| opendevreview | Helen Chen proposed openstack/neutron master: Add upstream EVPN multinode Zuul job https://review.opendev.org/c/openstack/neutron/+/994257 | 01:59 |
|---|---|---|
| lajoskatona | frickler, elodilles: Hi, I pushed -eol tags for x/networking-l2gw from ocata to 2023.2 , so I think it is now safe to delete those branches | 07:03 |
| frickler | lajoskatona: thanks, I'll take care of that, might not be today though | 07:20 |
| lajoskatona | frickler: thanks, if any questions or things in I can help appear just ping me | 07:28 |
| opendevreview | Lajos Katona proposed openstack/python-neutronclient master: Remove Vpnaas client code https://review.opendev.org/c/openstack/python-neutronclient/+/991550 | 07:34 |
| ralonsoh | hi folks, a quick fix for py314 UTs: https://review.opendev.org/c/openstack/neutron/+/994918 | 07:42 |
| ralonsoh | (this is mostly happening in py314 but also in py313) | 07:42 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: ovn: Fix HA_Chassis_Group race condition in port create/update https://review.opendev.org/c/openstack/neutron/+/995009 | 08:01 |
| opendevreview | alisafari proposed openstack/neutron stable/2026.1: bgp: Limit lrp_mac to the chassis https://review.opendev.org/c/openstack/neutron/+/995010 | 08:28 |
| opendevreview | Merged openstack/neutron master: tests: Fix random failure in ``test_router_update_with_too_many_routes`` https://review.opendev.org/c/openstack/neutron/+/994918 | 08:46 |
| opendevreview | Eduardo Olivares proposed openstack/neutron master: Add upstream EVPN multinode Zuul job https://review.opendev.org/c/openstack/neutron/+/994257 | 08:47 |
| opendevreview | Merged openstack/os-ken master: pre-commit: Bump versions https://review.opendev.org/c/openstack/os-ken/+/994681 | 08:55 |
| opendevreview | Merged openstack/ovsdbapp master: Don't call frozen_row if there are no matching events https://review.opendev.org/c/openstack/ovsdbapp/+/994611 | 09:16 |
| opendevreview | Elvira García Ruiz proposed openstack/neutron master: Add PVLAN documentation https://review.opendev.org/c/openstack/neutron/+/994760 | 09:30 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron-tempest-plugin master: DNM == Migrate QoS with placement tests from tempest https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/995020 | 09:47 |
| opendevreview | Merged openstack/os-ken master: zuul: Use openstack-python3-next-jobs template https://review.opendev.org/c/openstack/os-ken/+/994682 | 10:12 |
| opendevreview | yatin proposed openstack/neutron master: [UT] Update base_mac config override https://review.opendev.org/c/openstack/neutron/+/995028 | 10:59 |
| opendevreview | Merged openstack/neutron master: evpn: Add priority to EVPN callbacks https://review.opendev.org/c/openstack/neutron/+/994739 | 11:15 |
| opendevreview | Merged openstack/neutron-tempest-plugin master: evpn: Add API router tests for EVPN extension https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/987355 | 11:15 |
| ralonsoh | Hi folks, I can't access to the agenda, the website is down | 12:59 |
| ralonsoh | But I have the topics to be discussed | 12:59 |
| ralonsoh | #startmeeting neutron_drivers | 13:00 |
| opendevmeet | Meeting started Fri Jun 26 13:00:03 2026 UTC and is due to finish in 60 minutes. The chair is ralonsoh. Information about MeetBot at http://wiki.debian.org/MeetBot. | 13:00 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 13:00 |
| opendevmeet | The meeting name has been set to 'neutron_drivers' | 13:00 |
| froyo | o/ | 13:00 |
| ralonsoh | ping haleyb|out lajoskatona slaweq mlavalle bcafarel rubasov | 13:00 |
| ralonsoh | and anyone else that I don't recall now | 13:00 |
| ralonsoh | (as commented, I can't access now to the drivers agenda, the site is down) | 13:00 |
| mlavalle | \o | 13:01 |
| ralonsoh | slaweq, lajoskatona can you attend to the meeting? | 13:01 |
| ralonsoh | froyo, sorry, we don't have enough quorum | 13:03 |
| froyo | yeah, I see | 13:03 |
| ralonsoh | in any case, after reviewing https://bugs.launchpad.net/neutron/+bug/2156981 (that was the first topic) | 13:03 |
| ralonsoh | I don't think a spec would be needed | 13:03 |
| ralonsoh | it seems more like a refactor | 13:03 |
