14:00:30 #startmeeting networking 14:00:30 Meeting started Tue Oct 15 14:00:30 2024 UTC and is due to finish in 60 minutes. The chair is haleyb. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:30 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:30 The meeting name has been set to 'networking' 14:00:32 Ping list: bcafarel, elvira, frickler, mlavalle, mtomaska, obondarev, slaweq, tobias-urdin, ykarel, lajoskatona, jlibosva, averdagu, amotoki, haleyb, ralonsoh 14:00:39 o/ 14:00:45 \o 14:00:47 o/ 14:00:56 o/ 14:01:55 Liushy proposed openstack/neutron master: Only consider the IPv4 subnets when creating a FIP https://review.opendev.org/c/openstack/neutron/+/932405 14:02:09 o/ 14:02:11 o/ 14:02:11 o/ 14:02:40 ok, we can get started seem to have a mass of people 14:02:42 #topic announcements 14:02:52 Epoxy cycle PTG sessions (October 21-25) 14:02:57 hello 14:02:58 #link https://etherpad.opendev.org/p/oct2024-ptg-neutron 14:03:19 thanks to those who have added topics, please continue to add 14:03:29 late o/ 14:03:52 o/ 14:03:56 i need to add some myself, will create agendas for next week this weekend 14:04:21 Have Tue/Wed/Thu/Fri (if needed) timeslots 14:04:42 just a heads-up 14:04:46 TC meetings monday, and there is an eventlet meeting during the week as well 14:04:57 if we have xmeetings with Nova and the eventlet on wednesday 14:05:04 we should reserve these slots 14:05:50 ralonsoh: i will reach out to nova today, saw a ping from ironic late friday but i was offline for a long weekend 14:05:58 cool 14:06:08 #link https://ptg.opendev.org/ptg.html has the calendar with rooms and times 14:07:04 eventlet-removal is Wednesday at 13:00 and 14:00 14:07:34 for Ironic I suppose one of the topics will be this thread: https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/QS73CVSVRU66MN2MBGZ76WKJFVFM7LPC/ 14:07:51 haleyb: I did want to get some time to have neutron/ironic as well. 14:08:06 Sorry I'm late to the meeting. 14:08:27 the ironic core's asked me to schedule something with you guys on behalf of ironic 14:08:59 cardoe: yes, thanks for attending, can talk right after to pick a time. i've been offline 14:09:15 first should be to add a topic in the Neutron etherpad, I think, to have some context 14:10:15 ralonsoh: it has been a known topic of mine (another dhcp backend), i'm just slow at adding things 14:10:37 no problem, just asking to know the calendar and the topics 14:11:43 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 14:12:15 if no topics will just cancel, and we will not have one PTG week, we can discuss anything during a timeslot 14:13:22 ralonsoh: I added one a few weeks ago. It's not very speced out. 14:14:04 and just to pre-announce, the week after the PTG I am out (company event). will be looking for someone to lead this meeting October 29th 14:14:23 cardoe: i see one item in the agenda, i will add a little to that 14:14:39 I need to engage with you guys better and that's on me. Apologies. 14:14:47 haleyb, I can do it 14:14:59 Oct 29th, added to my calendar 14:15:09 ^^ that week i will be in europe timezone, so around, but just busy 14:15:13 ralonsoh: thanks! 14:15:56 if anyone is near the Hague i'm happy to meet up, but it's far from most i think 14:17:23 last announcement is about antelope 14:17:44 Elod sent an email to the list about Antelop transitioning to unmaintained 14:18:11 #link https://etherpad.opendev.org/p/2023.1-antelope-final-release-before-eom 14:18:55 so we should merge any changes in flight before too long 14:19:15 #link https://etherpad.opendev.org/p/neutron-stable-2023.1-eom 14:20:20 that was all the announcments i had, antyhing else? 14:21:11 #topic bugs 14:21:16 #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/F2RKF3GW52WWZY2CKKO5QXZISRRVVZG4/ 14:21:30 lajoskatona was the deputy last week, and there was just a single bug (?!) 14:21:39 exactly:-) 14:21:50 lucky you :) 14:21:52 and it has a patch 14:22:15 yes, so it is in good hands 14:22:25 https://review.opendev.org/c/openstack/neutron/+/931632 14:23:17 did anyone have other bugs to talk about? anything that needs attention? 14:24:13 ok 14:24:14 I would have one. Not sure if this is the right time or place. The bug is reported here: https://bugs.launchpad.net/neutron/+bug/2084536 14:24:22 and is related to https://bugs.launchpad.net/neutron/+bug/1995078 14:24:56 it has to do with using internal VLAN networks, connected to virtual routers in conjunction with Ironic baremetal (external) ports 14:25:34 I am willing to work on it if you folks think this can be added to neutron 14:26:37 gsamfira, we talked before about this bug. https://review.opendev.org/c/openstack/neutron/+/931892 is not the way to support this feature 14:26:43 you need a new L3 scheduler 14:26:50 something like this: https://review.opendev.org/c/openstack/neutron/+/872033 14:26:59 yes. The patch is a PoC, not intended to be merged as is 14:27:19 could be a POC but the approach is incorrect 14:27:27 but we still need to bind internal router ports to the same HA chassis group as the network it belongs to 14:27:36 ralonsoh: yes. I am aware. 14:27:46 but why the internal interfaces? 