| *** azbiswas has joined #openstack-neutron-ovn | 00:00 | |
| openstackgerrit | Aaron Rosen proposed openstack/networking-ovn: refactor sg/sgr event handing and add code coverage https://review.openstack.org/324956 | 00:08 |
|---|---|---|
| openstackgerrit | Aaron Rosen proposed openstack/networking-ovn: ignore.. testing patch below with ml2 https://review.openstack.org/324957 | 00:11 |
| *** azbiswas has quit IRC | 00:16 | |
| openstackgerrit | Aaron Rosen proposed openstack/networking-ovn: refactor sg/sgr event handing and add test coverage https://review.openstack.org/324956 | 00:17 |
| openstackgerrit | Aaron Rosen proposed openstack/networking-ovn: ignore.. testing patch below with ml2 https://review.openstack.org/324957 | 00:17 |
| *** doonhammer has quit IRC | 00:18 | |
| *** banix has quit IRC | 00:29 | |
| *** jckasper has joined #openstack-neutron-ovn | 00:30 | |
| *** jckasper has quit IRC | 00:35 | |
| *** banix has joined #openstack-neutron-ovn | 00:41 | |
| *** arosen12 has quit IRC | 01:15 | |
| *** stac has left #openstack-neutron-ovn | 01:23 | |
| *** jckasper has joined #openstack-neutron-ovn | 01:40 | |
| *** s3wong has quit IRC | 01:46 | |
| *** gangil has quit IRC | 01:48 | |
| *** gangil has joined #openstack-neutron-ovn | 02:19 | |
| *** yamamoto_ has joined #openstack-neutron-ovn | 02:48 | |
| *** azbiswas has joined #openstack-neutron-ovn | 03:18 | |
| *** banix has quit IRC | 03:21 | |
| *** azbiswas has quit IRC | 03:23 | |
| *** yamamoto_ has quit IRC | 03:45 | |
| *** jckasper has quit IRC | 04:26 | |
| *** yamamoto has joined #openstack-neutron-ovn | 04:30 | |
| *** chandrav has joined #openstack-neutron-ovn | 04:33 | |
| *** ssalagame has joined #openstack-neutron-ovn | 04:41 | |
| *** chandrav has quit IRC | 04:51 | |
| *** gangil has quit IRC | 04:58 | |
| *** ssalagame has quit IRC | 05:00 | |
| *** ssalagame has joined #openstack-neutron-ovn | 05:01 | |
| *** brad_behle has quit IRC | 05:10 | |
| *** karthikp has quit IRC | 05:11 | |
| *** ssalagame has quit IRC | 05:15 | |
| *** gongysh has joined #openstack-neutron-ovn | 05:28 | |
| *** ssalagame has joined #openstack-neutron-ovn | 06:07 | |
| *** ssalagame has quit IRC | 06:16 | |
| *** openstackgerrit has quit IRC | 06:17 | |
| *** openstackgerrit has joined #openstack-neutron-ovn | 06:18 | |
| *** numans has joined #openstack-neutron-ovn | 06:23 | |
| *** jckasper has joined #openstack-neutron-ovn | 06:27 | |
| *** jckasper has quit IRC | 06:31 | |
| *** openstackgerrit has quit IRC | 07:03 | |
| *** openstackgerrit has joined #openstack-neutron-ovn | 07:03 | |
| *** fzdarsky|afk has joined #openstack-neutron-ovn | 07:18 | |
| *** azbiswas has joined #openstack-neutron-ovn | 07:21 | |
| *** azbiswas has quit IRC | 07:25 | |
| *** armax has joined #openstack-neutron-ovn | 07:34 | |
| *** carl_baldwin has quit IRC | 07:58 | |
| *** fzdarsky|afk is now known as fzdarsky | 08:19 | |
| *** fkautz_ has joined #openstack-neutron-ovn | 08:37 | |
| *** fkautz_ has quit IRC | 08:37 | |
| *** fkautz_ has joined #openstack-neutron-ovn | 08:38 | |
| *** switchca1e has joined #openstack-neutron-ovn | 08:38 | |
| *** terryw has quit IRC | 08:39 | |
| *** fkautz has quit IRC | 08:39 | |
| *** switchcade has quit IRC | 08:39 | |
| *** otherwiseguy has joined #openstack-neutron-ovn | 08:39 | |
| *** fkautz_ is now known as fkautz | 08:43 | |
| *** fkautz is now known as Guest99301 | 08:43 | |
| *** armax has quit IRC | 08:45 | |
| *** armax has joined #openstack-neutron-ovn | 09:01 | |
| *** carl_baldwin has joined #openstack-neutron-ovn | 09:11 | |
| -openstackstatus- NOTICE: CI is experiencing issues with test logs, all jobs are currently UNSTABLE as a result. No need to recheck until this is fixed! Thanks for your patience. | 09:37 | |
| *** armax has quit IRC | 09:40 | |
| *** pcaruana has joined #openstack-neutron-ovn | 09:44 | |
| *** carl_baldwin has quit IRC | 10:02 | |
| -openstackstatus- NOTICE: CI is experiencing issues with test logs, all jobs are currently UNSTABLE as a result. No need to recheck until this is fixed! Thanks for your patience. | 10:09 | |
| *** ChanServ changes topic to "CI is experiencing issues with test logs, all jobs are currently UNSTABLE as a result. No need to recheck until this is fixed! Thanks for your patience." | 10:09 | |
