |
| *** samuelbercovici1 has joined #openstack-neutron | 09:47 | |
| *** samuelbercovici has quit IRC | 09:50 | |
| *** samuelbercovici1 is now known as samuelbercovici | 09:50 | |
| openstackgerrit | Berezovsky Irena proposed a change to openstack/neutron: Add L2 Agent side handling for non consistent security_group settings https://review.openstack.org/82729 | 09:51 |
| *** yamamoto_ has quit IRC | 09:52 | |
| *** samuelbercovici1 has joined #openstack-neutron | 09:53 | |
| *** pradipta_away is now known as pradipta | 09:54 | |
| *** samuelbercovici has quit IRC | 09:57 | |
| *** samuelbercovici1 is now known as samuelbercovici | 09:57 | |
| *** samuelbercovici1 has joined #openstack-neutron | 09:59 | |
| *** WackoRobie has joined #openstack-neutron | 09:59 | |
| *** jp_at_hp has joined #openstack-neutron | 10:01 | |
| *** samuelbercovici has quit IRC | 10:02 | |
| *** samuelbercovici1 is now known as samuelbercovici | 10:02 | |
| *** WackoRobie has quit IRC | 10:03 | |
| *** samuelbercovici1 has joined #openstack-neutron | 10:06 | |
| *** yfried1 has quit IRC | 10:06 | |
| *** amuller has quit IRC | 10:07 | |
| *** samuelbercovici has quit IRC | 10:08 | |
| *** samuelbercovici1 is now known as samuelbercovici | 10:09 | |
| *** matsuhashi has quit IRC | 10:09 | |
| *** samuelbercovici1 has joined #openstack-neutron | 10:12 | |
| *** overlayer has joined #openstack-neutron | 10:12 | |
| *** samuelbercovici has quit IRC | 10:15 | |
| *** samuelbercovici1 is now known as samuelbercovici | 10:15 | |
| *** yamahata has quit IRC | 10:15 | |
| *** Longgeek has joined #openstack-neutron | 10:16 | |
| *** yfried has joined #openstack-neutron | 10:19 | |
| *** yfried has quit IRC | 10:20 | |
| *** yfried has joined #openstack-neutron | 10:20 | |
| *** chandankumar_ has quit IRC | 10:22 | |
| *** samuelbercovici1 has joined #openstack-neutron | 10:27 | |
| *** b3nt_pin has joined #openstack-neutron | 10:28 | |
| *** b3nt_pin is now known as beagles | 10:28 | |
| *** samuelbercovici has quit IRC | 10:29 | |
| *** samuelbercovici1 is now known as samuelbercovici | 10:29 | |
| *** Trozz has joined #openstack-neutron | 10:31 | |
| openstackgerrit | A change was merged to openstack/neutron: Imported Translations from Transifex https://review.openstack.org/82435 | 10:33 |
| *** Trozz has quit IRC | 10:34 | |
| *** samuelbercovici1 has joined #openstack-neutron | 10:36 | |
| *** matsuhashi has joined #openstack-neutron | 10:39 | |
| *** networkstatic is now known as networkstatic_zZ | 10:40 | |
| *** samuelbercovici has quit IRC | 10:40 | |
| *** samuelbercovici1 is now known as samuelbercovici | 10:40 | |
| *** Longgeek has quit IRC | 10:41 | |
| *** Trozz has joined #openstack-neutron | 10:41 | |
| *** Trozz has quit IRC | 10:42 | |
| *** Jabadia_ has joined #openstack-neutron | 10:43 | |
| *** Jabadia has quit IRC | 10:43 | |
| *** xuhanp has quit IRC | 10:44 | |
| *** pcm_ has joined #openstack-neutron | 10:44 | |
| *** Trozz has joined #openstack-neutron | 10:44 | |
| *** svilgelm has joined #openstack-neutron | 10:45 | |
| *** Trozz has quit IRC | 10:45 | |
| *** yfried is now known as yfried|lunch | 10:47 | |
| *** Jianyong has quit IRC | 10:49 | |
| *** samuelbercovici1 has joined #openstack-neutron | 10:51 | |
| *** samuelbercovici has quit IRC | 10:55 | |
| *** samuelbercovici1 is now known as samuelbercovici | 10:55 | |
| *** tomoe_ has joined #openstack-neutron | 10:58 | |
| *** WackoRobie has joined #openstack-neutron | 10:59 | |
| *** WackoRobie has quit IRC | 11:04 | |
| *** samuelbercovici1 has joined #openstack-neutron | 11:04 | |
| *** Longgeek has joined #openstack-neutron | 11:06 | |
| *** samuelbercovici has quit IRC | 11:07 | |
| *** samuelbercovici1 is now known as samuelbercovici | 11:07 | |
| *** amuller has joined #openstack-neutron | 11:08 | |
| *** irenab has quit IRC | 11:09 | |
| *** roeyc has joined #openstack-neutron | 11:09 | |
| *** irenab has joined #openstack-neutron | 11:10 | |
| *** sphoorti_ has joined #openstack-neutron | 11:10 | |
| *** jgallard has quit IRC | 11:10 | |
| openstackgerrit | Sphoorti proposed a change to openstack/neutron: Add unit test for add_vxlan in test_linux_ip_lib https://review.openstack.org/80554 | 11:11 |
| *** leseb has quit IRC | 11:14 | |
| *** sphoorti has quit IRC | 11:14 | |
| *** leseb has joined #openstack-neutron | 11:14 | |
| *** tvardeman has joined #openstack-neutron | 11:15 | |
| *** tomoe_ has quit IRC | 11:16 | |
| openstackgerrit | Ann Kamyshnikova proposed a change to openstack/neutron: Set correct nullable parameter for columns https://review.openstack.org/82089 | 11:17 |
| *** Longgeek has quit IRC | 11:17 | |
| *** leseb has quit IRC | 11:19 | |
| *** Longgeek has joined #openstack-neutron | 11:19 | |
| *** saju_m has joined #openstack-neutron | 11:20 | |
| *** rkukura has quit IRC | 11:21 | |
| *** Trozz has joined #openstack-neutron | 11:23 | |
| *** Trozz has quit IRC | 11:23 | |
| *** sphoorti_ has quit IRC | 11:25 | |
| *** samuelbercovici1 has joined #openstack-neutron | 11:26 | |
| *** sphoorti_ has joined #openstack-neutron | 11:26 | |
| *** mwagner_ is now known as mwagner_notHere | 11:29 | |
| *** Trozz has joined #openstack-neutron | 11:29 | |
| *** samuelbercovici has quit IRC | 11:29 | |
| *** samuelbercovici1 is now known as samuelbercovici | 11:29 | |
| *** samuelbercovici1 has joined #openstack-neutron | 11:32 | |
| *** sphoorti_ has quit IRC | 11:33 | |
| *** sphoorti_ has joined #openstack-neutron | 11:33 | |
| *** WackoRobie has joined #openstack-neutron | 11:35 | |
| *** WackoRobie has joined #openstack-neutron | 11:35 | |
| *** samuelbercovici has quit IRC | 11:35 | |
| *** samuelbercovici1 is now known as samuelbercovici | 11:35 | |
| *** Trozz has quit IRC | 11:36 | |
| *** yamahata has joined #openstack-neutron | 11:40 | |
| *** samuelbercovici1 has joined #openstack-neutron | 11:41 | |
| *** sphoorti_ has quit IRC | 11:42 | |
| *** sphoorti_ has joined #openstack-neutron | 11:42 | |
| *** samuelbercovici has quit IRC | 11:45 | |
| *** samuelbercovici1 is now known as samuelbercovici | 11:45 | |
| *** sphoorti_ has quit IRC | 11:47 | |
| *** sphoorti_ has joined #openstack-neutron | 11:48 | |
| openstackgerrit | Sylvain Afchain proposed a change to openstack/neutron: L2 pop regular mac learning as parameter https://review.openstack.org/83053 | 11:51 |
| openstackgerrit | Sergey Lukjanov proposed a change to openstack/neutron: Start using oslosphinx theme for docs https://review.openstack.org/83054 | 11:52 |
| *** samuelbercovici1 has joined #openstack-neutron | 11:54 | |
| *** leseb has joined #openstack-neutron | 11:56 | |
| *** samuelbercovici has quit IRC | 11:57 | |
| *** samuelbercovici1 is now known as samuelbercovici | 11:57 | |
| *** mwagner_lap has quit IRC | 12:00 | |
| *** leseb has quit IRC | 12:00 | |
| *** samuelbercovici1 has joined #openstack-neutron | 12:02 | |
| *** Trozz has joined #openstack-neutron | 12:02 | |
| *** Trozz has quit IRC | 12:03 | |
| *** WackoRobie has quit IRC | 12:03 | |
| *** sphoorti_ has quit IRC | 12:04 | |
| *** sphoorti_ has joined #openstack-neutron | 12:05 | |
| *** samuelbercovici has quit IRC | 12:05 | |
| *** samuelbercovici1 is now known as samuelbercovici | 12:06 | |
| *** yamahata has quit IRC | 12:06 | |
| *** samuelbercovici1 has joined #openstack-neutron | 12:08 | |
| *** flwang has joined #openstack-neutron | 12:08 | |
| *** matsuhashi has quit IRC | 12:10 | |
| *** samuelbercovici has quit IRC | 12:11 | |
| *** samuelbercovici1 is now known as samuelbercovici | 12:11 | |
| *** matsuhashi has joined #openstack-neutron | 12:11 | |
| *** samuelbercovici1 has joined #openstack-neutron | 12:13 | |
| *** matsuhas_ has joined #openstack-neutron | 12:13 | |
| *** matsuhashi has quit IRC | 12:13 | |
| *** xianghui has quit IRC | 12:14 | |
| *** bashok has joined #openstack-neutron | 12:15 | |
| *** sphoorti_ has quit IRC | 12:15 | |
| *** samuelbercovici has quit IRC | 12:16 | |
| *** samuelbercovici1 is now known as samuelbercovici | 12:16 | |
| *** sphoorti_ has joined #openstack-neutron | 12:16 | |
| *** dims_ has quit IRC | 12:18 | |
| *** dims_ has joined #openstack-neutron | 12:21 | |
| *** sphoorti_ has quit IRC | 12:21 | |
| *** sphoorti_ has joined #openstack-neutron | 12:23 | |
| *** samuelbercovici1 has joined #openstack-neutron | 12:25 | |
| *** leseb has joined #openstack-neutron | 12:27 | |
| sphoorti_ | roeyc: should I make that change and submit a new patch ? the blank line was unintentional. | 12:28 |
| *** samuelbercovici has quit IRC | 12:28 | |
| *** samuelbercovici1 is now known as samuelbercovici | 12:28 | |
| *** tomoe_ has joined #openstack-neutron | 12:29 | |
| *** luqas has quit IRC | 12:30 | |
| *** samuelbercovici1 has joined #openstack-neutron | 12:32 | |
| roeyc | sphoorti_: yes please | 12:33 |
| sphoorti_ | okay roeyc . | 12:34 |
| akamyshnikova | salv-orlando, Hello, I'm researching database and models synchronization and found out that in TzNetworkBinding model incorrect nullable is set for vlan_id and phy_uuid. It sets nullable=True, when they should be NOT NULL according to database content http://paste.openstack.org/show/74224/ and that they are primary keys. But when I corrected the model unit tests started to fail as in some of them vlan_id expected as None. https://github | 12:34 |
| akamyshnikova | .com/openstack/neutron/blob/master/neutron/tests/unit/vmware/test_nsx_plugin.py#L265 Is it mistake in unit tests? Please, answer me when you'll have time. | 12:34 |
| *** samuelbercovici has quit IRC | 12:35 | |
| *** samuelbercovici1 is now known as samuelbercovici | 12:35 | |
| salv-orlando | it's likely to be a migration issue. They should not be nullable, even if I think the code never allows for nullifying them. I will investigate and push a patch if needed. | 12:36 |
| salv-orlando | thanks for reporting this. | 12:36 |
| *** sphoorti_ has quit IRC | 12:39 | |
| *** sphoorti_ has joined #openstack-neutron | 12:40 | |
| openstackgerrit | Sphoorti proposed a change to openstack/neutron: Add unit test for add_vxlan in test_linux_ip_lib https://review.openstack.org/80554 | 12:42 |
| akamyshnikova | salv-orlando, I have already do change with this https://review.openstack.org/82089. May be you should take a look at it? | 12:43 |
| openstackgerrit | Ann Kamyshnikova proposed a change to openstack/neutron: Set correct nullable parameter for columns https://review.openstack.org/82089 | 12:43 |
| *** leseb has quit IRC | 12:43 | |
| *** xuhanp has joined #openstack-neutron | 12:47 | |
| *** jecarey has quit IRC | 12:48 | |
| *** sphoorti_ has quit IRC | 12:49 | |
| *** sphoorti_ has joined #openstack-neutron | 12:50 | |
| *** jaypipes has quit IRC | 12:51 | |
| *** samuelbercovici1 has joined #openstack-neutron | 12:51 | |
| openstackgerrit | Oleg Bondarev proposed a change to openstack/neutron: Delete disassociated floating ips on external network deletion https://review.openstack.org/53364 | 12:53 |
| *** samuelbercovici has quit IRC | 12:54 | |
| *** samuelbercovici1 is now known as samuelbercovici | 12:54 | |
| *** sridhar has quit IRC | 12:55 | |
| *** sphoorti_ has quit IRC | 12:56 | |
| *** sphoorti_ has joined #openstack-neutron | 12:56 | |
| *** jaypipes has joined #openstack-neutron | 12:56 | |
| *** thuc has joined #openstack-neutron | 12:57 | |
| *** thuc_ has joined #openstack-neutron | 12:58 | |
| *** leseb has joined #openstack-neutron | 13:00 | |
| *** WackoRobie has joined #openstack-neutron | 13:01 | |
| *** sphoorti__ has joined #openstack-neutron | 13:01 | |
| *** nplanel_ has joined #openstack-neutron | 13:02 | |
| *** thuc has quit IRC | 13:02 | |
| *** sphoorti_ has quit IRC | 13:02 | |
| *** matsuhas_ has quit IRC | 13:07 | |
| *** jgallard has joined #openstack-neutron | 13:07 | |
| *** baoli has joined #openstack-neutron | 13:09 | |
| *** pradipta is now known as pradipta_away | 13:12 | |
| *** mwagner_lap has joined #openstack-neutron | 13:12 | |
| *** Jianyong has joined #openstack-neutron | 13:16 | |
| *** julim has joined #openstack-neutron | 13:16 | |
| *** Jianyong has left #openstack-neutron | 13:16 | |
| *** yamahata has joined #openstack-neutron | 13:18 | |
| *** jobewan has joined #openstack-neutron | 13:19 | |
| openstackgerrit | Yuriy Taraday proposed a change to openstack/neutron: Abstract out root_helper calls to classes https://review.openstack.org/82787 | 13:21 |
| *** irenab has quit IRC | 13:23 | |
| *** jecarey has joined #openstack-neutron | 13:25 | |
| *** tomoe_ has quit IRC | 13:28 | |
| *** leseb has quit IRC | 13:31 | |
| *** leseb has joined #openstack-neutron | 13:34 | |
| *** luqas has joined #openstack-neutron | 13:35 | |
| *** afazekas_ has joined #openstack-neutron | 13:36 | |
| *** afazekas has quit IRC | 13:39 | |
| *** sridhar has joined #openstack-neutron | 13:39 | |
| *** spandhe has joined #openstack-neutron | 13:44 | |
| *** armax has joined #openstack-neutron | 13:46 | |
| *** spandhe_ has joined #openstack-neutron | 13:47 | |
| *** flwang has quit IRC | 13:49 | |
| *** spandhe has quit IRC | 13:49 | |
| *** spandhe_ is now known as spandhe | 13:49 | |
| *** flwang has joined #openstack-neutron | 13:50 | |
| *** kedarkul has quit IRC | 13:51 | |
| *** Jianyong has joined #openstack-neutron | 13:51 | |
| *** samuelbercovici1 has joined #openstack-neutron | 13:58 | |
| *** rkukura has joined #openstack-neutron | 13:59 | |
| *** skraynev is now known as skraynev_going_h | 13:59 | |
| *** skraynev_going_h is now known as skraynev_afk | 13:59 | |
| *** dguitarbite has quit IRC | 14:00 | |
| *** thuc_ has quit IRC | 14:00 | |
| *** thuc has joined #openstack-neutron | 14:01 | |
| *** samuelbercovici has quit IRC | 14:01 | |
| *** samuelbercovici1 is now known as samuelbercovici | 14:01 | |
| *** beagles is now known as beagles_food | 14:03 | |
| *** baoli has quit IRC | 14:03 | |
| *** spandhe has quit IRC | 14:04 | |
| *** thuc has quit IRC | 14:05 | |
| *** thuc has joined #openstack-neutron | 14:05 | |
| *** thuc has quit IRC | 14:06 | |
| *** yfried|lunch has quit IRC | 14:06 | |
| *** baoli has joined #openstack-neutron | 14:06 | |
| *** thuc has joined #openstack-neutron | 14:07 | |
