14:00:09 <haleyb> #startmeeting networking 14:00:09 <opendevmeet> Meeting started Tue Nov 28 14:00:09 2023 UTC and is due to finish in 60 minutes. The chair is haleyb. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:09 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:09 <opendevmeet> The meeting name has been set to 'networking' 14:00:12 <elvira> o/ 14:00:33 <obondarev> o/ 14:00:40 <mlavalle> \o 14:01:06 <ykarel> o/ 14:01:10 <rubasov> o/ 14:01:34 <haleyb> alright lets get started 14:01:38 <haleyb> #topic announcements 14:02:19 <haleyb> according to https://releases.openstack.org/caracal/schedule.html we are past the C-1 milestone 14:02:21 <lajoskatona> o/ 14:02:55 <haleyb> C-2 milestone is Jan 8-12 which will come up quickly 14:03:41 <haleyb> Reminder: If you have a topic for the drivers meeting on Friday, please add it to the wiki @ https://wiki.openstack.org/wiki/Meetings/NeutronDrivers 14:03:51 <haleyb> there are currently no topics in the queue 14:04:33 <haleyb> Review days - talked last week about setting aside 2-3 days at intervals to get some in-flight reviews merged 14:04:38 <haleyb> #link https://doodle.com/meeting/organize/id/dN8opRDa 14:04:59 <haleyb> 5 people added their availability, with December 1st and 8th working for everyone, are people OK splitting across a week? 14:05:34 <haleyb> or if back-to-back is better i would propose December 7/8 14:05:59 * haleyb taps microphone to see if people are awake :) 14:06:08 <lajoskatona> yes sir 14:06:14 <frickler> I'm not directly involved, but splitting to allow authors to react sounds nice 14:06:15 <lajoskatona> for me either can work 14:06:18 <mlavalle> the split is fine 14:06:28 <lajoskatona> good reasoning, than split it 14:07:52 <haleyb> alright, i was thinking the same thing - so anyone that added their name, or anyone that needs reviews, try and set aside time this Friday and next for them 14:08:00 <haleyb> time in-between to fix-up and rebase, etc 14:08:16 <haleyb> #action haleyb to reply to ML thread with this info 14:08:52 <frickler> fnordahl: ^^ that might be interesting for your series, too 14:09:34 <haleyb> any other announcements? 14:10:10 <haleyb> #topic bugs 14:10:35 <haleyb> obondarev was deputy last week, isabek is this week, elvira next week 14:10:44 <haleyb> #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/W4JIPTXHVIN54T2FOPEYO6AGREF4JPCC/ 14:11:34 <haleyb> 4/5 bugs were picked-up and have patches 14:11:52 <isabek> o/ 14:11:55 <haleyb> so thanks for picking them up 14:12:09 <haleyb> isabek: are you ok being bug deputy this week? 14:12:22 <isabek> sure 14:12:33 <haleyb> ack, thanks! 14:12:50 <elvira> there are some PH on Spain next week, I will be out from wednesday until end of week :( 14:13:09 <bcafarel> late o/ 14:13:36 <elvira> I don't mind taking a look mon-tue and then on Mondaywhen I get back. Whatever is best for you folks 14:14:25 <haleyb> elvira: ack, i can watch for critical things the end of the week, unless someone wants to switch? 14:15:34 <lajoskatona> I can switch 14:15:46 <lajoskatona> my week would be the first of 24, and I am off that week 14:16:03 <haleyb> elvira: does that work? i can update if so 14:16:39 <elvira> I will also be off but I will have the laptop really nearby so no problem for me!! 14:16:42 <elvira> It works better 14:16:44 <elvira> :) 14:16:56 <lajoskatona> thanks, than I will be deputy next week 14:17:03 <haleyb> sold, will update 14:17:18 <haleyb> so the one bug in question is 14:17:23 <haleyb> #link https://bugs.launchpad.net/neutron/+bug/2044215 14:17:44 <haleyb> dns: Keyword is replaced by project_name instead of project_id 14:18:00 <haleyb> i don't exactly understand it either, submitter asked for more information 14:18:38 <haleyb> if it makes sense to anyone please feel free to add comments 14:19:20 <haleyb> are there any other bugs someone wants to discuss? 