| *** gongysh has quit IRC | 10:28 | |
| *** jckasper has joined #openstack-neutron-ovn | 10:28 | |
| *** jckasper has quit IRC | 10:33 | |
| *** yamamoto has quit IRC | 10:35 | |
| *** carl_baldwin has joined #openstack-neutron-ovn | 10:35 | |
| *** banix has joined #openstack-neutron-ovn | 10:41 | |
| *** banix has quit IRC | 10:53 | |
| *** palexster has joined #openstack-neutron-ovn | 10:57 | |
| *** armax has joined #openstack-neutron-ovn | 10:57 | |
| *** rtheis has joined #openstack-neutron-ovn | 11:04 | |
| *** palexster has quit IRC | 11:18 | |
| *** yamamoto has joined #openstack-neutron-ovn | 11:18 | |
| *** palexster has joined #openstack-neutron-ovn | 11:21 | |
| *** azbiswas has joined #openstack-neutron-ovn | 11:23 | |
| *** yamamoto has quit IRC | 11:24 | |
| *** azbiswas has quit IRC | 11:27 | |
| -openstackstatus- NOTICE: CI is experiencing issues with test logs, all jobs are currently UNSTABLE as a result. No need to recheck until this is fixed! Thanks for your patience. | 11:42 | |
| *** pcaruana has quit IRC | 11:45 | |
| *** yamamoto has joined #openstack-neutron-ovn | 11:52 | |
| *** yamamoto_ has joined #openstack-neutron-ovn | 11:52 | |
| *** yamamoto_ has quit IRC | 11:55 | |
| *** yamamoto has quit IRC | 11:56 | |
| *** jckasper has joined #openstack-neutron-ovn | 12:05 | |
| *** banix has joined #openstack-neutron-ovn | 12:09 | |
| *** banix has quit IRC | 12:24 | |
| *** yamamoto has joined #openstack-neutron-ovn | 12:25 | |
| *** yamamoto has quit IRC | 12:34 | |
| openstackgerrit | Richard Theis proposed openstack/networking-ovn: Convert core plugin to ML2 mechanism driver https://review.openstack.org/318095 | 12:40 |
| *** yamamoto has joined #openstack-neutron-ovn | 13:17 | |
| openstackgerrit | Na Zhu proposed openstack/networking-ovn: Sync static routes https://review.openstack.org/322461 | 13:23 |
| *** jckasper has quit IRC | 13:31 | |
| *** jckasper has joined #openstack-neutron-ovn | 13:31 | |
| *** jckasper has quit IRC | 13:38 | |
| *** jckasper has joined #openstack-neutron-ovn | 13:38 | |
| russellb | rtheis: sounds like CI is unstable at the moment, so recheck may not work | 13:48 |
| russellb | openstackstatus set our topic :) | 13:48 |
| rtheis | russellb: thanks, I missed that but should have guessed when I saw unstable in jenkins failure | 13:49 |
| russellb | all good, just letting you know it's not your fault :) | 13:50 |
| russellb | or an ovn bug afaik | 13:50 |
| russellb | hopefully resolved soon, would be cool if we can merge that! | 13:50 |
| rtheis | yes, I hope so too | 13:50 |
| russellb | thanks again for the heroic effort the last weeks | 13:51 |
| *** ChanServ changes topic to "http://docs.openstack.org/developer/networking-ovn/ -=- OVN meeting Thursdays 10:15am Pacific / 1:15pm Eastern #openvswitch -=- Tempest health: http://goo.gl/9MaBJC" | 14:00 | |
| -openstackstatus- NOTICE: Cleanup from earlier block storage disruption on static.openstack.org has been repaired, and any jobs which reported an "UNSTABLE" result or linked to missing logs between 08:00-14:00 UTC can be retriggered by leaving a "recheck" comment. | 14:00 | |
| *** salv-orlando has joined #openstack-neutron-ovn | 14:05 | |
| openstackgerrit | Merged openstack/networking-ovn: refactor sg/sgr event handing and add test coverage https://review.openstack.org/324956 | 14:29 |
| russellb | rtheis: really minor comments on the big ML2 patch, if you file a bug for some followup, i'm happy to merge | 14:30 |
| rtheis | thanks russellb: I'll take a look and see what's going on | 14:30 |
| russellb | k | 14:31 |
| *** ssalagame has joined #openstack-neutron-ovn | 14:33 | |
| *** ssalagame has quit IRC | 14:34 | |
| rtheis | russellb: It appears that any exceptions taken by mechanism driver will result in MechanismDriverError exception being raised. We are indeed taking exceptions while validating the ports in these tests. So I suspect 500 is the result of this. Now need to determine why core plugin gets 400 | 14:36 |
| russellb | ah, so it's a more general ML2-ism | 14:36 |
| russellb | i still think we should have a bug on it, but maybe it's not urgent | 14:36 |