| *** spandhe has joined #openstack-neutron | 14:07 | |
| *** samuelbercovici has quit IRC | 14:10 | |
| *** jgrimm has joined #openstack-neutron | 14:11 | |
| *** svilgelm has quit IRC | 14:12 | |
| *** WackoRobie has quit IRC | 14:13 | |
| *** otherwiseguy has joined #openstack-neutron | 14:15 | |
| *** amotoki has joined #openstack-neutron | 14:18 | |
| *** thuc has quit IRC | 14:23 | |
| *** thuc has joined #openstack-neutron | 14:23 | |
| *** Longgeek has quit IRC | 14:23 | |
| *** Longgeek has joined #openstack-neutron | 14:24 | |
| *** rkukura has quit IRC | 14:25 | |
| overlayer | folks, say I wanted to merge one of my existing physical LANs (connected to a physical router reachable from my openstack deployment) with neutron... is it currently possible or is there any work ongoing? | 14:25 |
| *** Longgeek has quit IRC | 14:25 | |
| *** otherwiseguy has quit IRC | 14:26 | |
| *** Longgeek has joined #openstack-neutron | 14:27 | |
| *** thuc has quit IRC | 14:28 | |
| *** dguitarbite has joined #openstack-neutron | 14:30 | |
| openstackgerrit | Miguel Angel Ajo proposed a change to openstack/neutron: fixes broken neutron-netns-cleanup https://review.openstack.org/80261 | 14:30 |
| *** tomoe_ has joined #openstack-neutron | 14:31 | |
| openstackgerrit | Akihiro Motoki proposed a change to openstack/neutron: Import request_id middleware from oslo https://review.openstack.org/83080 | 14:34 |
| openstackgerrit | Oleg Bondarev proposed a change to openstack/neutron: Reschedule router if new external gateway is on other network https://review.openstack.org/52884 | 14:34 |
| openstackgerrit | Yves-Gwenael Bourhis proposed a change to openstack/neutron: Make dnsmasq aware of all names https://review.openstack.org/52930 | 14:34 |
| openstackgerrit | Akihiro Motoki proposed a change to openstack/neutron: Import request_id middleware bug fix from oslo https://review.openstack.org/83080 | 14:35 |
| *** alexpilotti has joined #openstack-neutron | 14:39 | |
| *** Jianyong has quit IRC | 14:42 | |
| *** chandan_kumar has quit IRC | 14:42 | |
| *** chandankumar_ has joined #openstack-neutron | 14:44 | |
| *** skraynev_afk is now known as skraynev | 14:44 | |
| *** sphoorti__ has quit IRC | 14:45 | |
| *** chandankumar_ has quit IRC | 14:46 | |
| *** mestery has joined #openstack-neutron | 14:46 | |
| *** chandan_kumar has joined #openstack-neutron | 14:47 | |
| *** kmak_ has joined #openstack-neutron | 14:47 | |
| *** chandan_kumar has quit IRC | 14:48 | |
| *** chandan_kumar has joined #openstack-neutron | 14:48 | |
| *** thedodd has joined #openstack-neutron | 14:48 | |
| *** sphoorti has joined #openstack-neutron | 14:51 | |
| *** emagana has joined #openstack-neutron | 14:55 | |
| *** carl_baldwin has joined #openstack-neutron | 14:55 | |
| *** chandan_kumar has quit IRC | 14:55 | |
| *** chandan_kumar has joined #openstack-neutron | 14:56 | |
| *** dave_tucker_zzz is now known as dave_tucker | 14:57 | |
| *** sweston has quit IRC | 14:58 | |
| *** banix has joined #openstack-neutron | 14:59 | |
| *** emagana has quit IRC | 15:00 | |
| *** julim_ has joined #openstack-neutron | 15:00 | |
| *** beagles_food is now known as beagles | 15:01 | |
| *** julim has quit IRC | 15:01 | |
| *** networkstatic_zZ has quit IRC | 15:01 | |
| *** yfried has joined #openstack-neutron | 15:01 | |
| *** bvandenh has quit IRC | 15:02 | |
| *** chandan_kumar has quit IRC | 15:03 | |
| *** yamahata has quit IRC | 15:03 | |
| *** thuc has joined #openstack-neutron | 15:04 | |
| *** thuc_ has joined #openstack-neutron | 15:05 | |
| flwang | safchain: ping | 15:05 |
| *** mestery has quit IRC | 15:05 | |
| *** yamahata has joined #openstack-neutron | 15:05 | |
| *** otherwiseguy has joined #openstack-neutron | 15:08 | |
| *** thuc has quit IRC | 15:09 | |
| *** yamahata has quit IRC | 15:09 | |
| *** dguitarbite has quit IRC | 15:09 | |
| *** chandan_kumar has joined #openstack-neutron | 15:09 | |
| *** yamahata has joined #openstack-neutron | 15:10 | |
| *** thuc_ has quit IRC | 15:12 | |
| *** thuc has joined #openstack-neutron | 15:13 | |
| *** mflobo has joined #openstack-neutron | 15:15 | |
| openstackgerrit | Ihar Hrachyshka proposed a change to openstack/neutron: Synced rpc and gettextutils modules from oslo-incubator https://review.openstack.org/80998 | 15:16 |
| *** sweston has joined #openstack-neutron | 15:16 | |
| amotoki | HenryG: thanks for catching quotas db migration issue on folsom_initial. I am now investigating the situation. I remember I filed a bug about it. | 15:16 |
| *** yfried has quit IRC | 15:16 | |
| HenryG | amotoki: cool, let me know what you find | 15:17 |
| *** Longgeek has quit IRC | 15:17 | |
| amotoki | HenryG: https://bugs.launchpad.net/neutron/+bug/1277379 but my initial investigation has incorrect points. I am updating it now. | 15:18 |
| HenryG | amotoki: FYI cisco plugin is also affected | 15:20 |
| amotoki | HenryG: i see. is it because cisco plugin loads ovs plugin internally? | 15:20 |
| HenryG | amotoki: yup | 15:20 |
| *** kmak_ has quit IRC | 15:21 | |
| amotoki | HenryG: if my investigation is correct, the thing is simpler than i thought first :) | 15:21 |
| HenryG | amotoki: that would be nice | 15:23 |
| *** Longgeek has joined #openstack-neutron | 15:24 | |
| *** Longgeek has quit IRC | 15:24 | |
| openstackgerrit | Ihar Hrachyshka proposed a change to openstack/neutron: Synced rpc and gettextutils modules from oslo-incubator https://review.openstack.org/80998 | 15:24 |
| *** rkukura has joined #openstack-neutron | 15:26 | |
| amotoki | HenryG: i post a comment to bug 1277379 above. Could you check my comment? | 15:29 |
| * HenryG looks ... | 15:29 | |
| *** networkstatic has joined #openstack-neutron | 15:30 | |
| *** Longgeek has joined #openstack-neutron | 15:32 | |
| openstackgerrit | Akihiro Motoki proposed a change to openstack/neutron: nec plugin: allow to delete resource with ERROR status https://review.openstack.org/82143 | 15:33 |
| HenryG | amotoki: looks good to me, but I am no expert on this. We should get markmcclain and others ( salv-orlando maybe?) to weigh in. | 15:33 |
| *** Longgeek has quit IRC | 15:34 | |
| HenryG | amotoki: meanwhile I will apply the proposed changes and update the review, adding your bug as also fixed. OK? | 15:34 |
| *** Longgeek has joined #openstack-neutron | 15:34 | |
| amotoki | HenryG: sounds good. | 15:34 |
| salv-orlando | amotoki, HenryG: definetely fine. You need to cleanup stale resources. | 15:36 |
| amotoki | salv-orlando: thanks. I think it is better to deal with qutoas table bug in the same patch. | 15:36 |
| HenryG | salv-orlando: sorry, I don't get the "cleanup stale resources" part? Can you elaborate? | 15:37 |
| *** emagana has joined #openstack-neutron | 15:38 | |
| salv-orlando | amotoki: can we do a quick list of plugins needing hybrid plugging? ML2 with OVS, OVS, RYU, and then? | 15:38 |
| salv-orlando | does also NEC require hybrid? | 15:38 |
| amotoki | salv-orlando: yes. nec plugin uses it. | 15:38 |
| *** sacharya has joined #openstack-neutron | 15:38 | |
| HenryG | salv-orlando: since cisco plugin uses OVS, it also need hybrid | 15:39 |
| salv-orlando | HenryG: which one? I see only n1kv is setting port bindings (from what I gather) | 15:39 |
| amotoki | salv-orlando: cisco network plugins loads ovs plugin internally. | 15:40 |
| amotoki | bigswitch seems affected. it has two types of secgroup implementation: controller and agent-based. | 15:40 |
| salv-orlando | amotoki: right. So fixing OVS plugi fixes cisco as well. I don't know however if n1kv which is used as a switch also needs hybrid plugging | 15:41 |
| *** mflobo has quit IRC | 15:41 | |
| HenryG | salv-orlando: amotoki: I will have to check with the n1kv team how they do plugging | 15:42 |
| salv-orlando | sure np | 15:42 |
| HenryG | salv-orlando: amotoki is right for cisco with (nexus + ovs) sub-plugins uses OVS agent to plug | 15:42 |
| *** saju_m has quit IRC | 15:43 | |
| amotoki | salv-orlando: ML2 meeting will start in 20 minutes. Is it better to check which MDs are affected? | 15:44 |
| *** spandhe_ has joined #openstack-neutron | 15:44 | |
| *** spandhe has quit IRC | 15:44 | |
| *** spandhe_ is now known as spandhe | 15:44 | |
| salv-orlando | not a great idea to crash the ML2 agenda. Maybe that can be mentioned mention that in Open discussion | 15:47 |
| *** _cjones_ has joined #openstack-neutron | 15:48 | |
| *** spandhe has quit IRC | 15:51 | |
| amotoki | okay. vif_security bug will be covered in the agenda. if we have time i will check it. | 15:53 |
| *** Sudhakar has joined #openstack-neutron | 15:54 | |
| *** dfarrell07 has joined #openstack-neutron | 15:54 | |
| jlibosva | rkukura: salv-orlando: obondarev hi, may I ask you please to check https://review.openstack.org/#/c/83008/ ? - it is probably the last piece of puzzle for grenade | 15:56 |
| salv-orlando | yes sure | 15:56 |
| jlibosva | thanks! | 15:56 |
| rkukura | salv-orlando, amotoki: ML2 agenda is light, VIF security is last item, so can discuss their if you’d like | 15:58 |
| Sudhakar | salv-orlando,carl_baldwin : You had comments/concerns with patch https://review.openstack.org/#/c/77549/ ...if you have time .. we can discuss a little .. | 16:00 |
| *** rcurran has joined #openstack-neutron | 16:00 | |
| *** mestery has joined #openstack-neutron | 16:01 | |
| *** xuhanp has quit IRC | 16:01 | |
| carl_baldwin | Sudhakar: My reason for the -1 was mostly around the timing of the patch with respect to Icehouse RC. If there were some benchmarking or testing that showed that this patch is very valuable then that might justify it. | 16:02 |
| *** rudrarugge has joined #openstack-neutron | 16:02 | |
| *** bashok has quit IRC | 16:02 | |
| Sudhakar | we did scale testing on this fix and it helped us to scale more VMs on CNs | 16:03 |
| carl_baldwin | Do you have some data that you could put in to the patch? | 16:04 |
| Sudhakar | should i put along the test results from our testing?? | 16:04 |
| carl_baldwin | My other comments were around potentially simplifying the patch to achieve the same result. Might reduce risk but I did not give -1 for those comments because I realize that is more being critical of style rather than finding a real problem. | 16:04 |
| *** yamamoto has joined #openstack-neutron | 16:05 | |
| carl_baldwin | Sudhakar: I think it could help justify the inclusion of the patch sooner than later but I can't speak for the core devs. | 16:05 |
| *** mlavalle has joined #openstack-neutron | 16:06 | |
| *** mlavalle has quit IRC | 16:06 | |
| Sudhakar | I responded to your comments reg. the alternate approach as part of the review comment replies... | 16:06 |
| Sudhakar | carl_baldwin: Yes, I can understand... | 16:06 |
| carl_baldwin | Sudhakar: Okay, I have not had time yet to revisit the patch. I will today. | 16:06 |
| Sudhakar | carl_baldwin: Thanks :) | 16:06 |
| carl_baldwin | Glad to help. | 16:06 |
| Sudhakar | carl_baldwin: I will attach the test results to give more confidence... | 16:07 |
| carl_baldwin | cool | 16:07 |
| carl_baldwin | Sudhakar: The data might fit better in the bug report. | 16:09 |
| carl_baldwin | ... or a comment in the review. I don't know where the best place to include it is, actually. | 16:09 |
| Sudhakar | carl_baldwin: Yes, I thought the same too...will post a comment in the bug report... | 16:09 |
| carl_baldwin | Great. | 16:09 |
| *** Sukhdev has joined #openstack-neutron | 16:10 | |
| carl_baldwin | Sudhakar: btw, I don't see your responses in the review. Have you submitted your draft comments yet? | 16:10 |
| *** mlavalle has joined #openstack-neutron | 16:11 | |
| Sudhakar | carl_baldwin: err...let me check ... | 16:11 |
| Sudhakar | carl_baldwin: i saved them...and thought they are published.. | 16:11 |
| *** amuller has quit IRC | 16:13 | |
| Sudhakar | carl_baldwin: sorry...they are still marked as drafts... | 16:13 |
| Sudhakar | carl_baldwin: how do i submit ? | 16:13 |
| carl_baldwin | In the main review page under patch set 5, find the "Review" button and fill out the form and submit. | 16:14 |
| Sudhakar | carl_baldwin:ok..let me try.. | 16:14 |
| *** skraynev is now known as skraynev_afk | 16:14 | |
| *** Dseven has joined #openstack-neutron | 16:15 | |
| Sudhakar | carl_baldwin:done... thanks for helping out :) | 16:15 |
| carl_baldwin | Great, glad to help. I'll have a look in a bit. | 16:16 |
| *** pnavarro has quit IRC | 16:17 | |
| Sudhakar | carl_baldwin:sure.. | 16:17 |
| *** irenab has joined #openstack-neutron | 16:20 | |
| salv-orlando | Sudhakar: no concern on your patch. I think the code works (I don't think we can use it to say "closes bug" however given the very nature of the bug) | 16:22 |
| salv-orlando | but we need to get RC1 out of the way and I don't want to push patches which change logic and therefore might trigger regressions, unless they're fixing somehting which is now fundamentally broken | 16:22 |
| *** leseb has quit IRC | 16:22 | |
| salv-orlando | let | 16:22 |
| salv-orlando | let's give a few more spins, and then we'll see whether cut it for RC2, or then do early Juno and then backport | 16:23 |
| Sudhakar | carl_baldwin:agreed.. | 16:23 |
| *** Longgeek has quit IRC | 16:23 | |
| *** Longgeek has joined #openstack-neutron | 16:23 | |
| Sudhakar | hi salv-orlando | 16:24 |
| *** SumitNaiksatam has joined #openstack-neutron | 16:24 | |
| *** dfarrell07 has quit IRC | 16:28 | |
| openstackgerrit | Yuriy Taraday proposed a change to openstack/neutron: Abstract out root_helper calls to classes https://review.openstack.org/82787 | 16:29 |
| _cjones_ | kevinbenton. Thanks for the help over the past few days. I did a clean install and have everything running to my satisfaction. I've added my own test cases and I have a few questions regarding how these should be set up. If you wouldn't mind. | 16:30 |
| *** Longgeek has quit IRC | 16:30 | |
| *** afazekas_ has quit IRC | 16:30 | |
| *** Longgeek has joined #openstack-neutron | 16:31 | |
| *** Longgeek has quit IRC | 16:31 | |
| *** leseb has joined #openstack-neutron | 16:32 | |
| *** Longgeek has joined #openstack-neutron | 16:32 | |
