Tuesday, 2026-09-22

opendevreviewMerged openstack/neutron master: Do BGP Reconciliation on OVSDB lock acquisition  https://review.opendev.org/c/openstack/neutron/+/100580501:30
opendevreviewBrian Haley proposed openstack/neutron stable/2025.2: ovn: Add ip_substring_filtering API extension support  https://review.opendev.org/c/openstack/neutron/+/100666701:34
opendevreviewBrian Haley proposed openstack/neutron stable/2025.2: ovn: Add ip_substring_filtering API extension support  https://review.opendev.org/c/openstack/neutron/+/100666701:46
opendevreviewBrian Haley proposed openstack/neutron stable/2025.1: ovn: Add ip_substring_filtering API extension support  https://review.opendev.org/c/openstack/neutron/+/100667001:47
opendevreviewMiro Tomaska proposed openstack/neutron master: Expand FSM states to cover retry provisioning steps  https://review.opendev.org/c/openstack/neutron/+/100643101:57
opendevreviewBrian Haley proposed openstack/neutron master: Populate external subnet routing info when dhcp disabled  https://review.opendev.org/c/openstack/neutron/+/100597003:27
opendevreviewMerged openstack/neutron master: tests: skip mismatched BGP password test on FIPS  https://review.opendev.org/c/openstack/neutron/+/100650303:48
opendevreviewMerged openstack/neutron stable/2026.1: dhcp: only use local subnets for dnsmasq --dhcp-range  https://review.opendev.org/c/openstack/neutron/+/100622505:40
opendevreviewMerged openstack/neutron stable/2025.2: dhcp: only use local subnets for dnsmasq --dhcp-range  https://review.opendev.org/c/openstack/neutron/+/100622605:40
opendevreviewMerged openstack/neutron stable/2026.2: ovo: Fix ``qos_network_policy_id`` dirty flag after DB load  https://review.opendev.org/c/openstack/neutron/+/100622805:40
opendevreviewMerged openstack/neutron stable/2025.2: ovo: Fix ``qos_network_policy_id`` dirty flag after DB load  https://review.opendev.org/c/openstack/neutron/+/100623006:04
opendevreviewMerged openstack/neutron stable/2025.1: ovo: Fix ``qos_network_policy_id`` dirty flag after DB load  https://review.opendev.org/c/openstack/neutron/+/100623106:18
opendevreviewMerged openstack/neutron stable/2026.2: ovn, trunk: Skip trunk management for external type parent ports  https://review.opendev.org/c/openstack/neutron/+/100596206:18
opendevreviewMerged openstack/neutron stable/2026.1: ovn, trunk: Skip trunk management for external type parent ports  https://review.opendev.org/c/openstack/neutron/+/100596306:18
opendevreviewMerged openstack/neutron stable/2025.2: ovn, trunk: Skip trunk management for external type parent ports  https://review.opendev.org/c/openstack/neutron/+/100596506:19
opendevreviewMerged openstack/neutron stable/2025.1: ovn, trunk: Skip trunk management for external type parent ports  https://review.opendev.org/c/openstack/neutron/+/100596607:40
opendevreviewEduardo Olivares proposed openstack/neutron master: [bgp] Set OVN_BRANCH to v26.03.3  https://review.opendev.org/c/openstack/neutron/+/100557407:43
opendevreviewJimin Shin proposed openstack/neutron master: Fix metadata/OVN LB HM port MTU set to "None" for non-ML2 callers  https://review.opendev.org/c/openstack/neutron/+/100669909:46
opendevreviewMerged x/whitebox-neutron-tempest-plugin master: Make db sync work with master podified deployment  https://review.opendev.org/c/x/whitebox-neutron-tempest-plugin/+/100619709:47
opendevreviewEduardo Olivares proposed x/whitebox-neutron-tempest-plugin master: [DNM] POC whitebox test which will allow inspecting state of chassis  https://review.opendev.org/c/x/whitebox-neutron-tempest-plugin/+/100036611:38
opendevreviewEduardo Olivares proposed x/whitebox-neutron-tempest-plugin master: [DNM] POC whitebox test which will allow inspecting state of chassis  https://review.opendev.org/c/x/whitebox-neutron-tempest-plugin/+/100036611:39
opendevreviewJakub Libosvar proposed openstack/neutron master: bgp: Put virtual IPs to arp_proxy  https://review.opendev.org/c/openstack/neutron/+/100597411:50
haleyb_i'll start in a minute, system issues here13:01
haleyb_#startmeeting networking13:01
opendevmeetMeeting started Tue Sep 22 13:01:53 2026 UTC and is due to finish in 60 minutes.  The chair is haleyb_. Information about MeetBot at http://wiki.debian.org/MeetBot.13:01
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.13:01
opendevmeetThe meeting name has been set to 'networking'13:01
haleyb_Ping list: bcafarel, elvira, frickler, mlavalle, mtomaska, slaweq, ykarel, lajoskatona, jlibosva, haleyb, ralonsoh, cardoe13:01
mtomaskao/13:02
slaweqo/13:02
frickler\o13:02
bcafarelo/13:02
lajoskatonao/13:02
cardoeo/13:02
haleyb_#topic announcements13:03
haleyb_We are currently in Week R-1 of Hibiscus13:03
mlavalle\o13:03
haleyb_#link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/LEGGB6I4NVRBAYXHICPSXTFOFAFGIFIJ/13:04
haleyb_Final 2026.2 "Hibiscus" release: 30 September, 202613:04