| russellb | we should be able to report a proper failure for invalid input | 14:37 |
| russellb | but this is also binding:profile custom OVN-only stuff | 14:37 |
| russellb | in a perfect world, all API validation could be done at higher layers | 14:37 |
| russellb | this is sort of a hack | 14:37 |
| russellb | so.... | 14:37 |
| russellb | ¯\_(ツ)_/¯ | 14:37 |
| russellb | maybe it's not worth worrying about? | 14:37 |
| russellb | in random other news, ben acked guru's L3 gateway patches! | 14:37 |
| rtheis | russellb: I'll see if there's a better way to report this invalid input | 14:38 |
| rtheis | cool | 14:38 |
| russellb | ok, like i said, don't need to block on it | 14:38 |
| russellb | i just don't want to forget to look closer at it | 14:38 |
| rtheis | I'll open a bug to track it | 14:39 |
| openstackgerrit | Na Zhu proposed openstack/networking-ovn: Sync static routes https://review.openstack.org/322461 | 14:39 |
| russellb | sounds good | 14:39 |
| *** doonhammer has joined #openstack-neutron-ovn | 14:40 | |
| rtheis | russellb: opened https://bugs.launchpad.net/networking-ovn/+bug/1588848 | 14:44 |
| openstack | Launchpad bug 1588848 in networking-ovn "ML2 driver now returns 500 for invalid port bindings" [Undecided,Confirmed] - Assigned to Richard Theis (rtheis) | 14:44 |
| russellb | nice | 14:44 |
| russellb | +2 | 14:45 |
| rtheis | thanks | 14:46 |
| russellb | i think it's pretty amazing that this will be in newton-1! | 14:47 |
| russellb | sounds like newton-1 deadline was actually yesterday, oops | 14:51 |
| russellb | we'll still do it, not sure what the "late" impact is | 14:52 |
| russellb | sorry everyone .... | 14:54 |
| *** ssalagame has joined #openstack-neutron-ovn | 14:55 | |
| *** azbiswas has joined #openstack-neutron-ovn | 15:06 | |
| russellb | i checked with the release team, we won't be doing a newton-1 since we missed the window. we just need to be sure to get newton-2 before the deadline. i apologize for the mistake. | 15:08 |
| russellb | mestery: fyi ^ | 15:08 |
| *** rtheis has quit IRC | 15:29 | |
| *** rtheis has joined #openstack-neutron-ovn | 15:31 | |
| *** numans has quit IRC | 15:40 | |
| *** doonhammer has quit IRC | 15:40 | |
| *** doonhammer has joined #openstack-neutron-ovn | 15:54 | |
| *** azbiswas has quit IRC | 16:09 | |
| *** woodburn has joined #openstack-neutron-ovn | 16:17 | |
| *** gangil has joined #openstack-neutron-ovn | 16:18 | |
| *** woodburn has left #openstack-neutron-ovn | 16:19 | |
| *** ssalagame has quit IRC | 16:23 | |
| *** ssalagame has joined #openstack-neutron-ovn | 16:34 | |
| *** chandrav has joined #openstack-neutron-ovn | 16:41 | |
| *** ssalagame has quit IRC | 16:44 | |
| *** ssalagame has joined #openstack-neutron-ovn | 16:50 | |
| openstackgerrit | Merged openstack/networking-ovn: Convert core plugin to ML2 mechanism driver https://review.openstack.org/318095 | 17:06 |
| russellb | *gasp* | 17:07 |
| Sam-I-Am | heh | 17:07 |
| Sam-I-Am | question is, did these patches include docs with the config changes :) | 17:07 |
| *** arosen12 has joined #openstack-neutron-ovn | 17:09 | |
| rtheis | Sam-I-Am: the final patch did include doc updates with the config changes. If I missed anything, let me know | 17:09 |
| Sam-I-Am | rtheis: which # is the final patch? | 17:10 |
| rtheis | Sam-I-Am: https://review.openstack.org/318095 | 17:10 |
| *** salv-orlando has quit IRC | 17:11 | |
| *** salv-orlando has joined #openstack-neutron-ovn | 17:12 | |
| Sam-I-Am | rtheis: is the old method still available? | 17:14 |
| rtheis | Sam-I-Am: core plugin is gone | 17:16 |
| Sam-I-Am | hmm, which means downstream things ought to break something fierce soon | 17:17 |
| Sam-I-Am | guess there was no way for a transitional period? | 17:17 |
| Sam-I-Am | rtheis: i'm a bit confused by the contents of ml2_conf.ini, re tenant_network_types, firewall_driver, etc. | 17:18 |
| Sam-I-Am | looking at the gate job | 17:19 |
| rtheis | Sam-I-Am: what is your concern? | 17:21 |
| Sam-I-Am | ovn uses geneve for tenant networks. is that option just ignored? | 17:21 |
| rtheis | Sam-I-Am: I'll check | 17:22 |
| Sam-I-Am | and the firewall_driver references something bogus | 17:22 |
| rtheis | Sam-I-Am: tenant_network_types is used as part of the network segment allocation | 17:25 |