| *** devlaps has joined #openstack-neutron | 16:36 | |
| *** tomoe_ has quit IRC | 16:42 | |
| openstackgerrit | Henry Gessau proposed a change to openstack/neutron: Cisco APIC ML2 mechanism driver, part 1 https://review.openstack.org/73355 | 16:43 |
| *** mestery has quit IRC | 16:49 | |
| *** sacharya has quit IRC | 16:50 | |
| *** Longgeek has quit IRC | 16:54 | |
| *** pnavarro has joined #openstack-neutron | 16:54 | |
| *** chandan_kumar has quit IRC | 16:57 | |
| *** harlowja_away is now known as harlowja | 16:58 | |
| *** jlibosva has quit IRC | 17:00 | |
| *** nati_ueno has joined #openstack-neutron | 17:00 | |
| *** dave_tucker is now known as dave_tucker_zzz | 17:00 | |
| nati_ueno | hi | 17:01 |
| rkukura | nati_ueno: hi | 17:02 |
| kevinbenton | hey | 17:02 |
| nati_ueno | rkukura: hi | 17:02 |
| nati_ueno | kevinbenton: hey | 17:02 |
| kevinbenton | _cjones_: sure | 17:02 |
| *** gpizza has joined #openstack-neutron | 17:03 | |
| marun | nati_ueno: hi | 17:04 |
| *** baoli has quit IRC | 17:04 | |
| nati_ueno | marun: hi | 17:04 |
| nati_ueno | russellb: around? | 17:04 |
| Sukhdev | rcurran: Ping | 17:05 |
| nati_ueno | salv-orlando: around? | 17:05 |
| russellb | i am, just now reading through your mail though | 17:05 |
| nati_ueno | russellb: Thanks we will wait your question in here | 17:06 |
| russellb | nati_ueno: you see danpb's comment on the patch? | 17:07 |
| nati_ueno | russellb: let me check | 17:08 |
| *** Sudhakar has quit IRC | 17:08 | |
| rcurran | Sukhdev, yes - | 17:08 |
| russellb | nati_ueno: or that won't work because it's not for all neutron cases, just for one driver? | 17:08 |
| nati_ueno | russellb: I think Salvatore's patch also works https://launchpadlibrarian.net/170772780/hack.patch | 17:09 |
| *** mestery has joined #openstack-neutron | 17:09 | |
| nati_ueno | russellb: please take a looks his comment on #6 https://bugs.launchpad.net/neutron/+bug/1297469 | 17:09 |
| rkukura | Are we now considering any patch that passes a minimal set of flags via binding:vif-details to the nova VIF object and to the GenericVIFDriver, or has that been absolutely ruled out for icehouse? | 17:09 |
| nati_ueno | rkukura: I don't think we can take vif-details way now | 17:10 |
| *** leseb has quit IRC | 17:10 | |
| *** salv-mobile has joined #openstack-neutron | 17:10 | |
| *** jistr has quit IRC | 17:10 | |
| nati_ueno | so I'm +1 for Salvatore's patch (this is equivalent with Daniel's suggestion) | 17:11 |
| russellb | nati_ueno: yes, if that works, that looks a lot better | 17:11 |
| russellb | no config changes | 17:11 |
| rkukura | My view is that if we do some simple nova-only short term hack, fine, but if we are trying to do anything we’ll live with long term, the vif-details approach is the way to go - I have not seen any objections to it | 17:11 |
| russellb | rkukura: yeah it's just a timing thing | 17:11 |
| kevinbenton | +1 to salv-orlando's | 17:11 |
| nati_ueno | russellb: yes. OK If you like this way, I'll fix this patch asap. (I'll push the patch in short, but give me 1,2 hours including test) | 17:11 |
| rkukura | nati_ueno: Can you pose he link to Salvatore’s patch? | 17:12 |
| russellb | nati_ueno: sounds good | 17:12 |
| nati_ueno | russellb: Thanks | 17:12 |
| kevinbenton | https://launchpadlibrarian.net/170772780/hack.patch | 17:12 |
| kevinbenton | rkukura: ^^ | 17:12 |
| salv-mobile | So I've missed the rest of the discussion. What are we ok with? | 17:12 |
| rkukura | kevinbenton: Thanks - noticed that right after asking | 17:12 |
| kevinbenton | salv-mobile: the hack.patch | 17:12 |
| salv-mobile | What about the other thing I tried? | 17:12 |
| salv-mobile | The non hack one | 17:12 |
| kevinbenton | salv-mobile: which one is that? | 17:13 |
| *** gpizza has quit IRC | 17:13 | |
| salv-mobile | It's still on the bug report | 17:13 |
| *** Jabadia_ has quit IRC | 17:13 | |
| salv-mobile | Too not still | 17:13 |
| _cjones_ | kevinbenton: Awesome. Give me a minute to compose my questions. | 17:13 |
| *** yfried has joined #openstack-neutron | 17:13 | |
| *** Jabadia has joined #openstack-neutron | 17:13 | |
| *** ygbo has quit IRC | 17:13 | |
| russellb | i'd rather not make API changes under this much time pressure | 17:13 |
| salv-mobile | Basically i just let nova use vif type to decide but add a new hybrid vif type | 17:13 |
| russellb | would rather have time to have it more completely thought out | 17:14 |
| salv-mobile | Russellb there no api change | 17:14 |
| salv-mobile | russellb: nova already leverage that attribute. The change is simply to add one more value to the list of possible vif types | 17:15 |
| *** dave_tucker_zzz is now known as dave_tucker | 17:15 | |
| rkukura | Long term, binding:vif_type plus binding:vif_details should be all that’s needed for neutron plugin/driver to control what nova’s GenericVIFDriver does. | 17:16 |
| *** rcurran has quit IRC | 17:16 | |
| *** baoli has joined #openstack-neutron | 17:16 | |
| *** itzikb has joined #openstack-neutron | 17:16 | |
| kevinbenton | salv-mobile: i don't like the extra VIF type as much. nova gets by with security groups using one vif type, but then neutron needs another one... | 17:16 |
| salv-mobile | Meh look at how many vif types are already there | 17:16 |
| *** sweston has quit IRC | 17:17 | |
| marun | So are we going with salv-mobile's workaround? | 17:18 |
| *** Jabadia has quit IRC | 17:18 | |
| rkukura | We already have support for vif-details on the neutron side for months - why is having nova pass this to the GenericVIFDriver any more work/risk than defining new vif_type values tjhat nova treats specially? | 17:18 |
| marun | It looks reasonable to me. | 17:18 |
| russellb | marun: the first one, yes | 17:18 |
| marun | great | 17:18 |
| marun | rkukura: because what do we do on reboot? | 17:18 |
| russellb | salv-mobile: but it's a hack we'd have to maintain support for, can't just remove it in a few weeks | 17:18 |
| russellb | salv-mobile: at least the first hack is invisible and we can remove it as soon as we're ready to | 17:18 |
| rkukura | marun: Explain? | 17:18 |
| marun | rkukura: doesn't nova replug on reboot? | 17:19 |
| marun | rkukura: and if so, how does it know what to do? | 17:19 |
| marun | rkukura: does it store those vif details/ | 17:19 |
| salv-mobile | I don't think we'll ever get to an agreement on any patch larger than 10 lines | 17:19 |
| marun | rkukura: or does it have to query neutron again? | 17:19 |
| rkukura | If nova replugs, it should be getting the neutron port, but if its not doing this, I see the issue | 17:19 |
| marun | rkukura: I'm honestly asking, it's a gap in my knowledge | 17:19 |
| salv-mobile | So maybe just go with the hack which at least does not require config changes | 17:19 |
| marun | russellb: ^^ thoughts on reboot? | 17:19 |
| marun | russellb: (same question on migration/resize/etc) | 17:20 |
| marun | +1 to the hack. | 17:20 |
| salv-mobile | In this way any fix you work out it might be easy to backport | 17:20 |
| marun | It's simple and doesn't violate the 'there should not be new vif drivers' mandate | 17:20 |
| salv-mobile | Might just be worth checking with a few plugin devs to see if forcing hybrid breaks their plugins | 17:21 |
| marun | rkukura: not anything we can't figure out, just not something that's simple enough to consider doing so late in cycle. | 17:21 |
| salv-mobile | Or on the other hand merge as it is and then wait for them screaming | 17:21 |
| rkukura | I’m completely OK with something like salv-orlando’s hack if its acknowledged as a short term solution and that we’ll work out whether the vif-details approach is the long-term solution for VIF security, as well as for things like SR-IOV that need this same information flow to the GenericVIFDriver. | 17:21 |
| marun | rkukura: +1 | 17:22 |
| marun | rkukura: now that we have more people engaged in this discussion hopefully we can push for it early enough in juno to iron out the kinks | 17:22 |
| openstackgerrit | Kyle Mestery proposed a change to openstack/neutron: Correct OVS VXLAN version check https://review.openstack.org/82931 | 17:22 |
| marun | rkukura: (engaged on both the nova and neutron sides) | 17:22 |
| russellb | marun: don't know, i'd have to look | 17:22 |
| russellb | and yes, totally short term | 17:23 |
| kevinbenton | salv-mobile: unless they were depending on security groups not functioning, i think it should be okay :) | 17:23 |
| marun | russellb: no worries, just curious. we'll have to get the details right for the proper fix. | 17:23 |
| russellb | avoiding the rest until it's not under pressure to avoid doing something without enough thinking on the proper fix | 17:23 |
| *** jmeridth has joined #openstack-neutron | 17:23 | |
| marun | russellb: are you ok with the salv-mobile's workaround then? | 17:23 |
| marun | frankly, I'd like to see this merge today if we're all in agreement so we focus on everything else. | 17:24 |
| HenryG | amotoki: ping re: quotas in folsom_initial | 17:24 |
| Dseven | is the proposed workaround to force hybrid plugging always, even when iptables filtering is not required? | 17:24 |
| marun | Dseven: no | 17:24 |
| marun | it's to use the facts that neutron is in use + what firewall driver is configured to decide whether to use hybrid plugging | 17:25 |
| marun | i.e. we have the ability to support what we need (selective use of hybrid plugging), just not in as straightforward a way configuration-wise | 17:25 |
| kevinbenton | the hack.patch does always plug hybrid though if neutron is in use, right? | 17:26 |
| marun | salv-mobile: or, hmm. do I have it wrong. | 17:26 |
| marun | kevinbenton: I was just realizing that. | 17:26 |
| marun | salv-mobile: doesn't that break vmware? | 17:26 |
| marun | kevinbenton: no, i'm wrong. | 17:26 |
| djoreilly | so what about the linuxbridge plugin? | 17:26 |
| salv-mobile | See? | 17:27 |
| rkukura | This only applies when vif_type = ovs, right? | 17:27 |
| marun | kevinbenton: if firewall is 'noop' OR neutron is used, hybrid plugging | 17:27 |
| salv-mobile | That's why I asked to query other plugin types | 17:27 |
| _cjones_ | kevinbenton: I have built a driver that makes use of mongodb. When I run my tests individually they all seem to pass, but when I run them as a suite, it looks like things are not timed correctly and I get failures. Is this likely a programming issue or a timing issue? | 17:27 |
| salv-mobile | Yes only ovs vif type | 17:27 |
| russellb | marun: yes, i am | 17:27 |
| marun | kevinbenton: so if neutron wants hybrid plugging -> configure noop firewall | 17:27 |
| marun | kevinbenton: if neutron doesn't want hybrid plugging -> configure valid firewall | 17:27 |
| marun | russellb: awesome. | 17:28 |
| *** itzikb has quit IRC | 17:28 | |
| kevinbenton | but the OR condition will always be met | 17:28 |
| kevinbenton | or self.is_neutron() | 17:28 |
| marun | salv-mobile: do you want to post that patch to a review so we can get nova eyes and get it merged? | 17:28 |
| *** itzikb has joined #openstack-neutron | 17:28 | |
| Dseven | marun: that's almost opposite of the current code | 17:28 |
| Dseven | which uses hybrid if anything other than the noop firewall driver is configured | 17:28 |
| marun | kevinbenton: so the logic is messed in the patch. you see how we can get it working though? | 17:28 |
| salv-mobile | I'm not at home now. But I think nati-ueni is doing that | 17:29 |
| marun | Dseven: a fact we'll have to live with | 17:29 |
| marun | Dseven: like I said, it's configuration chicanery to get the result we want because we've left things too late for a proper fix | 17:29 |
| banix | what if you really want the noop firewall (admiting i do not know about the context here but our plugin uses noop firewall; will this change affect us?) | 17:30 |
| Dseven | I'm hoping to end up with a non-iptables firewall solution, and I'd rather not have the overhead of the ethernet bridge and veth pair that I don't need | 17:30 |
| kevinbenton | marun: Dseven, banix, how do you leverage security groups? | 17:30 |
| _cjones_ | kevinbenton: Second part. My driver has a config portion: config = conf.CONF.ml2_mydriver. How do I get this installed/setup for the the test cases to be able to mock this class? | 17:30 |
| marun | nati_ueno: so you're posting a version of salvatore's patch? | 17:31 |
| banix | kevinbenton: in our case we do not support security groups | 17:31 |
| Dseven | I'm hoping to get a firewall solution that runs in a VM, and all traffic is sent through it via OVS | 17:31 |
| nati_ueno | marun: yep | 17:31 |
| marun | nati_ueno: awesome | 17:31 |
| marun | nati_ueno: thank you for your hard work to make this happen! | 17:31 |
| nati_ueno | marun: no This is my fault | 17:32 |
| marun | nati_ueno: we're all culpable | 17:32 |
| marun | nati_ueno: no one person is to blame. anybody could have noticed things were slipping and raised the issue at any point this cycle | 17:33 |
| marun | nati_ueno: the fact that it didn't happen till now is as much the rest of core's fault as it is yours. | 17:33 |
| marun | (we're all in this boat together, we need all hands to bail) | 17:33 |
| kevinbenton | _cjones_: for your second question, you need something in your test setup that calls the config setup | 17:34 |
| marun | Dseven: what are you working on and is it going to target icehouse? | 17:34 |
| *** itzikb has quit IRC | 17:34 | |
| *** itzikb has joined #openstack-neutron | 17:35 | |
| Dseven | working on a private cloud implementation (currently in development) and yes, targetting icehouse | 17:35 |
| marun | salv-mobile: can the suggested workaround support non-hybrid plugging for neutron? | 17:35 |
| salv-mobile | Nope otherwise I would not have called the attachment hack.patch | 17:36 |
| marun | salv-mobile: is there a way it could? | 17:36 |
| kevinbenton | _cjones_: https://github.com/bigswitch/neutron/blob/master/neutron/tests/unit/ml2/drivers/test_bigswitch_mech.py#L48 | 17:36 |
| marun | salv-mobile: i.e. neutron configures a firewall driver - any firewall driver - and hybrid plugging is used | 17:36 |
| *** nplanel_ has quit IRC | 17:36 | |