| froyo | yeah, it could be consider a refactor also yeah | 13:04 |
| ralonsoh | and this table Load_Balancer_Group has been there since 20.03, so stable enough and no incompatibilities with the current OVN support | 13:04 |
| ralonsoh | IMO, you can start working on it and try to present it next week | 13:04 |
| ralonsoh | I'm 100% sure this RFE will be well received | 13:04 |
| froyo | sure, I have already a patch started | 13:05 |
| froyo | basically a quick summary | 13:05 |
| ralonsoh | perfect! | 13:05 |
| froyo | PROBLEM: Current implementation associates individual Load_Balancers directly to Logical_Routers (lr_lb_add per LB). | 13:05 |
| froyo | In multi-router topologies with many LBs, this creates N×M associations scaling poorly in large deployments. | 13:05 |
| froyo | SOLUTION: Use OVN's Load_Balancer_Group table (available since OVN 21.09). One group per router | 13:05 |
| froyo | (octavia_router_{uuid}) containing all LBs, reducing N individual lr_lb_add to 1 group association. | 13:05 |
| ralonsoh | so if you have a POC, that will be better for the RFE presentation | 13:05 |
| froyo | "refactor" allows that improvement ^ | 13:05 |
| ralonsoh | cool | 13:06 |
| ralonsoh | so I'll close the meeting now and move the topics for the next week | 13:06 |
| ralonsoh | I hope we'll have more people attending | 13:06 |
| slaweq | Hi | 13:06 |
| froyo | ok, so next week I will do the same nutshell (better formatted) and link to the patch | 13:06 |
| slaweq | Sorry. I missed notifications | 13:06 |
| ralonsoh | ok, hold on | 13:06 |
| ralonsoh | we are now 3 drivers, I think that's enough | 13:07 |
| ralonsoh | froyo, please present the topic | 13:07 |
| froyo | ok | 13:07 |
| ralonsoh | #link https://bugs.launchpad.net/neutron/+bug/2156981 | 13:07 |
| lajoskatona | o/ | 13:07 |
| opendevreview | Merged openstack/neutron master: ovn: Handle missing ``neutron:network_name`` in sibling LRPs https://review.opendev.org/c/openstack/neutron/+/993617 | 13:07 |
| lajoskatona | sorry, I lost my clock..... | 13:07 |
| froyo | so the RFE I'm proposing is linked above | 13:07 |
| froyo | RFE for Load_Balancer_Group support in ovn-octavia-provider | 13:07 |
| froyo | Basically urrent implementation associates individual Load_Balancers directly to Logical_Routers (lr_lb_add per LB). In multi-router topologies with many LBs, this creates N×M associations scaling poorly in large deployments. | 13:08 |
| froyo | and new implementation (refactor could match here) will use OVN's Load_Balancer_Group table (available since OVN 21.09). One group per router (octavia_router_{uuid}) containing all LBs, reducing N individual lr_lb_add to 1 group association. | 13:09 |
| froyo | the functionality keeps the same but we would have several benefits: | 13:09 |
| froyo | - scalability: one single association vs N individual LB associations | 13:10 |
| froyo | - Cleaner data model: router -> LBs relation | 13:10 |
| froyo | - OVN native solution | 13:10 |
| ralonsoh | I was going to ask about the functionality, but you already answered that | 13:11 |
| froyo | and that is basically what I have | 13:11 |
| ralonsoh | you'll handle the existing LBs, right? | 13:11 |
| mlavalle | so it boils down to taking advantage of an optimization in our underlying technology (OVN). I think it is a no brainer | 13:12 |
| ralonsoh | migrating them to the LBG soluion | 13:12 |
| froyo | yeah, initially will send a patch with new logic, and in chain a maintenance task to update existing LBs | 13:12 |
| ralonsoh | perfect | 13:12 |
| froyo | mlavalle, yeah! | 13:12 |
| ralonsoh | so IMO there is no need for spec, as mlavalle commented, this is a logical and expected proposal | 13:12 |
| slaweq | yeah, I agree with that | 13:13 |
| lajoskatona | sounds great , and good that you push it | 13:13 |
| ralonsoh | so no need for voting, I'll update the RFE approving it | 13:13 |
| ralonsoh | thanks froyo!! | 13:13 |
| froyo | thank you folks! | 13:13 |
| slaweq | there is no any api or other user facing changes there, it is just internal implementation changes AFAIU, so for me it is fine to go | 13:13 |
| mlavalle | to formalize my position: | 13:13 |
| mlavalle | +1 | 13:13 |
| froyo | slaweq, correct! | 13:14 |