| Sam-I-Am | yeah, in the reference arch | 17:28 |
| Sam-I-Am | ovn supports geneve and technically stt | 17:28 |
| *** azbiswas has joined #openstack-neutron-ovn | 17:28 | |
| Sam-I-Am | so how is this working with vxlan? | 17:28 |
| mestery | It shouldn't work with VXLAN, it should be using geneve | 17:30 |
| Sam-I-Am | thats why the config file is confusing me | 17:31 |
| rtheis | Sam-I-Am: good question, it does work but I think that is because the segment network type isn't used by the driver | 17:32 |
| Sam-I-Am | pretty sure ovn's tunnel type is configured in ovs | 17:32 |
| Sam-I-Am | same thing with the firewall driver? | 17:33 |
| rtheis | I think so but haven't checked that one yet | 17:35 |
| mestery | rtheis: ++ I agree with that | 17:35 |
| Sam-I-Am | question is can we leave out these bogus options that do not apply | 17:35 |
| Sam-I-Am | they will create confusion | 17:35 |
| *** ssalagame has quit IRC | 17:36 | |
| rtheis | I'm wondering if we should be using tenant_network_types in the driver | 17:36 |
| rtheis | rather the segment | 17:37 |
| Sam-I-Am | right now we do this | 17:38 |
| Sam-I-Am | ovs-vsctl --no-wait set open_vswitch . external-ids:ovn-encap-type="geneve" | 17:38 |
| *** ssalagame has joined #openstack-neutron-ovn | 17:38 | |
| Sam-I-Am | there's not a lot of config-file-able things in ovn, they're mostly ovs commands | 17:38 |
| Sam-I-Am | it would be nice to wrap those ovs commands in config file things | 17:38 |
| Sam-I-Am | theoretically vxlan is supported for vteps, but that wouldnt be a tenant network | 17:40 |
| Sam-I-Am | its sort of an odd case because i dont think neutron itself supports vxlan provider nets | 17:41 |
| *** s3wong has joined #openstack-neutron-ovn | 17:45 | |
| *** arosen12 has quit IRC | 18:03 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/networking-ovn: Updated from global requirements https://review.openstack.org/325461 | 18:14 |
| *** thumpba has joined #openstack-neutron-ovn | 18:23 | |
| rtheis | Sam-I-Am: I've opened https://bugs.launchpad.net/networking-ovn/+bug/1588935 related to the firewall/security issue that you pointed out | 18:33 |
| openstack | Launchpad bug 1588935 in networking-ovn "ML2 driver does not support enable_security_group configuration" [Undecided,New] - Assigned to Richard Theis (rtheis) | 18:33 |
| Sam-I-Am | thx | 18:34 |
| *** switchca1e is now known as switchcade | 18:35 | |
| *** uday has joined #openstack-neutron-ovn | 18:43 | |
| *** chandrav has quit IRC | 18:47 | |
| *** chandrav has joined #openstack-neutron-ovn | 18:52 | |
| *** gangil has quit IRC | 18:57 | |
| *** doonhammer has quit IRC | 19:02 | |
| *** azbiswas has quit IRC | 19:04 | |
| *** chandrav has quit IRC | 19:04 | |
| rtheis | Sam-I-Am: looks easy enough to fix tenant_network_types so it is set to geneve, but wondering if we can/should use the segmentation ID when creating the resulting ports | 19:04 |
| * russellb catching up on backlog | 19:05 | |
| russellb | the choice of geneve vs something else is opaque to neutron with OVN | 19:05 |
| russellb | neutron has no input | 19:05 |
| russellb | in terms of Neutron, ideally it's either unspecified and we get to do what we want, or if it *is* explicitly specfied, i suppose the best thing is to only accept Geneve and reject anything else | 19:06 |
| rtheis | russellb: so the ovn ml2 driver is safe to ignore what neutron generates for segments | 19:06 |
| russellb | segmentation ID? | 19:06 |
| rtheis | vni | 19:06 |
| russellb | if neutron generates something, yes, we should ignore it, at least for now | 19:06 |
| russellb | there's nowhere to put it, OVN decides that on its own | 19:06 |
| rtheis | ok | 19:07 |
| russellb | we could add the ability to specify it, but that's a little awkward | 19:07 |
| Sam-I-Am | so... | 19:07 |
| *** fzdarsky has quit IRC | 19:07 | |
| russellb | we need to support reading the segmentation_id for a vlan provider network | 19:07 |
| russellb | i assume that is still there? | 19:07 |
| rtheis | russellb: that is | 19:07 |
| Sam-I-Am | russellb: with the conventional ovs agent, tenant network type is somehow passed to ovs | 19:07 |
| russellb | ok | 19:07 |
| Sam-I-Am | russellb: i dont see why ovn is any different | 19:08 |