| salv-mobile | Not in a million years. I clarified that on the bug repo | 17:36 |
| marun | salv-mobile: oops, the reverse | 17:36 |
| marun | salv-mobile: what repo? | 17:37 |
| marun | salv-mobile: all I have is the patch | 17:37 |
| salv-mobile | Report not repo | 17:37 |
| marun | salv-mobile: so why 'not in a million years'? | 17:37 |
| marun | salv-mobile: nmind, i'll go read bug report | 17:37 |
| *** sphoorti has quit IRC | 17:38 | |
| *** baoli has quit IRC | 17:38 | |
| _cjones_ | kevinbenton: Perfect. Got that. I think that should work. | 17:38 |
| kevinbenton | _cjones_: for your first question, what kind of failures are you getting? are they DB conflicts or something? you might need to add something to the UT cleanup to clear out your backend stuff after each test | 17:38 |
| Sukhdev | rkukura: Ping | 17:41 |
| rkukura | Sukhdev: hi | 17:41 |
| *** sweston has joined #openstack-neutron | 17:41 | |
| marun | Dseven: It doesn't sound like we have any option but to add the intermediary bridge. | 17:41 |
| Sukhdev | rkukura: Wanted to chat with you regarding the idea of sync in ML2 | 17:41 |
| marun | Dseven: at least for icehouse. it may be we can backport something better early in juno | 17:41 |
| _cjones_ | kevinbenton: No, they appear as timing errors. https://gist.github.com/cjones-/eef389050e6145a30ca9 | 17:42 |
| rkukura | Sukhdev: sure | 17:42 |
| Sukhdev | rkukura: If you think it is worthwhile to have a session on this topic - I would love to work with you | 17:42 |
| Dseven | marun: OK, I think I can live with it for a while ... or maybe I'll hack around it | 17:42 |
| _cjones_ | kevinbenton: This test just tries to create 100 tenants and save them to mongodb. | 17:43 |
| Sukhdev | rkukura: perhaps we can jointly work on it - wanted to hear your thoughts | 17:43 |
| rkukura | Sukhdev: OK, you are welcome to take the lead on this, and I’ll assist | 17:43 |
| marun | Dseven: if you're willing to maintain support for it, non-hybrid plugging is certainly doable | 17:43 |
| *** baoli has joined #openstack-neutron | 17:43 | |
| Dseven | marun: I actually have a side-issue where non-hybrid plugging doesn't seem to work with libvirt+Xen | 17:43 |
| rkukura | Sukhdev: Is the idea to draft a BP and discuss it at the summit? | 17:44 |
| marun | Dseven: ah, xen :) | 17:44 |
| marun | Dseven: you have my condolenses | 17:44 |
| Dseven | marun: yeah, I like to make my life *really* difficult | 17:44 |
| kevinbenton | marun: what about adding another "neutronfirewalldriver" to nova that extends the noop class | 17:44 |
| marun | Dseven: not that xen isn't awesome, but given the poor support it gets in upstream testing it's always going to be an uphill battle | 17:44 |
| Dseven | marun: for reference https://bugs.launchpad.net/neutron/+bug/1293693 | 17:44 |
| marun | kevinbenton: that would be great but again, not likely this cycle | 17:45 |
| Dseven | marun: Yeah, I know, but for .... non-technical reasons .. I'm stuck with it | 17:45 |
| marun | kevinbenton: we're not looking for a solution beyond the workaround at this point | 17:45 |
| Sukhdev | rkukura: I was going to say the other way around :-) but, am OK either way | 17:45 |
| rkukura | Sukhdev: Want to start an etherpad or google doc to describe the issue, organize ideas, …? | 17:45 |
| marun | kevinbenton: it's not for technical reasons that we're pursuing the workaround, it's to minimize friction getting something merged to nova | 17:45 |
| kevinbenton | marun: so it would satisfy people that want non-hybrid plugging | 17:46 |
| Sukhdev | rkukura: yes, how about google doc? | 17:46 |
| marun | kevinbenton: I think they'll have to live with that in the short term | 17:46 |
| rkukura | Sukhdev: sure | 17:46 |
| marun | kevinbenton: if everyone was so concerned, they might have wanted to push on this issue a bit sooner | 17:46 |
| Sukhdev | I can throw in some initial thoughts and then you can jump in and add/clean? | 17:46 |
| marun | kevinbenton: now, it's too late. | 17:46 |
| marun | kevinbenton: (for icehouse) | 17:46 |
| rkukura | Sukhdev: grest | 17:46 |
| Sukhdev | rkukura: I will write it up over the weekend and send it to you. | 17:47 |
| rkukura | Sukhdev: great! | 17:47 |
| marun | Dseven: i hear both amazon and rackspace would be on kvm if they could... | 17:47 |
| Sukhdev | rkukura: cool - will ping you once I have something on the doc | 17:47 |
| Sukhdev | rkukura: early next week | 17:47 |
| Dseven | marun: I guess I would have pushed the issue a bit sooner, except the logic as it stands in the code today seems to actually support what I want to do, aside from the bug mentioned above | 17:47 |
| marun | Dseven: which bug? | 17:48 |
| *** leseb has joined #openstack-neutron | 17:48 | |
| Dseven | marun: the libvirt+Xen one - https://bugs.launchpad.net/neutron/+bug/1293693 | 17:48 |
| Dseven | marun: Basically, when non-hybrid plugging is used, the external-ids info doesn't get set on the OVS port, so neutron ignores the port and doesn't set the VLAN tag | 17:49 |
| marun | fun | 17:49 |
| Dseven | yeah | 17:50 |
| marun | well, on the upside hybrid plugging will always be used :) | 17:50 |
| marun | so things should work. just a bit slower | 17:50 |
| *** morganfainberg_Z is now known as morganfainberg | 17:50 | |
| marun | Dseven: are you not intending to support security groups then? | 17:51 |
| Dseven | yeah, the hackaround sortof mitigates that bug ... but I am a bit worried about the bridge being a bottlebeck | 17:51 |
| marun | Dseven: or are you working on a plugin that does the work itself? | 17:51 |
| marun | Dseven: I don't think it's that signifacant | 17:51 |
| Dseven | we're working with a firewall vendor to develop a soluton that runs in a VM (on each server) and all traffic is piped through it via OVS | 17:51 |
| Dseven | TBH, I don't know yet how that's going to plug into neutron, etc. | 17:52 |
| *** dave_tucker is now known as dave_tucker_zzz | 17:52 | |
| jogo | looks like there may be a new neutron bug in the gate | 17:53 |
| kevinbenton | marun: https://review.openstack.org/#/c/83148/ | 17:53 |
| jogo | http://status.openstack.org/elastic-recheck/data/uncategorized.html and click on 24 hours | 17:53 |
| *** itzikb1 has joined #openstack-neutron | 17:54 | |
| *** itzikb has quit IRC | 17:54 | |
| jogo | marun: ^ | 17:55 |
| kevinbenton | russelb: https://review.openstack.org/#/c/83148/ | 17:56 |
| marun | jogo: cripes, | 17:56 |
| *** saju_m has joined #openstack-neutron | 17:57 | |
| marun | kevinbenton: you do realize that he's busy as heck right now? | 17:57 |
| marun | kevinbenton: why are you pinging him when we've already chosen a path? | 17:57 |
| russellb | yes let's stick with the is_neutron() hack | 17:58 |
| jogo | arosen: ^^ | 17:58 |
| kevinbenton | marun: sorry. just seemed like like a painful limitation in icehouse for the people that didn't want the hybrid mode | 17:58 |
| *** roeyc has quit IRC | 17:58 | |
| russellb | yes it's unfortunate | 17:58 |
| marun | kevinbenton: like I said, we can deal with that fallout and see if a workaround/backport is possible next release | 17:58 |
| russellb | but it's also very unfortunate that lack of testing let it get until days before RC to bring it up | 17:58 |
| marun | kevinbenton: but this is a wake up call as much as anything - folks need to pay attention to severe issues and not just pass the buck | 17:59 |
| jogo | hmm this may be on the nova side | 17:59 |
| *** crc32 has joined #openstack-neutron | 18:00 | |
| kevinbenton | marun: i agree, but now the ones that take the performance hit are the only ones that legitimately didn't have to pay attention to this issue :) | 18:01 |
| kevinbenton | marun: since they aren't using the firewall | 18:01 |
| marun | kevinbenton: who doesn't have to pay attention? | 18:01 |
| marun | kevinbenton: involvement in neutron means 'involvement' | 18:01 |
| russellb | folks not thinking they have to care is a pretty core part of neutron's problem | 18:01 |
| *** luqas has quit IRC | 18:02 | |
| rkukura | marun, russellb, Dseven, salv-orlando: I hate to keep beating an almost dead horse, but the port_filter flag in binding:vif_details is already there in the current nuetron code, I think with all plugins. With no change on the neutron side, couldn’t we come up with a simple enough and low enough risk patch to nova so that the GenericVIFDriver uses this flag to decide whether or not to do the hybrid bridging with vif_type is ‘ovs | 18:02 |
| russellb | and is causing neutron to suffer greatly overall | 18:02 |
| marun | rkukura: you keep telling us this. we hear you. the answer is 'not this cycle' | 18:02 |
| kevinbenton | marun, russellb: i'm just referring to a regular neutron user | 18:02 |
| russellb | kevinbenton: heh, OK. | 18:02 |
| * russellb a little sore on the subject | 18:02 | |
| marun | kevinbenton: well, they suffer for our mistakes, sadly. | 18:02 |
| russellb | rkukura: interested in putting together a patch to consider? | 18:03 |
| rkukura | marun: I’m suggesting a compromise, where we leave neutron as-is, and just let nova use the data it already gets. | 18:03 |
| russellb | if there's data we can use for this already, that does sound even better | 18:04 |
| marun | rkukura: if nova calls neutron on every vif plug, so that the port details is always available, then it could be an option. | 18:04 |
| marun | rkukura: please verify that first | 18:04 |
| rkukura | russellb: I’ll take a closer look at what the nova code does between the port_create/port_update and calling the GenericVIFDriver. I’m not very familiar with this code. | 18:04 |
| kevinbenton | rkukura, marun: isn't that already necessary for selecting the vif_type? | 18:06 |
| rkukura | If nova is getting binding:vif_type, its also getting binding:vif_details. | 18:08 |
| *** Sukhdev has quit IRC | 18:10 | |
| *** dave_tucker_zzz is now known as dave_tucker | 18:12 | |
| HenryG | arosen: ping | 18:12 |
| HenryG | amotoki: ping | 18:13 |
| *** dave_tucker is now known as dave_tucker_zzz | 18:15 | |
| *** Jabadia has joined #openstack-neutron | 18:16 | |
| *** Jabadia has quit IRC | 18:16 | |
| *** luqas has joined #openstack-neutron | 18:16 | |
| *** Jabadia has joined #openstack-neutron | 18:17 | |
| *** luqas has quit IRC | 18:20 | |
| kevinbenton | _cjones_: are all of the tests editing the same backend DB? | 18:21 |
| *** thuc has quit IRC | 18:23 | |
| beagles | rkukura, I think the crux of your proposal is being missed. Why not point out what the change would be to vif driver would be if it were keyed off of vif_details? | 18:25 |
| carl_baldwin | emagana: ping | 18:25 |
| *** salv-mobile has quit IRC | 18:25 | |
| rkukura | beagles: Working on a patch to do just that | 18:25 |
| beagles | rkukura, if we can presume that it is always available to network_info, etc. ... the real difference would be just that | 18:25 |
| jogo | marun arosen salv-orlando: https://bugs.launchpad.net/neutron/+bug/1297992 | 18:25 |
| *** salv-mobile has joined #openstack-neutron | 18:25 | |
| *** jgallard has quit IRC | 18:25 | |
| rkukura | beagles: exactly | 18:26 |
| beagles | rkukura, my proposal only differed in that something nova specific be included in the model as opposed to the vif_details, so the gist is essentially the same and more of a design/cohesion nit | 18:26 |
| beagles | rkukura, and I think that element is being overlooked by the implied complexity of the mysterious vif_details dict ;) | 18:27 |
| rkukura | probably less code involved to just append vif_details to the VIF object | 18:27 |
| *** mlavalle has quit IRC | 18:28 | |
| kevinbenton | rkukura: are the vif_details not currently passed to nova as part of the VIF object? | 18:28 |
| openstackgerrit | Arvind Somya proposed a change to openstack/neutron: Cisco APIC ML2 mechanism driver, part 2 https://review.openstack.org/73372 | 18:28 |
| beagles | could be... why not fling it and see if it sticks. We have been operating under the premise that the orig patch was too.. something.. but perhaps we are throwing baby out with the bathwater | 18:29 |
| rkukura | kevinbenton: They are passed to nova as a port attribute, but nova isn’t appending vif_details to the VIF object. | 18:29 |
| beagles | rkukura, for example.. if the notion of the vif_details as an element of the vif object is abhorrent then it could either be backed off to something more specific or simply "no-way-man" for now | 18:30 |
| *** thuc has joined #openstack-neutron | 18:30 | |
| beagles | rkukura, i'm just trying to cover the bases now ... so I don't have to ask myself why I didn't push harder again later on :) | 18:31 |
| *** pnavarro has quit IRC | 18:31 | |
| *** jpich has quit IRC | 18:33 | |
| rkukura | beagles: Should have something concrete to discuss soon. | 18:34 |
| kevinbenton | rkukura: are you posting a patch? | 18:35 |
| rkukura | kevinbenton: yes | 18:36 |
| *** blogan has joined #openstack-neutron | 18:37 | |
| *** _sweston_ has joined #openstack-neutron | 18:37 | |
| *** sweston has quit IRC | 18:38 | |
| Dseven | BTW, is the use-case that requires an urgent fix for icehouse consisely described somewhere? i.e. why is Hybrid plugging required when the iptables firewall driver is not in use? | 18:38 |
| marun | Dseven: you never did answer my question regarding security groups | 18:39 |
| marun | Dseven: how are you intending to support them? | 18:39 |
| *** beagles is now known as beagles_brb | 18:40 | |
| *** salv-mobile has quit IRC | 18:40 | |
| Dseven | marun: we're working with a firewall vendor to implement firewalling that runs in a VM on each server, with all instance traffic piped through it by OVS .... but I'm not sure how that will related to security groups yet | 18:40 |
| Dseven | marun: so maybe it will involve a custom firewall driver .. I don't know yet | 18:41 |
| marun | Dseven: lots of work | 18:41 |
| marun | Dseven: definitely not going for the easy path :) | 18:41 |
| *** flwang has quit IRC | 18:41 | |
| Dseven | marun: pffft ... easy is no fun ..... or something | 18:41 |
| marun | Dseven: Accidental complexity is never fun | 18:42 |
| *** salv-mobile has joined #openstack-neutron | 18:42 | |