14:28:03 ralonsoh: would you consider this an RFE? or just a bug? 14:28:10 you need the external GW of the router to be scheduled in the same GW chassis as the external ports 14:28:11 because we want to be able to connect VLAN networks to geneve networks without going through a provider network 14:28:42 and we may have several VLAN networks connected to the same vrouter and several geneve networks 14:28:56 ok, we can have this conversation offline, not during this meeting 14:29:00 with security groups enforced 14:29:03 sure 14:29:12 thanks! 14:29:46 * cardoe wants to join that too. 14:30:15 ok, moving on 14:30:46 ykarel is the bug deputy this week, although i know he is out a couple of days 14:30:54 or did someone swap? 14:31:08 I did 14:31:08 Did we not change it in the end? I can swap 14:31:41 Oh cool, we should reflect it on the table then 14:32:01 no problem, I'll talk to yatin next week 14:32:28 ralonsoh: i think you just did it, but you can do it again, will have to figure out a week to swap yatin back in 14:32:49 sure, next time I'll ask him 14:32:53 mtomaska is in the schedule for next week 14:33:04 ACK 14:33:11 thanks! 14:33:36 and the bug count stayed stable this week at 730 14:33:57 #topic community-goals 14:34:16 eventlet deprecation is clearly top of mind, and we will discuss next week at PTG 14:34:42 nice! I hope you have time to read the document and add any comment 14:34:46 ralonsoh: are there any patches ready for merge since last week? 14:34:53 oh, did you send it to the ML? 14:35:01 yes and are approved 14:35:15 in any case, the big chunk of work must be defined next week 14:35:34 we'll have time to create tasks and assign them 14:36:25 ah, it was in the "#eventlet-removal - Progress & Update" thread 14:36:32 yes 14:36:36 I added some lines to the etherpad about stadiums, not much new information just to have all in one place 14:36:53 https://etherpad.opendev.org/p/neutron-eventlet-deprecation 14:37:12 #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/YAX2X7C6FCDDNN4SBFN7BNSUA44Z6VXA/ 14:37:18 that's the ML thread if you can't find it 14:37:45 lajoskatona: right, and the etherpad, thanks 14:39:19 ralonsoh: really good summary of the eventlet things in Neutron, thanks for it 14:39:25 yw 14:40:32 lajoskatona: is there an update on horizon work? no pressure, just asking since i have not looked at patches 14:40:45 nothing this week 14:40:56 ack 14:41:14 #topic on-demand 14:41:14 Merged openstack/neutron stable/2024.2: Optimize the SG rule retrieval https://review.opendev.org/c/openstack/neutron/+/932330 14:41:19 Merged openstack/neutron stable/2024.1: Optimize the SG rule retrieval https://review.opendev.org/c/openstack/neutron/+/932332 14:41:29 nothing from me, thanks 14:41:40 i did see a topic from kgube_ 14:41:46 Yes 14:41:59 so, AWS and google are both providing link-local NTP servers to instances, similar to the EC2 metadata IP 14:42:19 i see AWS uses the "common" .123 address 14:42:49 this seems nicer than providing local NTP servers via provider network, as is currebntly required in openstack 14:43:17 the Sovereign Cloud Stack project is currently considering contributing a feature like that, if there is interest 14:44:04 kgube_: i don't know antying about Sovereign, i'm assuming it's openstack-based? 14:44:07 but we are not sure if this is even a neutron topic, or if it could be instead part of the nova metadata service 14:45:04 Yes, it's basically an effort to create a standardized Openstack Setup, that allows easier migration between compliant providers 14:45:14 qq: do you propose that openstack provides this NTP service or just be a proxy? 14:45:29 (same as metadata in Neutron) 14:46:06 I think a proxy would be more realistic 14:46:12 right 14:46:25 so I would start creating a LP bug for this RFE 14:46:31 So maybe the ability to define link-local proxies for different services would be the RFC? 14:46:48 And the metadata service could then be a consumer of that just like NTP? 14:47:14 we do (neutron) much more than being a proxy for metadata 14:47:32 so I would avoid touching the metadata feature 14:47:54 i would agree with ralonsoh regarding an RFE. I could see a provider maybe wanting this, but how it's implemented (metadata entries, dhcp options, etc changes?) and/or config options 14:48:26 exactly, and if you do this during this week, we can discuss it during the PTG 14:48:32 if the PTL agrees on this 14:48:59 yes, it would be a good topic if we have an RFE 14:49:16 would this be a topic for a cross session with nova? 14:49:36 I don't see the link with Nova in this RFE 14:49:48 alright 14:50:41 kgube_: i would only see a link if there is some new metadata value required, but i'm guessing it's more something in the dhcp response pointing at this proxy 14:51:02 but that's what the RFE is for :) 14:51:28 yeah, I think NTP is well supported by DHCP 14:51:43 yup. It's option 004 IIRC 14:52:02 and there is currently no metadata key for it 14:53:03 ahh. It's 042 https://www.greyware.com/software/domaintime/v5/installation/dhcp.asp 14:53:06 trying to be somewhat similar to gcp/aws might be good, i just don't know what they are doing under the hood 14:53:12 Option 042 actually. 004 is deprecated. 14:53:19 yup 14:53:22 Yeah. 14:53:24 feeling old 14:53:58 The reason I suggest making it generic is that technically AWS and GCP provide more than just NTP. I believe they also do DNS. 14:54:26 So allow it to be extensible to add DNS in the future potentially. 14:55:18 we have to start with an RFE, then we can discuss it further 14:55:39 are there any other topics? 14:56:18 ok, thanks for attending everyone, next week is in person at the PTG 14:56:22 #endmeeting