| russellb | Sam-I-Am: because ovn is different? | 19:08 |
| Sam-I-Am | well, it is, but isn't it just ovs underneath the covers when it comes to tunnels? | 19:08 |
| russellb | neutron is directly configuring ovs in that case | 19:08 |
| russellb | OVN is hiding a lot of that | 19:08 |
| Sam-I-Am | right now we hand-run an ovs-vsctl command on every ovn controller node to set the tunnel type and endpoint ip | 19:08 |
| russellb | who is we | 19:09 |
| russellb | OVN doesn't run ovs-vsctl | 19:10 |
| Sam-I-Am | we = deployment tools | 19:10 |
| Sam-I-Am | ovs-vsctl --no-wait set open_vswitch . external-ids:ovn-encap-type="geneve" | 19:10 |
| Sam-I-Am | that thing | 19:10 |
| russellb | oh | 19:10 |
| Sam-I-Am | and also the ovn-encap-ip | 19:10 |
| russellb | yes | 19:10 |
| russellb | i think it's safe to assume that's always just geneve though | 19:10 |
| russellb | stt is technically in the code, but i don't know if anyone has ever even tried it | 19:10 |
| russellb | and it doesn't really have a future since it's not in the upstream kernel | 19:11 |
| Sam-I-Am | people coming from ML2 are going to expect tenant_network_types to do something | 19:11 |
| Sam-I-Am | well, ml2 + conventional agents | 19:11 |
| russellb | ok | 19:11 |
| rtheis | We can enforce geneve is set | 19:11 |
| russellb | geneve should be fine | 19:11 |
| russellb | i guess vlan should work too | 19:11 |
| russellb | that's just the same thing as vlan provider networks, except letting neutron dynamically pick VLAN IDs | 19:11 |
| Sam-I-Am | vlan shouldnt work... | 19:11 |
| russellb | why not | 19:11 |
| Sam-I-Am | theres no vlan tenant networks in ovn | 19:12 |
| russellb | vlan tenant network and a vlan provider network are the same thing | 19:12 |
| russellb | from ovn's perspective | 19:12 |
| russellb | and in neutron, in practice it's really a matter of how it gets created | 19:12 |
| russellb | but they are mostly the same internally... | 19:12 |
| russellb | afaik... | 19:12 |
| Sam-I-Am | the difference is who creates it | 19:13 |
| * russellb nods | 19:13 | |
| Sam-I-Am | and things like router:external | 19:13 |
| Sam-I-Am | currently, if a tenant creates a network and vlan type is default, it picks a vlan range out of the ml2 config | 19:13 |
| russellb | right | 19:13 |
| Sam-I-Am | and there's some sort of mapping to a bridge | 19:13 |
| russellb | i think that should work, and if it doesn't, it shouldn't be much work to make it work | 19:13 |
| russellb | since it's all the same infrastructure in ovn required to vlan provider nets | 19:14 |
| Sam-I-Am | bridge mappings move from the l2 agent config to ovs-vsctl commands | 19:14 |
| russellb | i think the best thing to do re: conversion period was rip the bandaid off and clean up any fallout as quick as we can | 19:14 |
| russellb | having both would be really painful | 19:14 |
| * russellb nods | 19:15 | |
| russellb | ovn has the same style bridge mappings | 19:15 |
| russellb | network name <--> ovs bridge | 19:15 |
| *** ssalagame has quit IRC | 19:15 | |
| Sam-I-Am | i guess what i'm wondering is do we need some sort of ovn-controller (agent) config file that contains stuff like local_ip, bridge mappings, etc. | 19:15 |
| *** salv-orl_ has joined #openstack-neutron-ovn | 19:15 | |
| russellb | Sam-I-Am: bridge mappings exist, as an external_id right now | 19:16 |
| Sam-I-Am | right now its like yeahhhh its ml2, but most of the config is done by ovs-vsctl | 19:16 |
| Sam-I-Am | yeah | 19:16 |
| Sam-I-Am | but with the ovs agent, they're in openvswitch_agent.ini | 19:16 |
| russellb | i see the ovs-vsctl config and the config file as funcionalliy equivalent | 19:16 |
| russellb | they are both host-local configuration | 19:16 |
| Sam-I-Am | to me its easier to do config file management than command line management | 19:16 |
| Sam-I-Am | i'm thinking about this from the perspective of migrating ovs agent users to ovn | 19:17 |
| russellb | gotcha | 19:17 |
| Sam-I-Am | can we make this look like a more typical neutron config | 19:17 |
| Sam-I-Am | now that we're under ml2 | 19:17 |
| russellb | btw, Babu on my team has been working on a grenade job that tests the migration | 19:17 |