| marun | Dseven: in any case, you have lots of challenges to run through before the performance hit of a linux bridge is going to be a concern | 18:42 |
| marun | Dseven: I'd recommend monitoring the situation and continuing to push for a real solution/backport to icehouse. | 18:42 |
| marun | Dseven: by the time you're ready to worry about network performance, there should be a fix | 18:43 |
| marun | Dseven: sooner if rkukura's gambit ends up successful | 18:43 |
| Dseven | marun: that approach may work ... so long as whatever OVS tricky is involved doesn't conflict with the hybrid plugging | 18:43 |
| marun | Dseven: OVS trickery? | 18:43 |
| Dseven | trickery | 18:43 |
| openstackgerrit | Yuriy Taraday proposed a change to openstack/neutron: Abstract out root_helper calls to classes https://review.openstack.org/82787 | 18:43 |
| Dseven | marun: trickery to make all traffic for instances pass through this firewall VM | 18:44 |
| marun | Dseven: ah, right | 18:44 |
| nati_ueno | UT and devstack tested. please review https://review.openstack.org/#/c/82904/ | 18:44 |
| marun | Dseven: my view is that NFV like that is dino tech. do the filtering directly on ovs. | 18:45 |
| marun | Dseven: but vendors would disagree | 18:45 |
| *** yfried has quit IRC | 18:46 | |
| Dseven | marun: perhaps .. I'm not a firewall guru ... I'll try to press my peer (who is) a bit on why we're going down this road, but he seems to believe it's the right one | 18:46 |
| marun | “It is difficult to get a man to understand something, when his salary depends on his not understanding it.” -- Upton Sinclair | 18:47 |
| rkukura | marun, Dseven: There has been some progress on an ovs-firewall-driver BP, but it didn’t complete for icehouse. | 18:48 |
| *** overlayer has quit IRC | 18:48 | |
| marun | rkukura: I'm looking forward to that. But looking forward to ODL more ;) | 18:48 |
| Dseven | Oklahoma Dept of Libraries ? | 18:49 |
| Dseven | :) | 18:49 |
| emagana | carl_baldwin: pong | 18:50 |
| salv-orlando | jogo: are you positive the traceback for bug 1297992 is common to most uncategorized failures? | 18:50 |
| salv-orlando | jogo: just looking whether I just need to look at the bug or all failures | 18:51 |
| banix | nati_ueno: Does this affect plugins that do not support firewall by setting it to noop? | 18:51 |
| salv-orlando | hi folks I am back. | 18:51 |
| marun | Dseven: I'm assuming you're just joshing but in case not - opendaylight | 18:51 |
| salv-orlando | So is there already a patch to revieW? | 18:51 |
| banix | nati_ueno: Does this affect plugins that do not support security groups by setting the firewall to noop? | 18:51 |
| carl_baldwin | emagana: Hey, a sudo patch has been discovered that helps reduce sudo overhead when agents run commands with sudo root wrap. I was wondering if there is a place in the documentation where we document OS tweaks like this. | 18:52 |
| *** BuSerD has joined #openstack-neutron | 18:52 | |
| marun | banix: yes | 18:52 |
| Dseven | marun: thanks, I'll check that out ... google didn't make it obvious | 18:52 |
| marun | banix: all neutron plugins will use hybrid plugging with the proposed patch | 18:52 |
| nati_ueno | banix: you are right | 18:52 |
| *** jecarey has quit IRC | 18:53 | |
| salv-orlando | marun, nati-ueno: so which approach was chosen? and most importantly, where's the patch. I'm sorry I ended up offline for most of the past hours. | 18:53 |
| marun | Dseven: opensource SDN controller, allows more complex use of OVS | 18:53 |
| marun | salv-orlando: your workaround is the proposed solution | 18:53 |
| banix | marun, nati_ueno and by hybrid plugging you mean the vif driver used? Forgive my ignorance please. | 18:53 |
| Dseven | marun: Yep, I see it now .. will explore | 18:53 |
| nati_ueno | banix: yes | 18:53 |
| marun | salv-orlando: rkukura is investigating if a vif details solution is possible, but until he determines that it is, we're working on the assumption that your workaround is the way to go | 18:54 |
| _cjones_ | kevinbenton: As for the configuration issue. It appears that I need to get my new configuration in oslo. Pointers of how to get this portion done? I'll be golden. | 18:54 |
| salv-orlando | marun: what about complaint about ovs-based plugins which do not want to use hybrid plugging? | 18:54 |
| _cjones_ | kevinbenton: https://gist.github.com/cjones-/1f534d6fa09421a46309 | 18:54 |
| marun | salv-orlando: they're SOL | 18:54 |
| salv-orlando | SOL? | 18:54 |
| salv-orlando | missing this acronym | 18:54 |
| marun | salv-orlando: s*** out of luck | 18:54 |
| emagana | carl_baldwin: excellent point! | 18:54 |
| banix | nati_ueno: and if we need the Generic driver, will explicitly setting that in nova configuration would be enough? | 18:54 |
| emagana | the patch is merged right? | 18:55 |
| marun | salv-orlando: out of curiosity, where is mark? | 18:55 |
| Dseven | I believe that hybrid plugging is implement by the Generic VIF driver .. it's just logic within the driver that's being tweaked | 18:55 |
| salv-orlando | I think nati-ueno initially did a vif_details patch, so why is now rkukura reinvestigating what nati-ueno already did? | 18:55 |
| carl_baldwin | emagana: The patch is found in a newer released version of sudo. Is that what you are asking? | 18:55 |
| marun | salv-orlando: he's like a dog with a bone | 18:55 |
| nati_ueno | banix: this fix don't need configuration change | 18:55 |
| emagana | carl_baldwin: yes! | 18:56 |
| salv-orlando | if vif_details is the way, ignore my comments, address nova concerns around merging this late in the release cycle and go ahead. | 18:56 |
| banix | nati_ueno, marun: or we won't be able to use the Generic driver? | 18:56 |
| marun | salv-orlando: if he can come up with something simple enough that nova is willing to accept it, then great | 18:56 |
| carl_baldwin | emagana: I don't have all the details at my fingertips but I can provide them in an email or something. | 18:56 |
| rkukura | salv-orlando: I suggested a compromise where we simply use the existing flag neutron already puts in vif_details to control whether or not the hybrid bridge is created in nova | 18:56 |
| emagana | carl_baldwin: please, do that and I will file a BP/bug to make sure we document it | 18:56 |
| banix | nati_ueno: what if we want to use the generic driver? (Sorry i dont know the context and tying to figure out if I need to do anything with our plugin.) | 18:56 |
| marun | salv-orlando: we're not adding a new driver | 18:56 |
| marun | oops, banix ^^ | 18:56 |
| rkukura | So nati_ueno’s finer-grained control could be added later | 18:57 |
| *** WormMan has joined #openstack-neutron | 18:57 | |
| salv-orlando | I think the simplest thing would be to let neutron tell nova what to use to plug, without changing the API. But still it was said here that it's too much of a change. | 18:57 |
| carl_baldwin | emagana: Great, I will provide it in the next day. | 18:57 |
| nati_ueno | banix: newest patch is using Generic driver | 18:57 |
| emagana | carl_baldwin: Thanks! | 18:57 |
| carl_baldwin | Thank you. | 18:57 |
| salv-orlando | so yeah, I guess there is always the trick to skip the hybrid plugging | 18:57 |
| salv-orlando | of willfully setting a wrong firewall driver in nova.conf | 18:57 |
| *** flwang has joined #openstack-neutron | 18:57 | |
| banix | marun, nati_ueno : thanks; still confused but will follow the discussion. | 18:57 |
| salv-orlando | so that it won't do hybrid plugging but won't either configure nova firewalling because neutron is in place | 18:58 |
| nati_ueno | banix: no worries, I'm confusing too X( | 18:58 |
| marun | salv-orlando: did you see kevinbenton's proposal? | 18:58 |
| marun | salv-orlando: creating a neutron-specific no-op firewall driver? | 18:58 |
| *** yfried has joined #openstack-neutron | 18:58 | |
| marun | salv-orlando: not something nova will accept this cycle, but maybe a good workaround for next | 18:58 |
| kevinbenton | marun: i already overrode that patch with what rkukura was getting at :-) | 18:59 |
| *** salv-mobile has quit IRC | 18:59 | |
| marun | too many cooks | 19:00 |
| *** yfried has quit IRC | 19:01 | |
| marun | salv-orlando: I'm of the opinion that we should just force hybrid plugging for everyone. | 19:01 |
| *** yfried has joined #openstack-neutron | 19:01 | |
| marun | salv-orlando: maybe then the folks impacted by the change would actually sit up and pay attention. | 19:01 |
| marun | salv-orlando, rkukura: dividing our attention between different fixes is the reason we're in this mess in the first place. | 19:02 |
| salv-orlando | marun: why are they at fault for anything? | 19:02 |
| marun | salv-orlando: if they aren't paying attention, they're not sufficiently involved. | 19:02 |
| jogo | salv-orlando: I did a spot check the traceback was in many uncategorized failures | 19:02 |
| salv-orlando | jogo: thanks | 19:02 |
| marun | salv-orlando: just like most people (including me) sit back while a core group of people (including you) work on fixing critical core issues. | 19:02 |
| jogo | I have an e-r patch up so once it lands we will know | 19:02 |
| *** jecarey has joined #openstack-neutron | 19:02 | |
| rkukura | marun: I think the lesson to learn is to do these things in small steps as soon as possible | 19:02 |
| marun | rkukura: I think the lesson is that we're not sufficiently organized to pay attention to things that matter. | 19:03 |
| marun | rkukura: what you said -> in the small | 19:03 |
| marun | rkukura: what i said -> in the large | 19:03 |
| salv-orlando | marun: by this standard shalle we start randomly breaking plugins where there aren't frequent contributions to make them pay attention? | 19:03 |
| marun | rkukura: I would agree that both views are necessary, but we're hurting in the large more than the small. | 19:03 |
| *** luqas has joined #openstack-neutron | 19:03 | |
| marun | salv-orlando: not randomly | 19:04 |
| marun | salv-orlando: but this bug has been brought up in what, every damn meeting this cycle? | 19:04 |
| marun | salv-orlando: it's not like the situation hasn't been very public | 19:04 |
| *** thuc has quit IRC | 19:05 | |
| marun | salv-orlando: like the testing problems of the past year, people don't sit up and pay attention unless it impacts them | 19:05 |
| marun | salv-orlando: i.e. 'my patches can't merge because of persistent test failures' | 19:05 |
| *** thedodd has quit IRC | 19:05 | |
| marun | salv-orlando: or 'my network performance is taking a hit because of vif plugging issues'! | 19:05 |
| salv-orlando | marun: I'm not talking about the specific fact that a group of plugins will be negatively affected. I'm simply advocating that stating that people who will be affected will somehow be punished for their laziness makes no sense. | 19:07 |
| salv-orlando | As an example NSX plugin will be affected, and if this merges we'll have to devise a strategy | 19:08 |
| salv-orlando | still I'm not saying no you don't merge this because NSX latencies will be hit | 19:08 |
| *** rotbeard has quit IRC | 19:08 | |
| *** markmcclain has joined #openstack-neutron | 19:08 | |
| salv-orlando | on the other hand, putting too many people on the same issue is going to be worse than putting no people on it | 19:09 |
| marun | salv-orlando: +1 | 19:09 |
| salv-orlando | especially with the kind of very strongly-minded, highly-skilled and experienced people you find in communities like this | 19:09 |
| salv-orlando | some a**holes too | 19:09 |
| salv-orlando | but that would be mostly me | 19:09 |
| marun | you mean me? ;) | 19:11 |
| *** irenab has quit IRC | 19:17 | |
| *** jp_at_hp has quit IRC | 19:20 | |
| Dseven | so is the proposal still salv's approach? - i.e. basically if is_neutron() then do_hybrid | 19:20 |
| *** leseb has quit IRC | 19:24 | |
| enikanorov_ | markmcclain: hi, sorry for bothering again with https://review.openstack.org/#/c/81537/ could you please take a look | 19:26 |
| *** dave_tucker_zzz is now known as dave_tucker | 19:27 | |
| *** markwash has quit IRC | 19:27 | |
| *** thuc has joined #openstack-neutron | 19:27 | |
| *** thuc_ has joined #openstack-neutron | 19:28 | |
| *** SumitNaiksatam has quit IRC | 19:31 | |
| *** leseb has joined #openstack-neutron | 19:32 | |
| *** thuc has quit IRC | 19:32 | |
| *** leseb has quit IRC | 19:34 | |
| banix | marun, salv-orlando: Please ignore if busy. I am still confused and have a question regarding the patches being discussed for the plugging: Does the proposed solution which sets the plugging to hybrid prevents all Neutron plugins from using anything but the hybrid plugging? In other words, can we use the Generic plugging after/if/when this patch is merged? | 19:35 |
| Dseven | banix: there is no hybrid plugin .. there is a generic VIF plugin for Nova ... there is logic within that plug that decides what type of plugging to do | 19:36 |
| *** overlayer has joined #openstack-neutron | 19:36 | |
| Dseven | banix: sorry, too many plugs ;) ... ".. there is logic with the generic VIF driver in nova that decides what type of plugging to do" | 19:37 |
| Dseven | banix: https://github.com/openstack/nova/blob/master/nova/virt/libvirt/vif.py#L465 | 19:38 |
| banix | Dseven: thanks weren't at some point two different classes one LibvirtGenericVIFDriver and another for Hybrid? I see the latest code has only one. | 19:39 |
| marun | banix: the proposal is that _all_ plugins will be forced to use hybrid plugging | 19:39 |
| marun | banix: you don't have to consider vif drivers at all. | 19:39 |
| Dseven | banix: yes, I believe that old plugin was removed and everything collapsed into the generic VIF driver | 19:40 |
| marun | banix: the generic vif driver will force hybrid plugging if neutron is enabled. | 19:40 |
| banix | marun: what is the implications of that? I will look at the code but in case there is a short answer to this | 19:41 |
| marun | banix: the implication is that there will be a linux bridge between the vif and ovs bridge for all plugins | 19:41 |
| marun | banix: regardless if they intend to do iptables filtering | 19:41 |
