Friday, 2026-06-05

opendevreviewHelen Chen proposed openstack/neutron master: Created Single VxLAN Device management for EVPN  https://review.opendev.org/c/openstack/neutron/+/99016902:14
opendevreviewHelen Chen proposed openstack/neutron master: Integrated SVD into OVN Agent EVPN Extension  https://review.opendev.org/c/openstack/neutron/+/99109902:27
opendevreviewTerry Wilson proposed openstack/neutron master: Integrated SVD into OVN Agent EVPN Extension  https://review.opendev.org/c/openstack/neutron/+/99109902:29
opendevreviewTerry Wilson proposed openstack/neutron master: evpn: Pass VLAN ID in port binding events to FSM  https://review.opendev.org/c/openstack/neutron/+/99152802:29
opendevreviewBrian Haley proposed openstack/neutron master: WIP: Add project_id attribute for neutron-lib change  https://review.opendev.org/c/openstack/neutron/+/98696203:37
opendevreviewHelen Chen proposed openstack/neutron master: Created Single VxLAN Device management for EVPN  https://review.opendev.org/c/openstack/neutron/+/99016903:52
opendevreviewHelen Chen proposed openstack/neutron master: Integrated SVD into OVN Agent EVPN Extension  https://review.opendev.org/c/openstack/neutron/+/99109903:52
opendevreviewMerged openstack/networking-bagpipe master: Replace OVN Metadata agent with OVN agent  https://review.opendev.org/c/openstack/networking-bagpipe/+/98705005:56
ralonsohlajoskatona, hello! if you have 1 min, please check these 3 backports: https://review.opendev.org/c/openstack/neutron/+/99152306:27
ralonsohthanks!!06:27
opendevreviewBrian Haley proposed openstack/neutron-tempest-plugin master: Remove all usage of tenant/tenant_id  https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/98795906:31
* haleyb is not really here, he is sleeping :)06:32
opendevreviewRodolfo Alonso proposed openstack/neutron master: Use idl.has_lock instead of idl.is_lock_contended for OVSDB lock checks  https://review.opendev.org/c/openstack/neutron/+/99134606:46
ralonsohhehehe06:54
opendevreviewFelix Moebius proposed openstack/neutron master: [OVN] Fix race condition during floating ip deletion  https://review.opendev.org/c/openstack/neutron/+/98823507:00
opendevreviewMerged openstack/neutron-tempest-plugin master: Add scenario test for VM connectivity with allowed_address_pairs  https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/98413307:19
opendevreviewMerged openstack/neutron stable/2025.1: Fix tunnel network detection to use tunnel type driver ranges  https://review.opendev.org/c/openstack/neutron/+/99060407:19
zigoralonsoh: FYI, I rebased https://review.opendev.org/c/openstack/neutron/+/991523 for Epoxy, since your patch didn't apply on top of 2025.1. I have no idea how come you could push such patch in Gerrit, and how misteriously some commits are missing or WTF ... :P08:21
zigotest_port.py was a pain to manually apply ... :/08:22
zigo(currently building Debian package, ie: running unit tests...)08:23
zigoI need this up to Dalmatian, right?08:24
zigoGot 6 failures ... :(08:26
zigohttps://paste.opendev.org/show/bPf2fPkevAUhXsidmGvZ/08:26
zigoOk, got some more to backport in fact.08:31
opendevreviewThomas Goirand proposed openstack/neutron master: Replace keepalived-state-change by a dash script  https://review.opendev.org/c/openstack/neutron/+/83614008:41
opendevreviewFiorella Yanac proposed openstack/neutron-tempest-plugin master: Add OVN PVLAN scenario connectivity test  https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/99171508:43
opendevreviewFelix Moebius proposed openstack/neutron master: [OVN] Add migration routine for ovn_router_indirect_snat  https://review.opendev.org/c/openstack/neutron/+/99177109:01
zigoralonsoh: FYI, my rebased patch is here: https://salsa.debian.org/openstack-team/services/neutron/-/blob/debian/epoxy/debian/patches/OSSA-2026-021_Fix_port_RBAC_policies_to_require_network_ownership.patch?ref_type=heads09:14