14:19:29 <gtema> https://bugs.launchpad.net/neutron/+bug/1975828 14:19:49 <gtema> sorry for jumping in: please check the last comment. Sadly bug expired, but I posted update recently 14:20:23 <gtema> there is huge difference in API when listing ports (from project with 1500 ports) between admin and user call 14:20:37 <gtema> and I ended up in thinking that it is policy evaluation engine 14:20:59 <haleyb> gtema: thanks for bringing this up 14:21:16 <haleyb> so when you say policy i'm assuming rbac? 14:21:25 <gtema> yes 14:22:38 <haleyb> we can un-expire it, is there anyone that has cycles to look into it? 14:23:00 <haleyb> rodolfo is out for a while 14:23:24 <opendevreview> Luis Tomas Bolivar proposed openstack/ovn-bgp-agent master: Avoid race when deleting VM with FIP https://review.opendev.org/c/openstack/ovn-bgp-agent/+/901963 14:23:33 <fnordahl> frickler: (indeed, I will read up on the e-mail thread, and I'm available until Dec 15th so hopefully we can get some ground covered before then!) 14:23:54 <opendevreview> Miguel Lavalle proposed openstack/neutron master: Router flavors and service type for OVN https://review.opendev.org/c/openstack/neutron/+/883988 14:24:03 <opendevreview> Merged openstack/ovn-bgp-agent master: Ensure withdrawn events are only processed in relevant nodes https://review.opendev.org/c/openstack/ovn-bgp-agent/+/901962 14:24:43 <haleyb> gtema: i will change the status just not sure if anyone can pick it up 14:25:31 <opendevreview> Luis Tomas Bolivar proposed openstack/ovn-bgp-agent stable/2023.2: Ensure withdrawn events are only processed in relevant nodes https://review.opendev.org/c/openstack/ovn-bgp-agent/+/902065 14:25:43 <gtema> ok. I will try myself to play deeper with it. It looks like there is no caching when policies are being evaluated. Just wanted to know whether somebody knows something about it 14:26:29 <haleyb> sorry i don't, first time i had seen it 14:26:56 <haleyb> last thing on topic of bugs 14:27:02 <gtema> ok, thks anyway 14:27:36 <haleyb> Current bug count this week: 762, up 2 from last week - so good we really didn't increase 14:28:12 <haleyb> #topic specs 14:28:22 <haleyb> #link https://review.opendev.org/q/project:openstack%252Fneutron-specs+status:open 14:29:10 <haleyb> i see racosta has uploaded a new version of the OVN-IC spec, i have not had time to review yet 14:29:16 <haleyb> #link https://review.opendev.org/c/openstack/neutron-specs/+/891204 14:29:22 <haleyb> will review today 14:30:18 <haleyb> and the other two (BGP related) also had updates 14:30:51 <haleyb> hopefully people will get around to reviews :) 14:31:51 <haleyb> moving on... 14:31:55 <haleyb> #topic community_goals 14:33:00 <haleyb> lajoskatona: think you're the only one with a goal on the list, neutronclient deprecation 14:33:08 <haleyb> #link https://review.opendev.org/c/openstack/horizon/+/891205 14:33:19 <lajoskatona> yes, I am strugling with the one horizon patch 14:33:32 <lajoskatona> to fix failing tests 14:33:41 <lajoskatona> so slow progression 14:34:25 <haleyb> thanks for taking on the work to begin with 14:34:52 <haleyb> #topic on_demand 14:35:21 <lajoskatona> I added a topic for OpenAPI 14:35:32 <lajoskatona> the mail from gtema: https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/BU5JH7FJ5VPS75QWPW47IYSZOFDI5R2W/ 14:35:54 <gtema> right, thanks 14:36:19 <gtema> I can give some more infos on the topic, but few more links to be more preciese 14:36:26 <haleyb> "Neutron is very special" 14:36:34 <lajoskatona> :-) 14:36:43 <haleyb> gtema: sure, go ahead 14:37:09 <gtema> https://github.com/gtema/openstack-openapi/blob/main/specs/network/v2.yaml - this is a spec that I was able to generate from neutron sources 14:37:30 <gtema> https://github.com/gtema/openstack-codegenerator/blob/main/codegenerator/generator/openapi/neutron.py - this is the code I generated it with 14:37:51 <gtema> https://gtema.github.io/openstack-openapi/network.html - current draft for rendering spec in the OpenStack style 14:37:57 <gtema> everything is a work in progress 14:38:30 <gtema> but generally the question nr 1) is Neutron team interested in having openapi doc produced from sources so that api-ref can be actually deprecated as such 14:38:37 <lajoskatona> is this for the API or for what we have in SDK? 14:38:46 <gtema> and also all clients/sdks/etc can be generated by various tools 14:39:04 <lajoskatona> ok so this is for the APi which we have in n-lib 14:39:13 <gtema> well, this is currently not connected with SDK directly, but rather extracted from