14:00:38 <haleyb> #startmeeting networking 14:00:38 <opendevmeet> Meeting started Tue Feb 11 14:00:38 2025 UTC and is due to finish in 60 minutes. The chair is haleyb. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:38 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:38 <opendevmeet> The meeting name has been set to 'networking' 14:00:40 <haleyb> Ping list: bcafarel, elvira, frickler, mlavalle, mtomaska, obondarev, slaweq, tobias-urdin, ykarel, lajoskatona, jlibosva, averdagu, haleyb, ralonsoh 14:00:43 <mlavalle> \o 14:00:46 <obondarev> o/ 14:00:52 <ralonsoh_> hello 14:00:53 <ykarel> o/ 14:00:56 <mblue_> o/ 14:01:14 <bcafarel> o/ 14:01:23 <sahid> o/ 14:01:26 <elvira> o/ 14:01:36 <haleyb> #announcements 14:01:37 <opendevreview> Rodolfo Alonso proposed openstack/neutron master: [L3] Do not track the fixed IP / FIPs in the keepalived configuration https://review.opendev.org/c/openstack/neutron/+/941247 14:01:42 <haleyb> We are in week R-7 of Epoxy 14:01:49 <frickler> o/ 14:01:52 <haleyb> E-3 milestone at week R-5 (Feb 24) 14:02:00 <lajoskatona> o/ 14:02:05 <haleyb> #link https://releases.openstack.org/epoxy/schedule.html 14:02:16 <haleyb> So about 2 weeks until feature freeze 14:03:05 <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 14:03:14 <haleyb> And of course a reminder to use the priorities dashboard for patches in the "ready to merge" state 14:03:27 <haleyb> Please put reviews there or ping people here 14:04:40 <haleyb> We previously merged change to remove linuxbridge driver, i've been trying to merge the dibbler removal patch this week but so far the gate has said 'no', will try again after meeting 14:05:23 <haleyb> was there any progress fixing things that broke due to linuxbridge? 14:05:55 <haleyb> i also proposed n-t-p and devstack changes yesterday 14:06:11 <lajoskatona> som with bagpipe 14:06:14 <haleyb> #link https://review.opendev.org/q/topic:%22remove-linuxbridge-agent%22 14:06:35 <lajoskatona> https://review.opendev.org/c/openstack/networking-bagpipe/+/940765 14:06:57 <lajoskatona> I will change the topic to have all under the same 14:06:59 <haleyb> lajoskatona: thanks, i see that is needed for mine 14:07:33 <lajoskatona> I pushed another one to make the job non-voting temporary 14:07:57 <haleyb> +1 14:08:01 <lajoskatona> https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/941218 14:08:56 <haleyb> last announcement is it's PTL/TC election season for 2025.2 cycle, please read the ML email for further information 14:09:58 <haleyb> i have no other announcements, anyone else? 14:10:22 <lajoskatona> today will be a meeting with TC and foundation I suppose 14:10:35 <lajoskatona> https://lists.openinfra.org/archives/list/foundation@lists.openinfra.org/thread/3B7OWPRXB4KD2DVX7SYYSHYYRNCKVV46/ 14:10:40 <lajoskatona> https://board.openinfra.org/strategic-consideration 14:10:50 <lajoskatona> about merging to/with Linuxfoundation 14:11:12 <lajoskatona> there will be another one this week for other timezones, today is 1600UTC 14:11:21 <haleyb> oh, yes, right after this meeting, so i will make sure to end on-time 14:12:19 <haleyb> hopefully everyone saw that email and responses 14:12:47 <haleyb> moving on 14:12:51 <haleyb> #topic bugs 14:13:01 <haleyb> bcafarel was the deputy last week, his report is at 14:13:08 <haleyb> #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/ST54KVP3RCC5XOZSWBHJ65M3Y3EZFME4/ 14:13:44 <haleyb> the two critical have patches proposed/merged, thanks for that 14:14:00 <haleyb> as do the two high ones 14:14:39 <haleyb> and 2/3 of medium have patches as well 14:15:17 <cardoe> I was hoping I could get a +1 for https://review.opendev.org/c/openstack/python-openstackclient/+/940595 which is a fix for routers. 14:15:28 <haleyb> and ralonsoh picked-up the eventlet-removal one, thanks 14:15:34 <haleyb> #link https://bugs.launchpad.net/neutron/+bug/2097257 14:15:40 <haleyb> [eventlet-removal] Use ``ThreadPoolExecutor`` for ``periodics.PeriodicWorker`` 14:15:51 <ralonsoh> ^ proposed by otherwiseguy 14:16:27 <lajoskatona> cardoe: will check it, thanks 14:16:34 <otherwiseguy> ralonsoh: on that one, I don't know if we need it or not. things seemed to work without it after not monkey patching 14:16:35 <haleyb> ack, it didn't have a patch linked in it but i do remember one now that you mention it 14:17:17 <ralonsoh> otherwiseguy, we'll see that in the CI, checking the periodicity of the method call 14:17:20 <ralonsoh> as you did before 14:18:06 <otherwiseguy> not sure why the patch changed the frequency of periodic calls while eventlet was still monkeypatching--but only when the api was active. was weird. 