| marun | banix: so unnecessary overhead | 19:41 |
| Dseven | marun: why is it that we need to do this? is there some use case other than iptables filtering that still requires the hybrid plugging ? | 19:42 |
| marun | Dseven: no | 19:42 |
| *** SumitNaiksatam has joined #openstack-neutron | 19:42 | |
| rkukura | marun, Dseven, russellb, kevinbenton, salv-orlando: Potential nova security group fix is: https://review.openstack.org/#/c/83190/ | 19:42 |
| banix | marun: thanks much; i think I understand this now better; so this will apply to all vifs that are of type OVS? | 19:43 |
| marun | banix: correct | 19:43 |
| Dseven | marun: so what is the problem we're trying to solve then? i.e. what does not work with the code as-is ? | 19:44 |
| marun | Dseven: we need to force hybrid plugging for any plugin that relies on iptables filtering between vif and ovs bridge | 19:44 |
| marun | Dseven: ml2+ovs - the default plugin tested in the gate - simply doesn't support security groups without this fix | 19:45 |
| marun | Dseven: and plugins that use the same approach are also broken | 19:45 |
| banix | marun: Then that would potentially have implications beyond the performance. May be not but i am wondering. In our plugin we have a controller managing and setting up the rules for OVSs we have and am wondering if the linux bridges lead to any issues…. | 19:46 |
| Dseven | marun: what type of plugin are we talking about? Not the firewall_drver in nova ? | 19:46 |
| marun | banix: i'm not sure why it would unless you're touching the vif | 19:46 |
| marun | Dseven: neutron plugin | 19:46 |
| marun | banix: the linux bridge is a passthrough unless iptables-based security groups are enabled in neutron | 19:47 |
| marun | vif <-> linux bridge <-> ovs port | 19:47 |
| Dseven | marun: OK .. so there are some neutron plugins that depend on iptables, independently of the IptablesFirewallDriver in nova ? | 19:47 |
| marun | <-> -> veth pairs | 19:47 |
| marun | Dseven: correct. | 19:48 |
| marun | Dseven: it's basically the equivalent of the nova functionality | 19:48 |
| Dseven | marun: OK... I think I get it | 19:48 |
| banix | marun: ok; and I suppose a single vif will still show up as a single port on the OVS and nothing else will be seen by the OVS; and setting the firewall to noop will make sure there are no iptable rules affecting traffic, etc | 19:48 |
| marun | banix: correct | 19:49 |
| Dseven | I wonder if it might not be simpler to just have a nova config option to force hybrid plugging, rather than trying to come up with some logic for it last-minute | 19:49 |
| banix | marun: thanks for taking the time to explain this here. | 19:49 |
| marun | Dseven: we discussed that, a new config option is not an option this late in cycle | 19:49 |
| marun | banix: np :) | 19:49 |
| Dseven | marun: hmmm... but a chance in logic is? ...... | 19:49 |
| Dseven | s/chance/change/ | 19:50 |
| marun | Dseven: a change in logic is internal-only | 19:50 |
| marun | Dseven: a config option has to be documented | 19:50 |
| marun | Dseven: in any case, it's not our call | 19:50 |
| Dseven | so we can pretend it never happened? ;) | 19:50 |
| marun | Dseven: we're trying to merge something to nova - they decide what goes in. | 19:50 |
| marun | Dseven: so we have to make them happy | 19:50 |
| *** skraynev_afk is now known as skraynev | 19:54 | |
| *** claudiub has joined #openstack-neutron | 19:54 | |
| Dseven | marun: what neutron plugin(s) depend on iptables? Trying to rationalise how this relates to ML2 | 19:55 |
| marun | Dseven: I don't know off the top of my head | 19:55 |
| *** otherwiseguy has quit IRC | 19:57 | |
| markmcclain | Dseven: OVS+ml2 was the only one that relied on the old hybrid driver | 19:58 |
| *** pcm_ has quit IRC | 19:59 | |
| Dseven | markmcclain: I'm missing a piece somewhere .... ML2 doesn't handle layer 3 filtering, does it ? | 20:00 |
| markmcclain | it does not, but it implements the neutron security group extension | 20:01 |
| Dseven | and it does that by configuring iptables ? | 20:02 |
| markmcclain | so previously the OVS plugin used a hybrid driver to implement security groups | 20:02 |
| *** pcm_ has joined #openstack-neutron | 20:02 | |
| *** pcm_ has quit IRC | 20:02 | |
| *** sacharya has joined #openstack-neutron | 20:03 | |
| *** pcm_ has joined #openstack-neutron | 20:03 | |
| *** julim_ has quit IRC | 20:03 | |
| markmcclain | there are folks working on proposals for changing open implementation from the hybrid to using flow mods | 20:04 |
| markmcclain | but that work is for Juno | 20:04 |
| Dseven | OK .. so .. meanwhile ... where is the code that interfaces with iptables? | 20:05 |
| rkukura | markmcclain: I posted a nova-only patch that uses the port_filter flag neutron currently includes in binding:vif_details. In juno, ML2 could control that based on which firewall driver is used on the compute node. | 20:06 |
| *** beagles_brb is now known as beagles | 20:06 | |
| *** thedodd has joined #openstack-neutron | 20:08 | |
| kevinbenton | salv-orlando: this was the other approach that i had brought up. https://review.openstack.org/#/c/83148/ | 20:10 |
| Dseven | rkukura: isn't get_firewall_required() supposed to return True or False? You have it returning port_filter | 20:11 |
| kevinbenton | Dseven: port_filter is a boolean | 20:11 |
| Dseven | oh, ok | 20:11 |
| *** saju_m has quit IRC | 20:11 | |
| rkukura | Dseven: the ‘port_filter’ key in binding:vif_details maps to a boolean | 20:12 |
| markmcclain | rkukura, kevinbenton, nati_ueno: ping | 20:15 |
| openstackgerrit | Henry Gessau proposed a change to openstack/neutron: Remove auto-generation of db schema from models at startup https://review.openstack.org/40296 | 20:15 |
| kevinbenton | markmcclain: pong | 20:15 |
| markmcclain | Ok been stuck on a plane all day… Why do we have 3 patches? | 20:15 |
| rkukura | markmcclain: pong | 20:15 |
| kevinbenton | markmcclain: mine was just a hack if the other stuff didn't work | 20:16 |
| rkukura | markmcclain: If you are counting mine and kevinbenton’s, these are now very similar | 20:16 |
| kevinbenton | rkukura: i replaced the duplicate of yours back to my original one | 20:16 |
| rkukura | markmcclain: It seemed the vif_details approach was being rejected on the basis of too much change in nova, so kevinbenton and I both wanted to show it was very straightforward | 20:17 |
| kevinbenton | rkukura: but as long as the nova folks are good with merging the vif logic, mine isn't needed | 20:17 |
| rkukura | kevinbenton: what do yu mean by “merging the vif logic”? | 20:17 |
| kevinbenton | rkukura: your approach | 20:18 |
| rkukura | ok | 20:18 |
| *** sacharya has quit IRC | 20:18 | |
| rkukura | markmcclain: So my patch is basically what nati_ueno has been working towards, but without introducing the fine-grained control over VIF secuity. Its just on or off. | 20:19 |
| kevinbenton | markmcclain: the other patch is one marun had settled on had a side effect of forcing everyone to use hybrid plugging mode | 20:20 |
| markmcclain | forcing everyone to us hybrid plugging is super disruptive this late | 20:20 |
| markmcclain | s/us/use/ | 20:20 |
| kevinbenton | yeah, that's what prompted me to push up my patch. then rkukura suggested using the vif_details | 20:21 |
| kevinbenton | which is even better | 20:21 |
| markmcclain | rkukura: looks like unit test changes are needed for yours | 20:22 |
| *** sacharya has joined #openstack-neutron | 20:23 | |
| *** crc32 has quit IRC | 20:23 | |
| *** dfarrell07 has joined #openstack-neutron | 20:24 | |
| nati_ueno | markmcclain: pong | 20:25 |
| *** jprovazn is now known as jprovazn_afk | 20:25 | |
| markmcclain | nati_ueno: mainly trying to figure why we have 3 proposals | 20:25 |
| nati_ueno | markmcclain: three!? | 20:26 |
| *** tvardeman has quit IRC | 20:26 | |
| markmcclain | yes | 20:26 |
| russellb | heh, yeah, rkukura posted a new one | 20:27 |
| russellb | nati_ueno: was hoping you could tell me what you thought ... | 20:27 |
| russellb | new one is https://review.openstack.org/#/c/83190/ | 20:27 |
| nati_ueno | russellb: I think we should choise simplest one.. | 20:27 |
| russellb | simplest would be the is_neutron() hack i guess | 20:28 |
| nati_ueno | I agree | 20:28 |
| russellb | but i'm really curious about this new patch | 20:28 |
| russellb | because if the API already had the info needed to do this ... | 20:28 |
| russellb | then i'm confused why the bug has been open for like 9 months? | 20:28 |
| russellb | not sure what i'm missing i guess | 20:28 |
| markmcclain | russellb: I'm with you on this one too | 20:29 |
| rkukura | nati_ueno: This is similar to what you were proposing, but with just the current single port_filter flag rather than the fine grained flags | 20:29 |
| *** saju_m has joined #openstack-neutron | 20:29 | |
| nati_ueno | russellb: I'm working on fixing this issue. Mainly we are discussing vif security api. | 20:29 |
| nati_ueno | russellb: but neutron side ins't fixed yet.. | 20:30 |
| rkukura | in hindsight, we should have got the mechanism into both sides for passing vif_details, and dealt with the fine grained VIF security as a separate followon step | 20:30 |
| russellb | OK | 20:31 |
| russellb | so, the simple patch for now then ... | 20:31 |
| russellb | fine with me. | 20:31 |
| rkukura | russellb: Which is the simple patch? | 20:31 |
| russellb | heh, hard to track it all | 20:32 |
| *** jobewan has quit IRC | 20:32 | |
| Dseven | simple patch is Salv's approach? | 20:32 |
| russellb | yes, i thin | 20:32 |
| russellb | though he had 2 approaches | 20:32 |
| russellb | so it's nice and confusing | 20:32 |
| rkukura | nati_ueno: If we go with using the port_filter key in binding:vif_details, is work needed on the neutron side? | 20:32 |
| Dseven | this one https://launchpadlibrarian.net/170772780/hack.patch | 20:33 |
| russellb | https://review.openstack.org/#/c/82904/ | 20:33 |
| russellb | yeah | 20:33 |
| nati_ueno | rkukura: I think so,, but i'm not sure | 20:33 |
| nati_ueno | rkukura: I'm fixing ut for 82904, give me 1 hour | 20:33 |
| rkukura | nati_ueno: I know that ML2 passes True for the L2 agents that need the iptables firewall driver | 20:34 |
| nati_ueno | may be, if russelib is OK, we can do more better solution,, but I think it is safe start from simple step | 20:34 |
| russellb | if you want something for icehouse, the simpler the better, yes | 20:35 |
| rkukura | I can’t argue that 82904 isn’t simpler than 83190, but not by much | 20:35 |
| *** devlaps has quit IRC | 20:35 | |
| banix | markmcclain: wouldn't your comment "forcing everyone to us hybrid plugging is super disruptive this late" above apply to this patch? | 20:36 |
| rkukura | I’m concerned that the is_neutron() test is just too restrictive. | 20:36 |
| *** jlibosva has joined #openstack-neutron | 20:36 | |
| Dseven | too broad .. yeah | 20:36 |
| *** saju_m has quit IRC | 20:37 | |
| markmcclain | yeah because a blanket use of hybrid would impact all vendors | 20:37 |
| rkukura | So 83190, at worst case, lets each plugin return set port_filter to either True or False. | 20:37 |
| russellb | markmcclain: that's what the simplest fix does | 20:37 |
| russellb | markmcclain: forces that for all neutron | 20:37 |
| rkukura | The plugin can either hard-code this (as most do right now), or be more clever | 20:38 |
| marun | russellb: if we could get in rkukura's better fix, that would be preferable | 20:38 |
| marun | russellb: fallback is the 'force hybrid' workaround | 20:38 |
| marun | russellb: it's your call, really | 20:38 |
| russellb | marun: it does seem preferable, if it works | 20:38 |
| russellb | you guys tell me if it works | 20:38 |
| russellb | i don't have time to test this stuff out | 20:38 |
| marun | (this is where in-tree functional testing would be really useful) | 20:39 |
| russellb | but we've got to merge *something* by sometime tomorrow, or it may miss rc1 | 20:39 |
| marun | hint, hint: https://review.openstack.org/#/c/72585/ | 20:39 |
| markmcclain | marun: in tree functional test would have never caught this | 20:39 |
| marun | markmcclain: nonsense | 20:39 |
| russellb | well, you know ... testing of this functionality at all would have helped :-p | 20:39 |
| markmcclain | this is an integration problem | 20:39 |
| russellb | markmcclain: right | 20:39 |
| marun | markmcclain: the patch in question runs it against an integration environment | 20:39 |
| nati_ueno | ok I think we should have no more discussion time. markmcclain: russellb: could you decide which way we go? | 20:39 |
| marun | but from the neutron tree | 20:39 |
| marun | i.e. pre-merge testing rather than the post-merge of tempest | 20:40 |
| markmcclain | marun: tempest can be run pre-merge | 20:40 |
| markmcclain | there are ways to do it ask arosen | 20:40 |
| russellb | ok, so https://review.openstack.org/#/c/83190 ... | 20:41 |
| russellb | this is ready to go on the neutron side already? | 20:41 |
| *** devlaps has joined #openstack-neutron | 20:41 | |
| marun | rkukura: that's a yes, right? ^ | 20:42 |
| kevinbenton | markmcclain: isn't tempest run pre-merge already as part of the jenkins tests? | 20:42 |
| russellb | pretty sure rkukura said yes before | 20:42 |
| marun | kevinbenton: but we can't merge tempest tests until neutron functionality exists | 20:42 |
| markmcclain | kevinbenton: marun is wanting to pair the code change and tests together | 20:42 |
| nati_ueno | kevinbenton: marun: let's discuss ci issue later | 20:42 |
| marun | i'm talking out of my ass, really. in this case in-tree tests make zero sense. | 20:43 |
| marun | not when nova is involved. it's only functional api testing we should be running in an integration environment | 20:43 |
| russellb | let's fix the bug first :) | 20:43 |
| nati_ueno | OK so we choose 83190 ? | 20:43 |
| marun | russellb: waiting on rkukura to answer | 20:43 |
| openstackgerrit | Kevin Benton proposed a change to openstack/neutron: Disable XML tests on Py26 https://review.openstack.org/81865 | 20:43 |
| marun | nati_ueno: i think the goal is to get it tested and if it can pass muster we can use it | 20:44 |