| russellb | that we'll be able to iterate on | 19:17 |
| russellb | we could add config file support to ovn-controller, sure | 19:17 |
| russellb | we talked about it at one point | 19:18 |
| russellb | for the reason you suggest, being a bit easier to deal with | 19:18 |
| *** salv-orlando has quit IRC | 19:18 | |
| russellb | but nobody took it on | 19:18 |
| russellb | i'm supportive of that though if someone wanted to work on it | 19:18 |
| Sam-I-Am | i think its important for getting people on board, documentation, and migrations | 19:19 |
| *** ssalagame has joined #openstack-neutron-ovn | 19:19 | |
| rtheis | I can open a bug to track this config file work if you'd like | 19:20 |
| Sam-I-Am | rtheis: i dont think geneve has id ranges, or at least ones that are configurable in ovn | 19:20 |
| Sam-I-Am | rtheis: yes, pls do | 19:20 |
| russellb | correct, they are not configurable in OVN | 19:20 |
| rtheis | Sam-I-Am: will do | 19:20 |
| russellb | and i'm not particularly interested in letting neutron specify them | 19:21 |
| russellb | i don't see the value | 19:21 |
| rtheis | so we basically ignore what neutron specifies | 19:21 |
| Sam-I-Am | so we probably wouldnt need anything in a [ml2_type_geneve] that i can think of | 19:21 |
| Sam-I-Am | and thats fine | 19:21 |
| russellb | i think arosen had a geneve type driver at one point | 19:21 |
| Sam-I-Am | we solve that issue with docs "you dont need these" | 19:21 |
| russellb | not sure if that merged | 19:21 |
| Sam-I-Am | yeah, its around | 19:21 |
| *** fzdarsky has joined #openstack-neutron-ovn | 19:21 | |
| Sam-I-Am | but in the case of ovn, we'd just doc it as "not configurable for reason X" | 19:22 |
| Sam-I-Am | people will be curious about securitygroup (which rtheis already posted a bug for), ml2_type_vlan, ml2_type_flat, and tenant_network_types | 19:23 |
| russellb | what's the sec group issue? | 19:23 |
| rtheis | russellb: https://bugs.launchpad.net/networking-ovn/+bug/1588935 | 19:23 |
| openstack | Launchpad bug 1588935 in networking-ovn "ML2 driver does not support enable_security_group configuration" [Undecided,New] - Assigned to Richard Theis (rtheis) | 19:23 |
| Sam-I-Am | and from the agent perspective, bridge_mappings and local_ip | 19:23 |
| rtheis | looks like we don't honor security groups being disabled | 19:23 |
| russellb | Sam-I-Am: we have those, it's just ovsdb and not config file, right? | 19:23 |
| russellb | rtheis: just being able to disable them? | 19:24 |
| russellb | ok, seems easy enough | 19:24 |
| russellb | i think we have some code for seeing if port security is disabled? | 19:24 |
| rtheis | I think so, and cleaning up the devstack deployment | 19:24 |
| russellb | so the check would be similar ... | 19:24 |
| Sam-I-Am | russellb: some of them. i'm not aware of tenant network vlan ranges anywhere in ovn. | 19:24 |
| russellb | Sam-I-Am: OVN lets you specify a VLAN ID for a VLAN network | 19:24 |
| russellb | which is all we need from OVN to do it | 19:24 |
| Sam-I-Am | tenants cant specify vlan ids | 19:25 |
| Sam-I-Am | only admins can | 19:25 |
| russellb | from our driver perspective, a vlan ID is given, maybe an admin set it, maybe neutron made it up, right? | 19:25 |
| russellb | however a VLAN ID is decided upon | 19:25 |
| russellb | we have a way to tell OVN what it is | 19:25 |
| russellb | and all is happy | 19:25 |
| *** thumpba_ has joined #openstack-neutron-ovn | 19:26 | |
| rtheis | so we only do that when we have PHYSICAL_NETWORK ... should we do the same for vlan tenant networks? | 19:26 |
| Sam-I-Am | so if we support network_vlan_ranges, ml2 and ovn Just Work ? | 19:27 |
| russellb | conceptually anyway :) ... rtheis, yes, that's the idea | 19:27 |
| rtheis | cool | 19:27 |
| russellb | Sam-I-Am: conceptually we should be able to, sounds like we have a fix to make based on rtheis' comment | 19:27 |
| Sam-I-Am | i was sort of under the impression that all non-provider nets use geneve | 19:27 |
| russellb | but it shouldn't be a ton of work | 19:27 |
| rtheis | I'll open another bug for that | 19:27 |
| Sam-I-Am | but theres an audience that would probably make use of vlan tenant nets | 19:27 |
| russellb | rtheis: thanks | 19:27 |
| *** thumpba has quit IRC | 19:28 | |