zigoIt only applies on top of Neutron 26.0.3, my rebase on 26.0.0 fails unit testing for the moment.09:14
opendevreviewMerged openstack/neutron master: Add PVLAN OVN Driver  https://review.opendev.org/c/openstack/neutron/+/98515009:17
opendevreviewMerged openstack/neutron-tempest-plugin master: [CI][unmaintained] Pin designate tempest plugin  https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/99166709:18
ralonsohI'll check it today09:37
opendevreviewRodolfo Alonso proposed openstack/neutron stable/2026.1: Fix cross-project access to router conntrack helpers  https://review.opendev.org/c/openstack/neutron/+/99177609:39
opendevreviewRodolfo Alonso proposed openstack/neutron stable/2025.2: Fix cross-project access to router conntrack helpers  https://review.opendev.org/c/openstack/neutron/+/99177709:39
opendevreviewRodolfo Alonso proposed openstack/neutron stable/2025.1: Fix cross-project access to router conntrack helpers  https://review.opendev.org/c/openstack/neutron/+/99177809:39
opendevreviewMerged openstack/neutron unmaintained/2024.1: [unmaintained only] Drop fullstack fips job  https://review.opendev.org/c/openstack/neutron/+/99154009:41
opendevreviewMerged openstack/neutron stable/2025.1: Allow service role to create/update port device_id  https://review.opendev.org/c/openstack/neutron/+/99160710:20
opendevreviewMerged openstack/neutron unmaintained/2023.1: [unmaintained only] Drop fullstack fips job  https://review.opendev.org/c/openstack/neutron/+/99154110:20
opendevreviewLajos Katona proposed openstack/neutron-dynamic-routing master: CI: remove py310 job and use desc for starting worker  https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/99062312:11
opendevreviewThomas Goirand proposed openstack/neutron master: Replace keepalived-state-change by a dash script  https://review.opendev.org/c/openstack/neutron/+/83614012:43
opendevreviewFiorella Yanac proposed openstack/neutron-dynamic-routing master: Fix Pass desc to add_agent_status_check_worker for DR agent checks  https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/99185012:44
wilkmarcinhello o/12:54
haleybwilkmarcin: hi, i'll start the meeting in a minute12:59
wilkmarcinack13:00
haleyb#startmeeting neutron_drivers13:00
opendevmeetMeeting started Fri Jun  5 13:00:27 2026 UTC and is due to finish in 60 minutes.  The chair is haleyb. Information about MeetBot at http://wiki.debian.org/MeetBot.13:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.13:00
opendevmeetThe meeting name has been set to 'neutron_drivers'13:00
haleybPing list: ykarel, mlavalle, mtomaska, slaweq, tobias-urdin, lajoskatona, haleyb, ralonsoh, cardoe13:00
mlavalle\o13:00
ralonsohhi13:00
lajoskatonao/13:01
wilkmarcino/13:01
mtomaskao/13:01
haleybalright we can get started13:01
haleybthere are two items on the agenda today13:01
haleybfirst one is13:01
haleyb#link https://bugs.launchpad.net/neutron/+bug/202648913:02
haleyb[RFE][quota] Implement "router_route" quota using the Neutron quota engine, replacing "max_routes" config option13:02
slaweqo/13:02
haleyb#link https://review.opendev.org/c/openstack/neutron/+/99102413:02
ralonsohAnd the gerrit links13:02
ralonsoh#link https://review.opendev.org/q/topic:%22bug/2026489%2213:02
ralonsoh(neutron, sdk and cli)13:02
ralonsohSo the proposal is simple: to remove the config option for router_routes, that is a static option in the config files13:03
ralonsohand add it to the quota engine, that makes it configurable without restarting13:03
ralonsohand per project (as any quota)13:03
ralonsohthat's the proposal13:03
ralonsohI think we don't even need a spec (IMO)13:04
mlavallegood proposal13:04
ralonsohany feedback on this?13:06
slaweqone question13:06
slaweqabout the old option - was it per project or per router?13:06