14:18:12 <otherwiseguy> probably a red herring 14:19:19 <haleyb> ok, i will let you two decide on the fate of any change there after testing 14:20:10 <haleyb> last one was on an older release 14:20:15 <haleyb> #link https://bugs.launchpad.net/neutron/+bug/2097610 14:20:54 <haleyb> i will add a note about reproducing on somethine 2023.2 or later 14:22:04 <haleyb> any other bugs to discuss? 14:22:40 <haleyb> deputy this week is elvira, next week is slaweq - that good with both of you? 14:22:48 <elvira> sure! 14:22:49 <ralonsoh> slaweq, is on PTO 14:22:56 <ralonsoh> I'll ping him next day 14:23:28 <haleyb> ack, thanks 14:24:00 <haleyb> #topic community goals 14:24:23 <haleyb> lajoskatona: neutronclient update update 14:24:29 <haleyb> #link https://review.opendev.org/q/topic:%22bug/1999774%22+status:open 14:24:55 <lajoskatona> no news this week 14:25:27 <lajoskatona> the patches from last week under review, but I had no time to work on this 14:25:55 <haleyb> ack, they all at least seem to be green and have reviews 14:26:18 <lajoskatona> yes quite good attention from Horizon team 14:26:49 <haleyb> and last is eventlet 14:26:58 <ralonsoh> yes I have a list of patches 14:26:59 <ralonsoh> https://review.opendev.org/c/x/whitebox-neutron-tempest-plugin/+/941224 14:26:59 <ralonsoh> https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/941212 14:26:59 <ralonsoh> https://review.opendev.org/c/openstack/neutron/+/941211 14:26:59 <ralonsoh> https://review.opendev.org/c/openstack/neutron/+/941202 14:26:59 <ralonsoh> https://review.opendev.org/c/openstack/neutron/+/940707 14:27:13 <ralonsoh> that should be enough to remove eventlet from the API 14:27:15 <ralonsoh> NOTE: 14:27:50 <lajoskatona> sahid also works on ovs-agent patches: https://review.opendev.org/q/topic:%22bug/2087939%22 14:28:02 <ralonsoh> I'm removing logger from the ML2/OVN jobs because (1) this is calling monkey_patch(), (2) this is a testing mech driver class, (3) this is not used in the jobs and (4) no user will call it in a production env 14:28:03 <lajoskatona> I try to follow these and help where I can 14:28:19 <ralonsoh> yes, OVS patches are the second part 14:28:40 <ralonsoh> but is there a CI job running without eventlet? 14:29:08 <sahid> ralonsoh: yes 14:29:12 <sahid> the last patch 14:29:20 <ralonsoh> https://review.opendev.org/c/openstack/neutron/+/937765 14:29:30 <sahid> https://review.opendev.org/c/openstack/neutron/+/937765 14:29:36 <sahid> yes 14:29:39 <ralonsoh> the CI is not passing 14:29:45 <sahid> still working on it 14:29:51 <sahid> all the previous are passing 14:29:57 <ralonsoh> ok, maybe we can merge the os-ken ones 14:30:02 <ralonsoh> and release a new version 14:30:22 <sahid> i think that will help to make progress 14:30:49 <ralonsoh> ok, let's focus on them. If there are further errors, we can release minor versions 14:31:01 <haleyb> ok, so start at this one 14:31:05 <haleyb> #link https://review.opendev.org/c/openstack/os-ken/+/938326 14:31:10 <lajoskatona> +1 14:31:11 <sahid> thank you 14:31:15 <haleyb> there is a relation change there 14:31:52 <haleyb> sahid: when those are done can you propose os-ken release? or ping me if you need any help 14:32:21 <sahid> sure I will 14:32:28 <sahid> and if needed will ping you aswell 14:32:33 <sahid> thank you 14:33:06 <haleyb> ack, thanks for all your work (and rodolfo and everyone else) 14:33:47 <haleyb> #topic on-demand 14:33:51 <haleyb> any other topics? 14:34:07 <haleyb> nothing on the wiki 14:34:21 <lajoskatona> nothing from me 14:35:21 <haleyb> ok, if nothing else have a good week, lets get some of these eventlet changes merged :) 14:35:21 <mlavalle> not from me either 14:35:26 <MengyangZhang[m]> I just a quick question, would the neutron community be interested in supporting applying QoS changes live to VMs 14:35:42 <ralonsoh> this is already supported 14:35:59 <ralonsoh> you can change the QoS settings of a port that is attached 14:36:42 <cardoe> I did have some questions. 