Neutron source and api-ref html 14:39:18 <haleyb> #link https://docs.openstack.org/api-ref/network/v2/index.html 14:39:33 <haleyb> should be similar to that i would guess 14:40:05 <gtema> right, this is going to be rendered this style, but maybe more modern. This is not a point 14:40:25 <slaweq> what about APIs provided by e.g. stadium projects? I don't see them in this example at all 14:40:50 <gtema> right, at the moment I took pure "neutron" and messed with what I could find 14:41:14 <gtema> vpnaas is also not there and that was a question towards you 14:41:49 <gtema> https://github.com/gtema/openstack-codegenerator/blob/main/codegenerator/generator/openapi/neutron.py#L79 - this is how I setup the neutron to access api router and inspect everything what it knows about 14:41:50 <slaweq> even for just neutron, APIs of running server may be different depending on enabled service plugins, mechanism drivers, etc. (as You already mentioned in the email) so will it be possible to somehow generate complete list of APIs always? 14:42:19 <gtema> that is the point - I need help in getting the configuration that enables "everything" 14:43:32 <lajoskatona> I dont think that is possible at least for all the stadiums in one setup 14:43:34 <gtema> second question would be: is it possible to alter exposed routes to not to expose things which are not working 14:44:09 <gtema> i.e. you see in the spec there is POST for /availability-zones or /auto-allocateted-topology 14:44:32 <gtema> this is because in reality those are exposed but if you try invoking them you will get weird validation errors, but not 404 14:46:18 <gtema> also, I found way to "convert" api_definition into the json schema, but there are certain limitations there and not all resources are having it at all, so that requires hardcoding 14:48:12 <gtema> lajoskatona - if it is not possible to have everything running at once (and this is why I stopped attempting that) we could try different approach of generating spec in few runs with different configs 14:48:19 <haleyb> i'm not sure i'm the best one to help, but my very generic question is there something neutron can change to make this work easier? 14:48:30 <lajoskatona> gtema: ack 14:48:38 <opendevreview> Merged openstack/ovn-bgp-agent master: Avoid race when deleting VM with FIP https://review.opendev.org/c/openstack/ovn-bgp-agent/+/901963 14:49:16 <gtema> haleyb: yes, there are certain things, but first I would like to know whether I have Neutron support generally in this initiative 14:50:30 <opendevreview> Luis Tomas Bolivar proposed openstack/ovn-bgp-agent stable/2023.2: Avoid race when deleting VM with FIP https://review.opendev.org/c/openstack/ovn-bgp-agent/+/902067 14:51:19 <haleyb> gtema: i would support it, what do others think? and where do you think it falls on the scale of small/medium/large so we don't overcommit? 14:51:55 <slaweq> generally I like that idea and I can try to help if I will have some cycles (but that may be hard currently) 14:52:16 <slaweq> so gtema if You will need any help, please ping me - I will try to help when possible 14:52:23 <lajoskatona> +1 from me, I can help also for the stadiums for example 14:52:28 <gtema> perfect. Thanks a lot. 14:52:57 <gtema> so from the gestimate: I think initially it is a small effort just to analyse and find ways for workarounds 14:53:31 <gtema> once that is done in next steps those workarounds may be moved to neutron code so that generation is more reliable, that may be medium 14:53:47 <gtema> but I really don't expect it to be a large stuff 14:54:47 <gtema> ok, thanks on the topic 14:55:01 <lajoskatona> thanks for working on this topic 14:55:02 <haleyb> gtema: ack, thanks, sounds good to me 14:56:11 <haleyb> any other topics for discussion? 14:56:15 <mlavalle> Just remind the team that router flavors with L3 OVN is ready for reviews: https://review.opendev.org/c/openstack/neutron/+/883988 14:56:36 <haleyb> mlavalle: thanks! 14:56:40 <mlavalle> I just rebased it 14:56:49 <haleyb> ykarel: and we have CI meeting today? 14:57:00 <ykarel> yes in 4 minutes over IRC 14:57:22 <lajoskatona> ack 14:57:41 <haleyb> ok, thanks for attending everyone 14:57:46 <haleyb> #endmeeting