ralonsohper system13:06
ralonsohone single config option for all projects and routers13:06
slaweqahh, so if e.g. someone had it set to 100 it meant that there can be max 100 routes created across all projects in the cloud, correct?13:07
slaweqand now it will be that each project can have 100 routes created13:07
ralonsohexactly13:08
slaweqso this is pretty big change from the operator PoV IMO - do you know why there was such global limit previously, any potential performance issues, or what?13:09
slaweqone more thing: you said it was global max value but help message in https://review.opendev.org/c/openstack/neutron/+/991024/4/neutron/conf/db/extraroute_db.py says that it was "per router"13:09
ralonsohthe performance is per router, as far as I know13:09
haleybit would have been a fixed 30 per-project before right? with all projects having the same value13:10
ralonsohhmmmm hold on (this is why I should not multitask)13:10
haleybwhere with this each can have it's own13:10
haleybor am i mis-understanding13:10
haleybor i guess max_routes per project13:11
ralonsohok ok, I think slaweq found an issue in the propsed patches13:11
ralonsohso the old config option was PER ROUTER13:11
slaweqdid I? :)13:11
ralonsohyes13:11
slaweqI just wanted to clarify what and how will change with this proposal13:11
ralonsohbecause the _validate_routes method (https://review.opendev.org/c/openstack/neutron/+/991024/4/neutron/db/extraroute_db.py) was checking the number of routes per router13:11
ralonsohahhhh nononono13:12
ralonsohpfffff13:12
ralonsohdelete your minds13:12
ralonsohso, the `max_routes` (old config option) was per router13:12
slaweqit's friday afternoon - I've already deleted almost all my minds :D13:12
ralonsohthis option was unique for all projects13:12
ralonsohthe quota value: it is configurable per project13:12
ralonsohso each project can have its own value13:12
ralonsohbut13:12
ralonsohthe router_routes limit is PER ROUTER13:13
ralonsohso the same behaviour: it is limiting the number of routes PER ROUTER13:13
ralonsoh(but you can define a value per project)13:13
slaweqok, now I have another question then13:13
mlavalledo we have evidence that operators are experiencing issues with the current limit?13:14
slaweqif you are saying that new option quota_router_route is actually number of routes per router not per project, how you want to track usage of this quota?13:14
ralonsohmlavalle, no13:14
mlavallewhy change it, then?13:14
ralonsohslaweq, ... that's a good question...13:14
ralonsohmlavalle, because it is wrong to have a quota defined in a config option, not in the quota engine13:15
ralonsohslaweq, so that is a good question: how to print the number of router_routes in use???13:16
slaweqpersonally I think that good approach could be to define new quota as per project so that owner of the project would be able to use those routes as they want across routers they have13:16
ralonsohbecause this is per resource (router), not per project13:16
ralonsohso the main issue here is how to return the usage13:16
slaweqthat would be changed behavior comparing to what we have now13:16
slaweqbut it would be more consistent with how resources other quota works, e.g. ports, subnets, etc.13:17
ralonsohyes, this is a gap between the other quotas13:17
ralonsohbecause usually we provide quota per project, not per resource13:17
ralonsohI never though about this, to be honest13:17
haleybbut a route is not the same as a floating ip13:17
ralonsohhaleyb, yeah but the point is that we report usage per project13:18
haleybit doesn't consume anything in the environment if i define 10 routes on a router13:18
ralonsohbut that doesn't make sense in this case13:18
slaweq[haleyb](https://matrix.to/#/@_oftc_haleyb:matrix.org) network also don't consume anything really :)13:18
slaweqor SG13:18
slaweqor most of other resources13:18