| rkukura | russellb, marun: The ML2 plugin should be setting the right port_filter value based on which mechanism driver binds the port. For openvswitch and linuxbridge, this should be all set. | 20:44 |
| marun | nati_ueno: and if there are unexpected side-effects, we can fall back to workaround | 20:44 |
| nati_ueno | marun: you get a point | 20:44 |
| markmcclain | ok now that rkukura has confirmed it we don't need a followup neutron change | 20:45 |
| marun | nati_ueno: good to have a safety net :) | 20:45 |
| markmcclain | looks like we just need to add UT | 20:45 |
| *** julim has joined #openstack-neutron | 20:45 | |
| nati_ueno | ok so go with 83190 ? | 20:45 |
| russellb | and i'd like someone to do functional testing | 20:45 |
| rkukura | If any other plugin or ML2 driver sets the wrong port_filter value, the fix is trivial. | 20:45 |
| nati_ueno | russellb: I'll do | 20:45 |
| russellb | nati_ueno: great thanks | 20:45 |
| russellb | nati_ueno: so that can be plan A | 20:45 |
| russellb | plan B the "force it for all of neutron" hack | 20:45 |
| rkukura | I’ll add the unit test, and have addressed russellb’s comment | 20:45 |
| marun | rkukura: is there something we should be doing that tempest check jobs won't cover? | 20:45 |
| nati_ueno | OK if 83190 isn't merged in Today, let's go plan B | 20:45 |
| rkukura | would be great if someone else could look into functional testing | 20:46 |
| nati_ueno | rkukura: could you add UT for it? | 20:46 |
| marun | rkukura: it would be helpful to clarify so that people's efforts are focused. | 20:46 |
| rkukura | I’ll base the UT on what nati_ueno proposed in his patch | 20:46 |
| markmcclain | russellb: +1 for that plan | 20:46 |
| russellb | marun: wasn't there a test submitted upstream for this? | 20:46 |
| russellb | marun: that's what lpeer told me i think ... | 20:46 |
| marun | russellb: yes | 20:47 |
| russellb | markmcclain: cool | 20:47 |
| russellb | marun: k | 20:47 |
| marun | russellb: good point. we can manually run it locally | 20:47 |
| marun | russellb: https://review.openstack.org/#/c/62702/ | 20:47 |
| russellb | yeah | 20:47 |
| rkukura | marun: What needs clarification? | 20:47 |
| russellb | or just manually test for now too | 20:47 |
| russellb | but just saying, as far as preventing regression in the future ... | 20:47 |
| marun | rkukura: what to test - deploiy your branch and yfried's tempest branch with devstack, and running his test, | 20:48 |
| marun | rkukura: more testing than that would seem unnecessary given that all the tempest jobs should give us good coverage of everything else | 20:49 |
| nati_ueno | marun: I'm setting up the test | 20:49 |
| salv-orlando | as arosen did, you might be able to do that test in the check queue pushing a devstack patch which changes stackrc | 20:49 |
| salv-orlando | so the results will be immediately shared | 20:49 |
| rkukura | marun: Could use a volunteer to look at the tempest testing of this | 20:50 |
| marun | arosen: help with that? ^ | 20:50 |
| marun | rkukura: sounds like nati_ueno's on it | 20:50 |
| rkukura | great | 20:50 |
| arosen | salv-orlando: yes see: https://review.openstack.org/#/c/78694/ | 20:50 |
| marun | arosen: ah, clever | 20:50 |
| marun | nati_ueno: are you looking at this? | 20:51 |
| arosen | marun: I probably won't have the cycles today to look at the tempest stuff :/ | 20:51 |
| marun | arosen: none needed | 20:51 |
| arosen | We're about to upgrade our cloud to havana from grizzly | 20:51 |
| marun | arosen: we already have the test written | 20:51 |
| marun | arosen: best of luck! | 20:51 |
| arosen | marun: ah i though you're help with that comment was related to rkukura | 20:52 |
| marun | arosen: nah, I just wanted instructions on how to run unmerged-tempest tests upstream | 20:53 |
| arosen | marun: yea i see now :) | 20:54 |
| nati_ueno | https://review.openstack.org/#/c/83210/ | 20:54 |
| nati_ueno | <-- Bob's patch and yfield patch | 20:54 |
| nati_ueno | so we will see the test result in that patch | 20:54 |
| marun | nati_ueno: sweet! | 20:54 |
| nati_ueno | marun: yes | 20:54 |
| *** leseb has joined #openstack-neutron | 20:54 | |
| *** Jabadia has quit IRC | 20:55 | |
| salv-orlando | if you allow me to do that, I've started also vmware nsx checks. I won't bark if that doesn't work, but at least I want toknow | 20:55 |
| salv-orlando | I mean technically I've started them even if you do not allow me to do that. | 20:56 |
| marun | salv-orlando: not like it impacts anything :) | 20:56 |
| salv-orlando | marun: nasty vendor trying to get his way into opensoruce stuff! | 20:57 |
| openstackgerrit | Aaron Rosen proposed a change to openstack/neutron: Subnets should be set as lazy='join' https://review.openstack.org/83213 | 20:57 |
| nati_ueno | rkukura: 83190 isn't working. I posted the comment to fix | 20:59 |
| nati_ueno | rkukura: I get error with devstack | 20:59 |
| salv-orlando | nati_ueno: could that be your env? I think rkukura said he tested it already with ml2 | 21:00 |
| nati_ueno | rkukura: please ping me when you fix it. I'll do test again. Also could you recheck https://review.openstack.org/#/c/83210/ | 21:00 |
| rkukura | I have not done any devstack testing of this | 21:00 |
| nati_ueno | salv-orlando: IMO, code is wrong. Could you try it in your env? VM won't boot | 21:00 |
| rkukura | nati_ueno: Thanks - looking at that now | 21:01 |
| nati_ueno | rkukura: Thanks | 21:01 |
| *** rudrarugge has quit IRC | 21:01 | |
| marun | rkukura: still allergic to devstack? ;) | 21:02 |
| marun | rkukura: I hope we'll be able to run it in a container in the near future, so at least virtualization isn't involved. | 21:03 |
| openstackgerrit | Mohammad Banikazemi proposed a change to openstack/neutron: Deals with fails in update_*_postcommit ops https://review.openstack.org/83217 | 21:03 |
| *** singhs has joined #openstack-neutron | 21:03 | |
| *** sacharya has quit IRC | 21:05 | |
| rkukura | nati_ueno: in get_config_bridge(), can you clarify what you think needs to be done? | 21:09 |
| nati_ueno | rkukura: we don't need conf.filtername if it is neutron with security group | 21:10 |
| *** sridhar has quit IRC | 21:10 | |
| rkukura | So this needs a different test than get_firewall_required()? | 21:10 |
| nati_ueno | rkukura: yes | 21:11 |
| nati_ueno | rkukura: please take a look https://review.openstack.org/#/c/82904/ | 21:11 |
| rkukura | OK, same solution OK? | 21:12 |
| nati_ueno | rkukura: if we use the value of port_filter, it is OK | 21:12 |
| nati_ueno | rkukura: instead of is_neutron | 21:12 |
| rkukura | But when vif_details is available, get_firewall_required() is returning the value of port_filter. | 21:13 |
| *** skraynev is now known as skraynev_afk | 21:14 | |
| *** jlibosva has quit IRC | 21:14 | |
| nati_ueno | rkukura: That's way won't work | 21:15 |
| nati_ueno | rkukura: Two issues are mixed in this. conf.fitlername and vif selection | 21:16 |
| *** otherwiseguy has joined #openstack-neutron | 21:16 | |
| rkukura | Any reason your code won’t work here: | 21:16 |
| rkukura | if self.get_firewall_required(vif) and not utils.is_neutron(): | 21:16 |
| rkukura | conf.filtername = name | 21:16 |
| kevinbenton | rkukura: conf.filename needs to not be set if get_firewall_required returns true because of the VIF_DETAIL_PORT_FILTER | 21:16 |
| *** jprovazn_afk has quit IRC | 21:16 | |
| nati_ueno | rkukura: yes | 21:17 |
| nati_ueno | rkukura: if we are going to use port_filter, we should use that value | 21:17 |
| rkukura | that’s what the current patch is doing, isn’t it? | 21:18 |
| *** sacharya has joined #openstack-neutron | 21:18 | |
| nati_ueno | rkukura: no | 21:18 |
| nati_ueno | rkukura: ok, so | 21:18 |
| nati_ueno | there is many combinations | 21:19 |
| *** lifeless has quit IRC | 21:19 | |
| nati_ueno | (1) nova security group + nova network (2) neutron security group + neutron (3) nova security group + neutron | 21:19 |
| *** jorgem has joined #openstack-neutron | 21:19 | |
| nati_ueno | (1) (3) needs conf.filtername to be set | 21:19 |
| nati_ueno | (2) needs conf.filtername = None | 21:20 |
| nati_ueno | my bad | 21:20 |
| nati_ueno | There is (4) | 21:20 |
| nati_ueno | (2-1) neutron security group + neutron + iptables (2-2) neutron security group + neutron without iptables | 21:20 |
| nati_ueno | sorry not (4) but (2-1), (2-2) | 21:20 |
| nati_ueno | (2-1),(2-2) needs conf.filtername =None | 21:20 |
| nati_ueno | and... | 21:21 |
| kevinbenton | if self.get_firewall_required(vif) and not vif['details'].get(network_model.VIF_DETAIL_PORT_FILTER) | 21:21 |
| nati_ueno | (2-1-1) OVS (2-1-2) Not OVS | 21:21 |
| nati_ueno | (2-1-1) needs hybrid plugging | 21:21 |
| nati_ueno | kevinbenton: Thats the way | 21:21 |
| rkukura | nati_ueno: OK | 21:22 |
| nati_ueno | super complicated.... | 21:22 |
| Dseven | is that safe if the port_detail is not set at all ? | 21:22 |
| Dseven | (my python-fu is not so strong) | 21:23 |
| marun | Dseven: dicts as objects suck | 21:23 |
| kevinbenton | the vif model sets detail to an empty dict | 21:23 |
| kevinbenton | https://review.openstack.org/#/c/83190/1/nova/network/model.py | 21:23 |
| marun | Dseven: but hopefully the folks here are going to wrap dict access in a method call | 21:23 |
| rkukura | The whole point here is that only the GenericVIFDriver should know or care what keys to look for in vif_details. That lets things like SR-IOV pass the info they need without having to muck with intermediate code. | 21:25 |
| marun | kevinbenton: still, bad practice to expose the implementation of vif details outside the model | 21:25 |
| marun | kevinbenton: preferable to encapsulate access to vif details with methods on the model | 21:25 |
| rkukura | Wrapping at the VIF driver level is fine, just shouldn’t have to scatter this across the code | 21:25 |
| *** crc32 has joined #openstack-neutron | 21:25 | |
| marun | rkukura: it needs to be encapsulated somewhere | 21:26 |
| rkukura | accessors in the model are OK I guess | 21:26 |
| marun | rkukura: at least, if anything is actually going to look at it. | 21:26 |
| *** ale____ has joined #openstack-neutron | 21:26 | |
| *** ale____ has quit IRC | 21:26 | |
| marun | rkukura: if, as you say, it gets passed down raw, then no need for special handling | 21:26 |
| marun | rkukura: I'm just hoping we can move away from things like vif_details['blah'] sprinkled everywhere | 21:27 |
| *** ale____ has joined #openstack-neutron | 21:27 | |
| kevinbenton | marun: i don't see where that's being done for anything else in the vif | 21:28 |
| *** ale____ has quit IRC | 21:28 | |
| marun | kevinbenton: even once is too many times | 21:28 |
| kevinbenton | marun: or do you mean specifically for the nested stuff inside the 'details' dict | 21:28 |
| blogan | markmcclain: will you be attending the lbaas meeting tomorrow? | 21:28 |
| marun | kevinbenton: I think so | 21:28 |
| marun | kevinbenton: if model.vif_details is the extent of it, fine | 21:29 |
| rkukura | marun, kevinbenton : Should this do it: | 21:29 |
| marun | kevinbenton: if it's model.vif_details['foo'], that's bad | 21:29 |
| rkukura | and not utils.is_neutron(): | 21:29 |
| rkukura | pasted wrong thing | 21:29 |
| markmcclain | blogan: I should be able to attend | 21:29 |
| rkukura | if (self.get_firewall_required(vif) and | 21:29 |
| rkukura | vif['details'].get(network_model.VIF_DETAIL_PORT_FILTER) is None): | 21:29 |
| kevinbenton | rkukura: if i understood nati_ueno, that second one needs to also check if the filter is True | 21:31 |
| kevinbenton | i mean make sure it's not True | 21:31 |
| *** luqas has quit IRC | 21:31 | |
| kevinbenton | so either False or None | 21:31 |
| nati_ueno | kevinbenton: good point | 21:31 |
| kevinbenton | oh, i see get_firewall_required will handle if it's false | 21:32 |
| Dseven | I'm wondering if it makes more sense to have a separate function like get_is_neutron_handling_the_filtering(), and call it from both places | 21:32 |
| rkukura | None means port_filter is not in vif_details | 21:32 |
| marun | Dseven: +1 | 21:32 |
| nati_ueno | Dseven: +1 | 21:32 |
| marun | rkukura: please ^ | 21:32 |
| nati_ueno | Dseven: my previous patch was doing that thing | 21:32 |
| rkukura | Checking for False means either not present or present and False | 21:32 |
| kevinbenton | yeah, checking for present and False is handled by get_firewall_required | 21:33 |
| kevinbenton | so your suggestion works | 21:33 |
| openstackgerrit | Arvind Somya proposed a change to openstack/neutron: Cisco APIC ML2 mechanism driver, part 2 https://review.openstack.org/73372 | 21:34 |
| rkukura | nati_ueno, kevinbenton, marun: If any of you have a really clear picture of what’s needed here, I’d be happy for you to push a new patch with the fix. | 21:37 |
| kevinbenton | rkukura: ok | 21:39 |
| rkukura | I’ll push my latest version in a second | 21:39 |
| nati_ueno | rkukura: Thanks | 21:40 |
| *** sacharya has quit IRC | 21:41 | |
| rkukura | nati_ueno, kevinbenton, marun: new patch pushed - if someone wants to pick this up and add a get_is_neutron_handling_the_filtering() function, and the unit tests for this and get_firewall_required(), that would be great. | 21:44 |
| kevinbenton | o/ | 21:44 |
| nati_ueno | rkukura: OK let me take it | 21:48 |
| kevinbenton | nati_ueno: i'm on it | 21:48 |
| nati_ueno | kevinbenton: Ah OK | 21:48 |
| nati_ueno | kevinbenton: Please take it :P | 21:49 |
| *** overlayer has quit IRC | 21:52 | |
| kevinbenton | hmm, what is the expected behavior when VIF_DETAIL_PORT_FILTER is set to false by neutron? | 21:53 |
| kevinbenton | that could happen from a plugin not supporting security groups, right? | 21:53 |
| kevinbenton | and it might want to offload to nova | 21:53 |
| kevinbenton | so if VIF_DETAIL_PORT_FILTER is False, get_firewall_required shouldn't return False automatically, but instead continue to check CONF.firewall_driver | 21:54 |
| kevinbenton | does that make sense? | 21:54 |
| Dseven | makes sense to me | 21:55 |
| *** leseb has quit IRC | 21:57 | |
| *** b3nt_pin has joined #openstack-neutron | 21:57 | |
| *** baoli has quit IRC | 21:57 | |