| ralonsoh | cool, that was fast, waiting for the patch (or patches) | 13:14 |
| froyo | ;) | 13:15 |
| ralonsoh | I have another topic, I couldn't add to the meeting agenda because is down | 13:15 |
| ralonsoh | #link https://bugs.launchpad.net/neutron/+bug/2158253 | 13:15 |
| ralonsoh | [RFE] Wait for ovn-controller OpenFlow rules during live migration | 13:15 |
| ralonsoh | this is a follow up of https://review.opendev.org/q/topic:%22bug/2069718%22 | 13:16 |
| ralonsoh | the goal: to allow the port binding only when the destination host was the OF rules installed | 13:16 |
| ralonsoh | after talking to OVN folks, that could be done now using the NB_Global.nb_cfg flag | 13:16 |
| ralonsoh | we can force an update of this value in the NB when we receive the port binding of the migrated port | 13:17 |
| ralonsoh | and wait for the destination chassis_private.nb_cfg to update this value | 13:17 |
| lajoskatona | so another OVN optimization? | 13:17 |
| ralonsoh | yes | 13:17 |
| ralonsoh | so at this point we'll know the rules are already written and we can unpause the VM and start transmitting | 13:18 |
| ralonsoh | that's basically all | 13:18 |
| ralonsoh | I have a POC: https://review.opendev.org/c/openstack/neutron/+/994296 | 13:18 |
| ralonsoh | I don't think that deserves a spec, but good testing | 13:18 |
| ralonsoh | and docs | 13:18 |
| lajoskatona | +1 | 13:19 |
| ralonsoh | do you have any question? you can also ask it in the patches too | 13:20 |
| lajoskatona | not from me, let's check the patches | 13:22 |
| mlavalle | to be clear, we are applying this to live migration? | 13:22 |
| ralonsoh | yes, only for OVN with non-dpdk ports | 13:22 |
| lajoskatona | perhaps worth cross checking with Nova, with some nova-next or whatever job they have for nice scenarios | 13:22 |
| ralonsoh | that's Neutron only, we need to send to Nova the vif-plugged event | 13:22 |
| slaweq | [lajoskatona](https://matrix.to/#/@_oftc_lajoskatona:matrix.org)++ to double check with Nova | 13:23 |
| ralonsoh | but we are responsible of when to send it | 13:23 |
| slaweq | but other than that it is also good for me as this is yet another internal optimization really | 13:23 |
| ralonsoh | I mean, the previous feature was a cross team effort | 13:23 |
| lajoskatona | I don't mean to have meeting or similar, I suppose a depends-on patch or similar can be enough to be on the positive side | 13:24 |
| mlavalle | yeah, we are using now NB_Global.nb_cfg | 13:24 |
| ralonsoh | yes, what we need is to ensure that this feature is working, using also the Nova CI jobs | 13:24 |
| ralonsoh | anything in tempest or n-t-p related to live migration | 13:25 |
| ralonsoh | this is why this feature needs to focus on testing and docs | 13:25 |
| mlavalle | as far as I know, Nova has a live migration job that is still using the "fake" interface up event | 13:26 |
| ralonsoh | I need to confirm that will Nova folks during the development of these patches | 13:26 |
| ralonsoh | local testing ("in my machine works"), these patches are working fine | 13:27 |
| mlavalle | one final question: this makes sense in the context where we are using the cross team optimization implemented a few month ago, correct? | 13:28 |
| ralonsoh | yes, that's is mandatory: we need to enable the ovs_creation_tap flag | 13:29 |
| ralonsoh | I proposed a patch to make it True by default | 13:29 |
| ralonsoh | and in 2027.1 it will be removed (assuming True by default always) | 13:29 |
| ralonsoh | https://review.opendev.org/c/openstack/neutron/+/994752 | 13:29 |
| mlavalle | ok, +1 from me | 13:29 |
| lajoskatona | +1 | 13:30 |
| ralonsoh | slaweq, ? | 13:31 |
| ralonsoh | I assume yes based on previous comments (I know he is in another meeting now) | 13:32 |
| slaweq | +1 | 13:32 |
| ralonsoh | cool! | 13:32 |
| ralonsoh | There are no more topics | 13:32 |
| ralonsoh | do you want to add anything else? | 13:32 |
| mlavalle | Please remember that next Friday is a holiday in the USA | 13:32 |
| mlavalle | so neither me nor haleyb|out will be here | 13:33 |
| ralonsoh | then I'll send a mail cancelling next meeting | 13:33 |
| ralonsoh | thank you folks, have a nice weekend | 13:33 |
| ralonsoh | #endmeeting | 13:33 |