14:36:45 <MengyangZhang[m]> okk, is there a doc for this, want to learn more about it 14:36:55 <cardoe> I'm an operator looking to use Neutron for managing our bare metal infra. 14:37:00 <cardoe> And specifically networking. 14:37:02 <ralonsoh> MengyangZhang[m], the Neutron API, let me find that 14:37:26 <ralonsoh> MengyangZhang[m], https://docs.openstack.org/api-ref/network/v2/index.html#quality-of-service 14:37:33 <cardoe> You guys have https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/5SX35AQZ65G6HWQ7WARLJTJK3ASAOWNU/ on your wiki which is what I'm looking to follow up on. 14:39:19 <cardoe> I'm specifically using L2VNI. There's been some out of tree tweaks around that by Juniper and others. There's also https://bugs.launchpad.net/ovn-bgp-agent/+bug/2017890 14:39:47 <cardoe> I'm just wondering how to best engage on my use case. I'd like to contribute it upstream as best as possible. 14:41:00 <ralonsoh> To be honest, I think Julia has a better perspective of this topic. I think you should first sync with Julia about what is the Ironic's plan about "decoupling" certain areas from Neutron 14:41:27 <haleyb> yes, i would agree was just trying to read that and look at wiki 14:41:46 <ralonsoh> the ovn-bgp-agent is for OVN only, I don't think that suits your requirements 14:42:07 <cardoe> I've been speaking with Julia about it. 14:42:26 <cardoe> I'm encouraging less of a decoupling and more of a lean in harder. :D 14:43:54 <haleyb> are there specific neutron changes for this effort? 14:44:15 <cardoe> Well just usage behaviors and maybe I'm understanding the features wrong. 14:44:40 <cardoe> So I've got a VXLAN fabric with leafs and edges. But some of my servers can sit on two separate fabrics. 14:45:13 <cardoe> Previously when we used VLAN, I could use physical_network as the fabric / separation. But VXLAN explicitly disallows physical_network. 14:47:02 <cardoe> With L2VNI, we'll have a VLAN mapped to that VXLAN on that switch pair serving the physical server for example as well. It would seem like the segments feature would work to create VLAN segment on a VXLAN network. But things feel more L3VNI. I'm not sure if it'll be a blocker. 14:47:41 <haleyb> so if there is a change required in neutron, it needs to be proposed via our RFE process, I have not seen any request in this area yet 14:47:48 <cardoe> I'm interested in using OVN with this as well and have it either join the fabric with something like... https://github.com/toreanderson/evpn_agent/ 14:48:11 <cardoe> I'll happily write them up. I just wanted to first ask if this use case was something that would be acceptable. 14:48:18 <haleyb> https://wiki.openstack.org/wiki/Meetings/NeutronDrivers 14:49:55 <haleyb> i am just not super familiar with this effort 14:50:41 <cardoe> In some cases there would just be neutron and nothing else. e.g. the BMC / iLO / iDRAC (whatever ya wanna call it) network. 14:51:00 <cardoe> I'll do a better job of writing this up and explaining it in the future. 14:51:20 <cardoe> I just was more curious around the acceptability of neutron as a data center network manager. 14:52:12 <haleyb> hmm, i guess i would not try and stretch neutron that far 14:52:40 <haleyb> but would have to see a proposal 14:53:30 <cardoe> okay. So basically Ironic runs its own DHCP server for certain cases still. I've been asking why not just use a neutron network with neutron supplied DHCP for the BMC network. 14:53:44 <cardoe> As one example. 14:54:05 <cardoe> I'll refine it better before I bring it up again. Sorry. 14:54:26 <cardoe> TheJulia: btw ^ 14:55:02 <haleyb> we only have a few minutes left, want to give people time if they wish to attend TC meeting 14:55:50 <frickler> 1600 UTC is 1h away fwiw 14:56:12 <haleyb> oh, i saw 10am ET 14:56:18 <haleyb> i have a hard stop either way 14:57:18 <haleyb> cardoe: thanks for bringing it up, please loop-back once you have more info from ironic team 14:57:43 <spatel> Folks, I have question related OVN chassis priority. I want to change chassis priority so my external port for SRIOV get schedule on specified chassis - https://paste.opendev.org/show/by6LCZQ8jSCAwlDeWv4K/ 14:58:05 <ralonsoh> spatel, we can talk after the meeting 14:58:07 <spatel> I have vlan tenant based networking I am not running any L3 vRouter 14:58:10 <haleyb> #endmeeting