ralonsohok, let me go back to the planning table. Maybe we can have an exception for this resource (somehow)13:19
ralonsohI'll check other projects quotas too13:19
slaweqso there are really 2 questions here: 1. do we really need quota for routes? 2. if yes, do we really need it to be per router or per project is better to be consistent with other resources13:19
haleybslaweq: true13:19
* haleyb sees the sdk and osc patches have already merged :(13:20
ralonsohI can revert them13:20
ralonsohso give me more time for this and think about how to report it13:20
ralonsohthe router_routes should be per router, because the problem is the amount of them in each router13:21
ralonsohat least when it was implemented for the L3 agent13:21
lajoskatonaso better to have a spec where we can discuss?13:21
ralonsohyes, I'll push an spec13:21
mlavallegood idea13:21
ralonsohwe can always abandon this idea, but let's discuss in the spec13:22
slaweqthx13:22
haleybralonsoh: or we make it 300 to match ROUTER * 3013:22
ralonsohif we have a limit per resource (and the resource is not the project) we should report that13:22
opendevreviewHelen Chen proposed openstack/neutron master: Created Single VxLAN Device management for EVPN  https://review.opendev.org/c/openstack/neutron/+/99016913:23
ralonsohI'll find a way13:23
lajoskatonathanks for checking and working on it13:23
mlavallethanks for the proposal13:24
haleybralonsoh: so should we revert the two that merged? i guess we don't want them released if things will change13:25
mlavalle++13:25
ralonsohhaleyb, yes, I'll propose the reverts for now13:26
haleybso i guess we should vote, it does seem like this should be in the quota engine somehow, so +1 from me for a spec13:27
mlavalle+1 for a spec13:28
lajoskatona+1 for spec13:29
slaweq+1 for a spec13:29
haleybalright that is four, i can mark it approved, and thanks for working on it Claude, err I mean Rodolfo :)13:30
ralonsohhehehe13:30
slaweqLOL13:31
ralonsoh(today I'm installing and adding documents to my own AnythingLLM deployment + qwen3.5 in local)13:31
haleybok, so the second agend item is regarding this bug13:32
haleyb#link https://bugs.launchpad.net/neutron/+bug/213673313:32
haleyband patch13:32
haleyb#link https://review.opendev.org/c/openstack/neutron/+/98956513:33
wilkmarcinyes, so this is related to the OVN driver. I'd like to know your opinions on this #idea: https://review.opendev.org/c/openstack/neutron/+/989565 The idea is to update GW chassis priorities after a highest priority chassis crash (and BFD failover) to reflect the actual state and avoid GW port bouncing13:33
wilkmarcinI provided more details in the commit message13:33
ralonsohSo I've added this topic in out internal meetings. The feedback I had is that the current preemptive behavior keeps the L3 scheduling, avoiding unbalanced chassis assignations13:35
haleybit allows the operator to then manually change the priorities back when they want (like during a planned event)13:35
ralonsohyeah, and now (in a few days) that will be possible via CLI13:36
ralonsohI think we can make this configurable, via CLI, per router 13:37
ralonsoha config option that allows demoting or not the failed chassis13:37
ralonsoh(default=current behavior)13:37
ralonsohor maybe just a general config know13:38
ralonsohknob*13:38
haleybit's tough since we want to keep things balanced if possible, but admins might want this control13:38
wilkmarcinralonsoh: if it was per router, I suppose it means db schema changes -> no possibility to backport it to anything older than master?13:39
slaweqeven though I am usually against new config knobs, I think in this case it may be good idea to add such as both solutions have own pros and cons IIUC, so operator may be able to choose how they want their routers to behave and what is more important for them13:39