| opendevmeet | Meeting ended Fri Jun 26 13:33:27 2026 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 13:33 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/neutron_drivers/2026/neutron_drivers.2026-06-26-13.00.html | 13:33 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/neutron_drivers/2026/neutron_drivers.2026-06-26-13.00.txt | 13:33 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/neutron_drivers/2026/neutron_drivers.2026-06-26-13.00.log.html | 13:33 |
| mlavalle | \o | 13:33 |
| lajoskatona | o/ Bye | 13:33 |
| froyo | o/ see u | 13:33 |
| slaweq | have a great weekend | 13:33 |
| slaweq | o/ | 13:33 |
| opendevreview | Lajos Katona proposed openstack/tap-as-a-service master: Extract DB model classes into models https://review.opendev.org/c/openstack/tap-as-a-service/+/995059 | 13:33 |
| opendevreview | Lajos Katona proposed openstack/tap-as-a-service master: Remove nested DB context decorators from private getter methods https://review.opendev.org/c/openstack/tap-as-a-service/+/995060 | 13:33 |
| opendevreview | Lajos Katona proposed openstack/tap-as-a-service master: Introduce OVO objects for TaaS resources https://review.opendev.org/c/openstack/tap-as-a-service/+/995061 | 13:33 |
| stephenfin | ralonsoh: slaweq: lajoskatona: Could one of you sanity check this for me? https://review.opendev.org/c/openstack/python-openstackclient/+/995003/ | 13:38 |
| ralonsoh | stephenfin, right now | 13:40 |
| stephenfin | Thanks. I've asked a few of the nova folks also. I assume it's fine but I've never done anything dual-stack'y like that | 13:41 |
| ralonsoh | stephenfin, from Neutron point of view is valid | 13:42 |
| stephenfin | ty 🙏 | 13:42 |
| slaweq | yes, I also think it is ok | 13:42 |
| slaweq | actually if you don't specify IP neutron will allocate for port one IPv4 and one IPv6 address if there are such subnets in the network | 13:43 |
| ralonsoh | ^^ exactly | 13:43 |
| opendevreview | Arnaud Morin proposed openstack/neutron master: Allow service role to create default SG for other project https://review.opendev.org/c/openstack/neutron/+/994744 | 13:52 |
| opendevreview | yatin proposed openstack/neutron master: [FT] Wait some time for lsp tag to be set https://review.opendev.org/c/openstack/neutron/+/995072 | 14:31 |
| opendevreview | Jakub Libosvar proposed openstack/neutron master: Fix ugly traceback when attaching a subnet to a router https://review.opendev.org/c/openstack/neutron/+/995076 | 14:41 |
| opendevreview | Fernando Royo proposed openstack/ovn-octavia-provider master: Implement Load_Balancer_Group support https://review.opendev.org/c/openstack/ovn-octavia-provider/+/995077 | 14:42 |
| opendevreview | Eduardo Olivares proposed openstack/neutron master: Add upstream EVPN multinode Zuul job https://review.opendev.org/c/openstack/neutron/+/994257 | 14:44 |
| opendevreview | Eduardo Olivares proposed openstack/neutron master: Add upstream EVPN multinode Zuul job https://review.opendev.org/c/openstack/neutron/+/994257 | 14:50 |
| opendevreview | Eduardo Olivares proposed openstack/neutron master: Add upstream EVPN multinode Zuul job https://review.opendev.org/c/openstack/neutron/+/994257 | 15:18 |
| mlavalle | slaweq: are you going to work on RPC for FWaaS RFE you just filed? Do I assign it to you? | 15:27 |
| opendevreview | Merged openstack/neutron-tempest-plugin master: Re-enable PUT body checks in PVLAN port update test https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/993590 | 16:19 |
| opendevreview | Elod Illes proposed openstack/networking-bgpvpn stable/2025.2: WIP: [CI][stable-only] Fix tox-py312 job https://review.opendev.org/c/openstack/networking-bgpvpn/+/988281 | 16:34 |
| opendevreview | Jakub Libosvar proposed openstack/neutron master: Add upstream EVPN multinode Zuul job https://review.opendev.org/c/openstack/neutron/+/994257 | 16:56 |
| opendevreview | Jakub Libosvar proposed openstack/neutron master: Add upstream EVPN multinode Zuul job https://review.opendev.org/c/openstack/neutron/+/994257 | 16:59 |
| opendevreview | Jakub Libosvar proposed openstack/neutron master: Add upstream EVPN multinode Zuul job https://review.opendev.org/c/openstack/neutron/+/994257 | 20:14 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!