| russellb | from ovn perspective, there's really no such thing as tenant network vs provider network | 19:28 |
| russellb | there's logical network, allowing OVN to implement it how it wants (typically geneve) | 19:28 |
| Sam-I-Am | rtheis russellb sorry i didnt think of these things earlier, but time... and i think doing the init conversion as quickly as possible was important | 19:28 |
| russellb | and then specifying that it's a flat network, or a vlan network | 19:28 |
| russellb | Sam-I-Am: no worries, these are just new things we need to work through in ML2 land | 19:28 |
| russellb | not regressions | 19:28 |
| Sam-I-Am | yeah | 19:30 |
| Sam-I-Am | i have an interesting perspective on things | 19:30 |
| Sam-I-Am | sometimes good, sometimes bad. | 19:30 |
| Sam-I-Am | as for the secgroup thing, noop should do what people expect it to do... disable secgroups. otherwise a dummy setting just lets ovn enable its secgroup model | 19:31 |
| Sam-I-Am | what i want to avoid is confusion between the native ovs secgroup driver and the ovn secgroup implementation | 19:31 |
| Sam-I-Am | thats where i fear we'll see confusion | 19:32 |
| russellb | this is a neutron-server side config item? | 19:32 |
| Sam-I-Am | good question :) | 19:33 |
| Sam-I-Am | it has gone through some changes recently | 19:33 |
| russellb | so 2 options ... enable_security_group and firewall_driver | 19:34 |
| rtheis | enable_security_group option under [securitygroup] in ml2_conf.ini. | 19:34 |
| russellb | leaving them to defaults should do what we normally expect | 19:34 |
| russellb | enable_security_group=False should make us disable security groups / not configure ACLs | 19:35 |
| rtheis | sounds good | 19:35 |
| russellb | and really, we should just ignore firewall_driver | 19:35 |
| russellb | it's not relevant at all | 19:35 |
| Sam-I-Am | yeah | 19:35 |
| Sam-I-Am | i think its just enable_security_groups | 19:36 |
| russellb | ok | 19:36 |
| Sam-I-Am | i think firewall_driver defaults to noop, which may be confusing to people | 19:36 |
| russellb | it defaults to empty / not set | 19:36 |
| russellb | noop means something a little different | 19:36 |
| russellb | it's weird | 19:36 |
| rtheis | looks like firewall_driver default is None | 19:36 |
| rtheis | we don't have a driver so that seems appropriate | 19:37 |
| russellb | yeah | 19:37 |
| russellb | we should just totally ignore that config item | 19:37 |
| rtheis | ok | 19:37 |
| Sam-I-Am | ok, thats good | 19:37 |
| Sam-I-Am | we can doc that one out | 19:38 |
| russellb | there's a *lot* of very mixed stuff in neutron between general abstraction and stuff very specific to the ref impl / neutron agents | 19:38 |
| russellb | this is an example of that | 19:39 |
| russellb | enable_security_group == a very generally useful config item, next to firewall_driver == a very ref impl specific thing | 19:39 |
| rtheis | yeah, definitely | 19:39 |
| russellb | ¯\_(ツ)_/¯ | 19:39 |
| Sam-I-Am | ok, i think we're on similar pages now | 19:40 |
| Sam-I-Am | if we put things into config files, converting from ovs agent to ovn should be a lot easier from a config management pov | 19:41 |
| Sam-I-Am | the underlying rebuilding of network bits... glhf :) | 19:41 |
| Sam-I-Am | but at least i can make the docs pretty | 19:41 |
| *** banix has joined #openstack-neutron-ovn | 19:41 | |
| *** ssalagame has quit IRC | 19:41 | |
| russellb | a lot easier from a technical perspective? or just "less different", so less scary to people | 19:42 |
| russellb | we're only talking about 3 or 4 config items | 19:42 |
| *** ssalagame has joined #openstack-neutron-ovn | 19:43 | |
| Sam-I-Am | well, more in line with the expections of configuration with ml2 | 19:43 |
| Sam-I-Am | rather than scary, somewhat obscure ovs-vsctl commands | 19:43 |
| russellb | this is the compute host / agent side | 19:44 |
| russellb | unrelated to ML2 | 19:44 |
| russellb | but i understand the scariness angle | 19:44 |
| russellb | but i also wonder if most people using ovs today are familiar with ovs-vsctl? | 19:45 |
| russellb | is it that scary? | 19:45 |
| russellb | probably to most folks that want to mess with it as litle as possible and just want it to work | 19:45 |