| nati_ueno | +1 | 21:58 |
| marun | nati_ueno: is it ever desirable for neutron to rely on nova for security groups ?? | 21:58 |
| marun | nati_ueno: how does that ever make sense when ovs is used? | 21:58 |
| Dseven | maybe there are legacy neutron plugins that have that reliance ? | 21:59 |
| kevinbenton | marun: this was the case with the BigSwitch plugin until IceHouse | 22:00 |
| *** beagles has quit IRC | 22:00 | |
| *** jmeridth has left #openstack-neutron | 22:00 | |
| *** b3nt_pin is now known as beagles | 22:00 | |
| marun | kevinbenton: really? | 22:00 |
| kevinbenton | marun: yeah, because it required us to add a security groups only agent | 22:01 |
| nati_ueno | marun: so there is three case (1) use nova security group (2) use neutron security group (3) don't use anything | 22:01 |
| marun | kevinbenton: so hybrid plugging but nova managing iptables? | 22:01 |
| *** carl_baldwin has quit IRC | 22:01 | |
| blogan | markmcclain: good, would like to hear your thoughts on the pros and cons of the object model discussion | 22:01 |
| kevinbenton | marun: yes | 22:01 |
| marun | kevinbenton: yikes | 22:01 |
| kevinbenton | marun: well that was just nova security groups | 22:02 |
| kevinbenton | not really a hack, just slow to adopt :-) | 22:02 |
| markmcclain | blogan: I should be caught up on the discussion by then | 22:03 |
| blogan | markmcclain: excellent! look forward to it | 22:03 |
| *** banix has quit IRC | 22:04 | |
| *** mlavalle has joined #openstack-neutron | 22:04 | |
| *** ale____ has joined #openstack-neutron | 22:07 | |
| *** overlayer has joined #openstack-neutron | 22:12 | |
| *** luqas has joined #openstack-neutron | 22:12 | |
| *** armax has quit IRC | 22:13 | |
| *** ale____ has quit IRC | 22:14 | |
| Dseven | kevinbenton, are you going to post your version of the patch to 83190 ? I think I'm ready to test it (ML2+OVS) when it's ready.... | 22:16 |
| kevinbenton | Dseven: shortly. working on UT now | 22:17 |
| Dseven | roger | 22:17 |
| *** mwagner_lap has quit IRC | 22:17 | |
| *** carl_baldwin has joined #openstack-neutron | 22:22 | |
| *** bjornar has quit IRC | 22:22 | |
| *** baoli has joined #openstack-neutron | 22:23 | |
| *** beagles has quit IRC | 22:25 | |
| *** b3nt_pin has joined #openstack-neutron | 22:25 | |
| *** b3nt_pin is now known as beagles | 22:25 | |
| *** baoli has quit IRC | 22:26 | |
| *** blogan has quit IRC | 22:28 | |
| nati_ueno | Dseven: type apt-get install -y sl;sl <-- This kills time we are waiting kevin :P | 22:29 |
| *** carl_baldwin has quit IRC | 22:29 | |
| kevinbenton | https://review.openstack.org/#/c/83190/ | 22:29 |
| marun | kevinbenton: can you please post it w/o unit tests first so we can target nati_ueno 's devstack change for upstream tempest validation? | 22:30 |
| kevinbenton | marun: i posted it with a unit test, would you like me to remove it? :-) | 22:30 |
| marun | :p | 22:30 |
| kevinbenton | nati_euno: sl is awesome :-) | 22:31 |
| nati_ueno | kevinbenton: he he he | 22:31 |
| marun | kevinbenton: please replace the line 174-175 with a self-documenting conditional | 22:32 |
| marun | kevinbenton: i.e | 22:32 |
| marun | filtername_required = (self.get_firewall….) | 22:33 |
| marun | if filtername_required | 22:33 |
| kevinbenton | ok | 22:33 |
| marun | kevinbenton: otherwise, lgtm | 22:34 |
| *** mlavalle has quit IRC | 22:34 | |
| marun | not sure why the definition of name isn't inside the conditional, since it's not used elsewhere, but can't really change something like that in this patch | 22:34 |
| nati_ueno | kevinbenton: One comment | 22:35 |
| kevinbenton | nati_ueno: yeah? | 22:36 |
| nati_ueno | kevinbenton: ah commented on the gerrit | 22:36 |
| kevinbenton | nati_ueno: checking | 22:36 |
| Dseven | lol ... just installed sl | 22:37 |
| nati_ueno | Dseven: he he he | 22:37 |
| Dseven | is there a gerp too? ;) | 22:37 |
| nati_ueno | Dseven: you can invent it :) | 22:38 |
| Dseven | OK, will have to come up with an acronym that works | 22:38 |
| nati_ueno | Dseven: :P | 22:39 |
| *** jgrimm has quit IRC | 22:39 | |
| nati_ueno | I confiremed the patch is working with devstack + ml2 + ovs | 22:40 |
| *** ale has joined #openstack-neutron | 22:41 | |
| *** thedodd has quit IRC | 22:45 | |
| Dseven | hmm, looks like I'm behind on some other changes to api.py | 22:47 |
| kevinbenton | nati_ueno: did you see my comment? | 22:51 |
| kevinbenton | nati_ueno: oh nevermind | 22:51 |
| kevinbenton | i had to make another slight change to the UT because it was mangling the self.vif_ovs | 22:51 |
| nati_ueno | kevinbenton: yep, you are right | 22:51 |
| kevinbenton | salv-orlando: ping | 22:53 |
| salv-orlando | yeah? | 22:53 |
| kevinbenton | i think 3 is covered. if someone wants to implement security groups without hybrid plugging, they could set cap_filter to false | 22:54 |
| kevinbenton | or is that an abuse of cap_filter? | 22:54 |
| kevinbenton | in the port bindings | 22:55 |
| *** overlayer has quit IRC | 22:55 | |
| salv-orlando | who would be this fool which implements security groups this way? | 22:55 |
| salv-orlando | anyway, it might work. But out of change. | 22:56 |
| salv-orlando | chance. | 22:56 |
| salv-orlando | because the intention of the code is that cap_port_filter=False then means no security groups | 22:56 |
| *** harlowja has quit IRC | 22:58 | |
| salv-orlando | assuming the current approach is approved VNIC_TYPE might be the VIF_DETAILS attribute to leverage. It can be HYBRID or NATIVE and for simplicity you can let nova assume it's HYBRID by default | 23:00 |
| salv-orlando | it will also solve the fact which now you're making the VIF driver select the plugging mode according to the capability of doing filters | 23:01 |
| *** harlowja has joined #openstack-neutron | 23:03 | |
| *** ale has quit IRC | 23:03 | |
| *** julim has quit IRC | 23:05 | |
| kevinbenton | salv-orlando: i re-uploaded. can you check to see if i responded to your comments adequately? | 23:06 |
| *** juice has quit IRC | 23:09 | |
| *** Trozz has joined #openstack-neutron | 23:10 | |
| *** dims_ has quit IRC | 23:12 | |
| kevinbenton | russelb: ping | 23:13 |
| *** luqas has quit IRC | 23:14 | |
| kevinbenton | rkukura: ping | 23:21 |
| *** jorgem has quit IRC | 23:22 | |
| *** nati_ueno has quit IRC | 23:23 | |
| *** nati_ueno has joined #openstack-neutron | 23:23 | |
| rkukura | kevinbenton: pong | 23:25 |
| kevinbenton | what is the VNIC_TYPE being sent in vif_details by ML2 | 23:26 |
| kevinbenton | rkukura: ^^ | 23:26 |
| *** juice has joined #openstack-neutron | 23:26 | |
| kevinbenton | it can change per binding type, correct? | 23:27 |
| rkukura | kevinbenton: binding:vnic_type is a separate attribute that is an input to neutron. binding:vif_details is an output from neutron | 23:27 |
| *** alexpilotti has quit IRC | 23:28 | |
| kevinbenton | rkukura: who defines vnic_type? | 23:29 |
| *** armax has joined #openstack-neutron | 23:30 | |
| *** dims_ has joined #openstack-neutron | 23:30 | |
| salv-orlando | kevinbenton: meh don't mind hten | 23:30 |
| rkukura | kevinbenton: Its there for SR-IOV. The user could define it directly to say that their VM expects a certain type of VNIC. | 23:30 |
| rkukura | Or eventually nova could do this, based on the user telling nova what it expects via —nic params | 23:31 |
| kevinbenton | ok, so salv-orlando, are you okay with the current patch for now then? | 23:32 |
| kevinbenton | salv-orlando, or do we want to add another field to portbindings? | 23:32 |
| salv-orlando | I am not really, but I am not the one to judge here. Frankly a bit pissed off that you've done your best to accomodate all plugins and only the NSX plugin is cut out. But again, it's just yet another vendor plugin and therefore I shall not complain. | 23:33 |
| salv-orlando | anyway, there is no other attribute you can use to specify the correct behaviour. So I guess there's nothing that can be done. | 23:33 |
| kevinbenton | salv-orlando: oh, i didn't realize this broke the NSX one still. | 23:33 |
| salv-orlando | it's not broken, but it would add a lot to traffic latency | 23:34 |
| salv-orlando | but I think the patch is already a lot bloated | 23:35 |
| kevinbenton | salv-orlando: ah, NSX is case 3 you mentioned above where it does filtering but without the bridge, right? | 23:35 |
| salv-orlando | it is. | 23:35 |
| salv-orlando | The patch is now probably already beyond what nova core would deem acceptable. | 23:35 |
| kevinbenton | why didn't you just say so? ;-) | 23:35 |
| salv-orlando | In a review you have to respect the review's context. And basically I just limited myself to say that there are plugins for which the patch is not yet satisfactory. But before you dig even further, consider whether this is acceptable for nova. | 23:37 |
| *** armitage81 has joined #openstack-neutron | 23:37 | |
| salv-orlando | because they are very strict about changes they take in RC phase | 23:37 |
| salv-orlando | and this is already at ~50 lines | 23:37 |
| kevinbenton | if we added a different portbindings field, it wouldn't change the bloat of this patch, but it would require a change on the neutron side as well | 23:37 |
| *** yamahata has quit IRC | 23:37 | |
| *** thuc_ has quit IRC | 23:37 | |
| kevinbenton | because this patch would just change to reference a different field | 23:37 |
| salv-orlando | I don't understand why you see changing the neutron side as asomething bad | 23:38 |
| *** mwagner_lap has joined #openstack-neutron | 23:38 | |
| salv-orlando | we have a lot more control than | 23:38 |
| salv-orlando | yet again, changes in the API won't be accepted (ie: you can't add another attribute) | 23:38 |
| *** thuc has joined #openstack-neutron | 23:38 | |
| kevinbenton | i didn't say changing neutron was bad, just will require a little more coordination | 23:39 |
| salv-orlando | so I guess it's endgame, since they idea of having making the hybrid a vif type of its own has been rejected, this is probably the best thing one could come up to. If NSX plugin is broken, it's my team that need to take care of that. | 23:39 |
| *** gdubreui has joined #openstack-neutron | 23:40 | |
| *** Trozz has quit IRC | 23:40 | |
| kevinbenton | would another field in VIF_DETAILS count as an attribute change? | 23:40 |
| claudiub | rkukura: hi | 23:41 |
| *** emagana has quit IRC | 23:41 | |
| claudiub | you talked yesterday with alexpilotti about a more complicated backport for a bug that was fixed within another blueprint | 23:42 |
| *** rkukura has quit IRC | 23:42 | |
| claudiub | could you take a look at the backport commits and say if it's ok? | 23:42 |
| *** thuc has quit IRC | 23:42 | |
| salv-orlando | kevinbenton: I think it would, but you can try since vif details is one of the hideous blobs when in theory you can stash anything you want, in theory you can claim it's not a change in the API | 23:43 |
| salv-orlando | and you would not need the neutron patch as in that case only the NSX plugin and probably some others (thinking of n1kv) might need an update | 23:44 |
| salv-orlando | well, I guess it's goodnight from me. | 23:44 |
| kevinbenton | salv-orlando: can you hang around for 10 mins? | 23:44 |
| salv-orlando | I can | 23:45 |
| mestery | salv-orlando: Any chance you can re-eyeball my critical bug fix here as well? https://review.openstack.org/#/c/82931/ | 23:50 |
| mestery | salv-orlando: I don't think this is a release blocker, per my discussion with markmcclain, but it is quite important to fix I believe for Ubuntu 14.04. | 23:51 |
| salv-orlando | sorry mestery I thought I had approved it | 23:51 |
| mestery | Yes, then amotoki gave it a -1 so I revved it :) | 23:51 |
| mestery | Also, I need one more core on it as well, so if anyone else has a moment, would be great to get this in! | 23:51 |
| mestery | This bug also led me to the fact we have a LOT of duplicate unit test code in places. Probably should clean that up at some point. | 23:52 |
| *** rkukura has joined #openstack-neutron | 23:52 | |
| rkukura | claudiub: yes | 23:52 |
| mestery | thanks salv-orlando | 23:52 |
| claudiub | rkukura: thanks. :) | 23:53 |
| * mestery goes hunting for another core for https://review.openstack.org/#/c/82931/. | 23:53 | |
| salv-orlando | mestery: np. hopefully we'll get another +2 soon | 23:53 |
| claudiub | rkukura: here's the first commit: https://review.openstack.org/#/c/83222/ | 23:53 |
| nati_ueno | hunted | 23:53 |
| claudiub | and here's the second one: https://review.openstack.org/#/c/83223/ | 23:53 |
| claudiub | had to add first the methods used in the second commit | 23:53 |
| HenryG | markmcclain: salv-orlando: just letting you know, https://review.openstack.org/40296 now has a more viable fix for quotas table thanks to amotoki. If you get a few minutes tomorrow please take a look. | 23:54 |
| rkukura | kevinbenton: The binding:vif_details attribute is intended exactly for this flow from neutron to VIF driver. | 23:54 |
| rkukura | kevinbenton: Not sure why using some different attribute is being suggested | 23:54 |
| rkukura | claudiub: My IRC client disconnected breifly and I may have missed part of the conversation right before I said “yes” | 23:55 |
| salv-orlando | sorry rkukura: that was my ignorance. I did not know port binding has input and output attributes | 23:55 |
| claudiub | rkukura: aah, i see, just one sec | 23:55 |
| claudiub | rkukura: you talked yesterday with alexpilotti about a more complicated backport for a bug that was fixed within another blueprint | 23:56 |
| claudiub | rkukura: could you take a look at the backport commits and say if it's ok? | 23:56 |
| rkukura | salv-orlando: binding:host_id, binding:profile, and binding:vnic_type are inputs, binding:vif_type and binding:vif_details are outputs | 23:56 |
| salv-orlando | yes, I must have missed the documentation | 23:58 |
| rkukura | claudiub: Looks reasonable at a glance. I’ll look closer as soon as I can | 23:58 |
| mestery | nati_ueno: Thanks man! Would you prefer I quickly rev for you concerns? I can do that if so. | 23:59 |
| claudiub | rkukura: ok, thanks. :) | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!