ralonsohyes13:39
mlavalle++ good point13:40
ralonsohwilkmarcin, so the config knob allows to backport (in downstream) this feature if needed13:40
lajoskatona+1 for cfg option to make it flexible for operators13:41
slaweqand making it "per router" may actually not give that same choice for operators, even if some routers will choose to behave in new way, load may not be balanced anymore13:41
ralonsohyes, "per router" config is flexible but could be a mess13:41
slaweqif we want to give operators choice: keep balanced load across gw chassis or have less breaks of the traffic then it has to be "all or nothing" IMO13:42
wilkmarcinralonsoh: ack, and yes we are thinking about backporting it all the way to yoga (!), if possible, so the global config option potentially is a better solution from that perspective13:42
ralonsohI don't think this requires a spec but we'll need good documentation (and release notes)13:43
slaweqyeah, spec is not needed for this change IMO13:43
wilkmarcinack, I will keep working on the patch and please provide comments/feedback . 13:44
wilkmarcinas I move along13:44
wilkmarcinthank you!13:45
mlavalleare we going to vote on this?13:45
ralonsoh+1 I like the idea13:45
slaweq+1 from me13:46
mlavalle+113:46
lajoskatona+113:46
mlavallewilkmarcin: you want it to be official, so the vote is a good thing13:47
haleyb+1 from me, but i am biased as i work with marcin and he's done a lot of work on this13:48
wilkmarcin+1 from me, to further pursue the topic13:48
haleybi know there isn't an rfe there, so if you're interested please follow the patch13:49
haleybgreat. that was all on the agenda13:49
haleybany other topics?13:49
lajoskatonano thanks13:50
* mlavalle will be on PTO next Friday13:50
haleybmlavalle: ack13:50
haleybok, thanks for attending everyone, have a nice weekend!13:50
haleyb#endmeeting13:50
opendevmeetMeeting ended Fri Jun  5 13:50:47 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)13:50
opendevmeetMinutes:        https://meetings.opendev.org/meetings/neutron_drivers/2026/neutron_drivers.2026-06-05-13.00.html13:50
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/neutron_drivers/2026/neutron_drivers.2026-06-05-13.00.txt13:50
opendevmeetLog:            https://meetings.opendev.org/meetings/neutron_drivers/2026/neutron_drivers.2026-06-05-13.00.log.html13:50
ralonsohbye13:50
mlavalle\o13:50
wilkmarcinthank you!13:50
slaweqhave a good one13:50
slaweqo/13:50
wilkmarcino/13:50
lajoskatonao/13:51
amnikhaleyb: hi brian. Can I say something?14:00
haleybamnik: sure14:01
opendevreviewTakashi Kajinami proposed openstack/neutron master: Validate [DEFAULT] setproctitle while loading config files  https://review.opendev.org/c/openstack/neutron/+/99165614:01
amnikUnfortunately, I am unable to proceed with my RFE due to a personal issue14:01
amnikWould it be possible to postpone my RFE?14:01
amnikhttps://review.opendev.org/c/openstack/neutron-specs/+/93606314:01
opendevreviewTakashi Kajinami proposed openstack/neutron-lib master: Add constants for setproctitle feature  https://review.opendev.org/c/openstack/neutron-lib/+/99166214:02
haleybamnik: i will mark it postponed and abandon the spec for now, can always restore it14:04
opendevreviewHelen Chen proposed openstack/neutron master: Created Single VxLAN Device management for EVPN  https://review.opendev.org/c/openstack/neutron/+/99016914:11
opendevreviewHelen Chen proposed openstack/neutron master: Integrated SVD into OVN Agent EVPN Extension  https://review.opendev.org/c/openstack/neutron/+/99109914:11
amnikhaleyb: I think, I was disconnected. Did you see my messages?14:13
opendevreviewElvira García Ruiz proposed openstack/neutron master: Fix Update response after updating pvlan-type  https://review.opendev.org/c/openstack/neutron/+/99187214:43