| russellb | and if they ever have to run ovs-vsctl, they're pissed off | 19:45 |
| russellb | because that means something wasn't working and they're debugging | 19:45 |
| Sam-I-Am | yeah, this is the equiv of some of the items in the openvswitch_agent.ini file | 19:47 |
| *** ssalagame has quit IRC | 19:47 | |
| Sam-I-Am | they edit file, run agent... done. | 19:47 |
| Sam-I-Am | the only ovs-vsctl command people expect to run is plugging a host interface into a provider bridge | 19:47 |
| Sam-I-Am | bridge mappings and local tunnel endpoint are in a config file | 19:48 |
| Sam-I-Am | people do fear ovs, so the more we can do to make it look more polished will help adoption | 19:54 |
| Sam-I-Am | a lot of this is from experience in #openstack, docs bugs, etc. | 19:55 |
| -openstackstatus- NOTICE: The infrastructure team is taking Gerrit offline for maintenance this afternoon, beginning shortly after 20:00 UTC. We aim to have it back online around 00:00 UTC. | 19:59 | |
| russellb | cool | 19:59 |
| russellb | i'm good with config file support | 20:00 |
| russellb | Sam-I-Am: funny story, i apparently said almost a year ago that i might work on it | 20:03 |
| russellb | i clearly forgot | 20:03 |
| russellb | http://openvswitch.org/pipermail/dev/2015-July/057871.html | 20:03 |
| Sam-I-Am | russellb: lol | 20:03 |
| Sam-I-Am | guess you just signed up | 20:04 |
| *** banix has quit IRC | 20:04 | |
| Sam-I-Am | re-signed up :) | 20:04 |
| russellb | heh, maybe .... | 20:04 |
| russellb | i need to finish what i've started first | 20:04 |
| russellb | and then maybe! | 20:04 |
| russellb | i'm happy for someone else to ... | 20:04 |
| -openstackstatus- NOTICE: Gerrit is offline for maintenance until 00:00 UTC | 20:10 | |
| *** ChanServ changes topic to "Gerrit is offline for maintenance until 00:00 UTC" | 20:10 | |
| *** gangil has joined #openstack-neutron-ovn | 20:11 | |
| *** ssalagame has joined #openstack-neutron-ovn | 20:12 | |
| *** banix has joined #openstack-neutron-ovn | 20:28 | |
| *** azbiswas has joined #openstack-neutron-ovn | 20:41 | |
| *** uday has quit IRC | 20:41 | |
| *** azbiswas_ has joined #openstack-neutron-ovn | 20:45 | |
| *** irenab has quit IRC | 20:47 | |
| rtheis | russellb, Sam-I-Am: https://bugs.launchpad.net/networking-ovn/+bug/1588966 and https://bugs.launchpad.net/networking-ovn/+bug/1588969, I attempted to capture our discussion in these 2 bugs. Feel free to update as needed. | 20:48 |
| openstack | Launchpad bug 1588966 in networking-ovn "ML2 driver should only support geneve and vlan tenant network types" [Undecided,New] - Assigned to Richard Theis (rtheis) | 20:48 |
| openstack | Launchpad bug 1588969 in networking-ovn "Support ovn-controller configuration file" [Undecided,New] | 20:48 |
| *** azbiswas has quit IRC | 20:48 | |
| *** chandrav has joined #openstack-neutron-ovn | 21:00 | |
| *** salv-orl_ has quit IRC | 21:06 | |
| *** salv-orlando has joined #openstack-neutron-ovn | 21:07 | |
| *** doonhammer has joined #openstack-neutron-ovn | 21:14 | |
| *** rtheis has quit IRC | 21:15 | |
| *** a_ta has quit IRC | 21:27 | |
| *** a_ta has joined #openstack-neutron-ovn | 21:27 | |
| *** doonhammer has quit IRC | 21:31 | |
| *** a_ta has quit IRC | 21:32 | |
| *** banix has quit IRC | 21:38 | |
| *** ssalagame has quit IRC | 21:40 | |
| *** ssalagame has joined #openstack-neutron-ovn | 21:42 | |
| *** doonhammer has joined #openstack-neutron-ovn | 22:05 | |
| *** azbiswas_ has quit IRC | 22:14 | |
| *** thumpba_ has quit IRC | 22:27 | |
| *** ssalagame has quit IRC | 23:01 | |
| *** gangil has quit IRC | 23:15 | |
| *** jckasper has quit IRC | 23:26 | |
| *** jckasper has joined #openstack-neutron-ovn | 23:27 | |
| *** thumpba has joined #openstack-neutron-ovn | 23:28 | |
| *** jckasper has quit IRC | 23:31 | |
| *** thumpba has quit IRC | 23:33 | |
| *** gangil has joined #openstack-neutron-ovn | 23:38 | |
| *** gangil has quit IRC | 23:38 | |
| *** gangil has joined #openstack-neutron-ovn | 23:38 | |
| *** gangil has quit IRC | 23:40 | |
| *** carl_baldwin has quit IRC | 23:54 | |
| *** banix has joined #openstack-neutron-ovn | 23:55 | |
| *** carl_baldwin has joined #openstack-neutron-ovn | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!