opendevreviewMerged openstack/neutron-dynamic-routing master: CI: remove py310 job and use desc for starting worker  https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/99062314:44
opendevreviewElvira García Ruiz proposed openstack/neutron master: Fix Update response after updating PVLAN properties  https://review.opendev.org/c/openstack/neutron/+/99187214:44
opendevreviewElvira García Ruiz proposed openstack/neutron master: Fix Update response after updating PVLAN properties  https://review.opendev.org/c/openstack/neutron/+/99187214:45
-opendevstatus- NOTICE: Gerrit on review.opendev.org is being upgraded to version 3.13 and will be offline starting at 1600 UTC. We have allocated an hour for the outage window lasting until 1700 UTC. https://lists.opendev.org/archives/list/service-announce@lists.opendev.org/thread/UIFRQR34WYMF3KDMNQFMO222DIK3HTUD/15:02
opendevreviewMerged openstack/neutron master: Fix cross-project access to router conntrack helpers  https://review.opendev.org/c/openstack/neutron/+/99158615:03
opendevreviewMerged openstack/neutron master: evpn: Implement DB model for evpn_vni attribute  https://review.opendev.org/c/openstack/neutron/+/98725015:47
opendevreviewBrian Haley proposed openstack/neutron-tempest-plugin master: Remove all usage of tenant/tenant_id  https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/98795915:59
-opendevstatus- NOTICE: Gerrit on review.opendev.org is being upgraded to version 3.13 and will be offline momentarily. We have allocated an hour for the outage window lasting until 1700 UTC. https://lists.opendev.org/archives/list/service-announce@lists.opendev.org/thread/UIFRQR34WYMF3KDMNQFMO222DIK3HTUD/16:01
opendevreviewIvan Anfimov proposed openstack/neutron master: fix small typo  https://review.opendev.org/c/openstack/neutron/+/99190116:54
opendevreviewIvan Anfimov proposed openstack/neutron-specs master: fix small typo  https://review.opendev.org/c/openstack/neutron-specs/+/99190216:57
opendevreviewIvan Anfimov proposed openstack/neutron-specs master: fix small typo  https://review.opendev.org/c/openstack/neutron-specs/+/99190216:58
opendevreviewMerged openstack/neutron stable/2025.1: Allow service role more RBAC access for Octavia  https://review.opendev.org/c/openstack/neutron/+/99155218:35
opendevreviewJakub Libosvar proposed openstack/neutron master: Created Single VxLAN Device management for EVPN  https://review.opendev.org/c/openstack/neutron/+/99016918:57
opendevreviewJakub Libosvar proposed openstack/neutron master: epvn: Split Svd into generic and Evpn specific  https://review.opendev.org/c/openstack/neutron/+/99193118:57
opendevreviewJakub Libosvar proposed openstack/neutron master: epvn: Split Svd into generic and Evpn specific  https://review.opendev.org/c/openstack/neutron/+/99193119:09
opendevreviewJakub Libosvar proposed openstack/neutron master: epvn: Split Svd into generic and Evpn specific  https://review.opendev.org/c/openstack/neutron/+/99193119:53
opendevreviewJakub Libosvar proposed openstack/neutron master: epvn: Split Svd into generic and Evpn specific  https://review.opendev.org/c/openstack/neutron/+/99193121:08
opendevreviewMiguel Lavalle proposed openstack/neutron master: [OVN] Implement multiple segments per host for routed networks  https://review.opendev.org/c/openstack/neutron/+/97511121:11
opendevreviewMiguel Lavalle proposed openstack/neutron master: [OVN] Agent extension for segment bridges  https://review.opendev.org/c/openstack/neutron/+/98201821:13
opendevreviewBrian Haley proposed openstack/neutron master: WIP: Add project_id attribute for neutron-lib change  https://review.opendev.org/c/openstack/neutron/+/98696221:19
opendevreviewBrian Haley proposed openstack/neutron master: WIP: Add project_id attribute for neutron-lib change  https://review.opendev.org/c/openstack/neutron/+/98696222:29

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!