haleyb_my first question is - did we merge anything into stable/2026.2 that would require an update to the RC ?13:05
haleyb_my workstation locked-up right before meeting so need to load tab myself13:05
haleyb_#link https://review.opendev.org/q/project:openstack/neutron+branch:stable/2026.213:06
slaweqI see couple of patches merged even today13:06
slaweqso I guess it would be good to have RC2, no?13:06
haleyb_yeah there is at least one on that list that would be good to have13:07
haleyb_so can i ask for reviews of the two that are left? one seems to be stuck and i'll recheck it13:08
lajoskatonathan let's have an rc213:08
lajoskatonaahh, so there are hanging ones also? 13:09
haleyb_one was, recheck seemed to trigger13:09
haleyb_i will propose a tag later today once they all merge13:09
haleyb_are there any other critical changes not on that list?13:10
haleyb_ok let me know if there is anything else by eod13:10
haleyb_Reminder to please use the priority dashboard for anything neutron/networking related13:11
haleyb_RP +1 for "ready to merge" changes, RP +2 for gate blockers and similar13:11
haleyb_#link https://tinyurl.com/59z278km13:11
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/NeutronDrivers13:11
haleyb_i will not be able to attend this Friday however, and not sure there is anything on the list anyway, so will just proactively cancel now13:13
slaweqthere is my RFE but we can even discuss it during PTG, no rush at all13:13
opendevreviewMerged x/whitebox-neutron-tempest-plugin master: Adapt ha router tests to 2026.2 ha_chassis_group move  https://review.opendev.org/c/x/whitebox-neutron-tempest-plugin/+/100644013:13
mlavalle+113:13
haleyb_slaweq: ah, right, forgot about that one13:14
haleyb_and finally, the PTG is coming up in October13:14
haleyb_Indri" PTG October 12-16, 202613:14
haleyb_PTG Website: https://openinfra.org/ptg/13:15
lajoskatonaohh, and there's an empty ehterpad also : https://etherpad.opendev.org/p/oct2026-ptg-neutron13:15
haleyb_yes, just noticed that too, will update with a header after meeting, but please start thinking about topics you want to discuss13:16
lajoskatona+1, thanks13:17
haleyb_that's all i had for announcements, any others?13:18
slaweqone quick question13:18
haleyb_sure13:18
slaweqI saw yesterday evening email that next week there will be OpenInfra Live for the 2026.2 release highlights13:19
slaweqanyone from our team wants to talk there maybe?13:19
slaweqnot that I want to do so, I just wanted to make sure we are all aware of it :)13:19
haleyb_slaweq: well, you brought it up :)13:20
mlavalleLOL13:20
lajoskatona:-)13:21
slaweqdamn, I knew it will not be good idea :D13:21
haleyb_i think we merged enough things i can present our highlights, guess it's part of my job :)13:21
slaweqthx haleyb_ :)13:21
haleyb_plus my PM here would appreciate it13:22
slaweqand seriously if needed, I can do it of course, I just didn't want to jump ahead of our great leader :)13:22
mlavallefearless leader!13:23
haleyb_hah, i think we are more like the Borg13:23
slaweqLOL13:23
haleyb_alright let's move on13:24
haleyb_#topic bugs13:24
haleyb_mlavalle was the deputy last week, his report is at13:25
haleyb_#link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/IGKOJMBONEPLNRKDIJWQ2QGOJVSESN7H/13:25
haleyb_it was a busy week13:25
haleyb_let's go through the unassigned13:25
haleyb_#link https://bugs.launchpad.net/neutron/+bug/216735113:25
haleyb_[n-t-p] Test "StatelessSecGroupDualStackSlaacTest.test_default_sec_grp_scenarios " failing randomly13:26
haleyb_looks like interface is not getting IPv6 address /13:27
haleyb_IPNetwork('2001:db8:0:1:f816:3eff:fef1:1087/64') not in [IPNetwork('10.1.0.5/28')]13:27
haleyb_not sure if it was discussed at CI meeting yesterday?13:28
haleyb_ok, will follow and see if there are any other occurences with more info13:29
haleyb_some unassigned in the list have been picked-up, looking for an unassigned one13:31
haleyb_this one seemed important and interestingg13:32
haleyb_#link https://bugs.launchpad.net/neutron/+bug/216762013:32
haleyb_[ovn] neutron-server worker permanently broken after an OVN NB reconnect onto fd >= 1024: "filedescriptor out of range in select()" then "TXN queue is full"13:32
haleyb_looks like an issue in python-ovs as it's still using eventlet and not doing poll()13:33
haleyb_i actually saw a proposed change on the ovn/ovs list too13:34
haleyb_jimin3shin[m] did provide a workaround we could put in neutron13:35
haleyb_#link https://launchpadlibrarian.net/877844980/impl_idl_ovn.py13:35
cardoeand that's an annoying package to deal with for many of the packaging options since many of the OpenStack installation mechanisms use the distro package and not from PyPi.13:35
haleyb_might need to ping terry on that, and check the OVS patch13:36
cardoeI've seen a number of reoccurring bug reports against kolla / OpenStack Helm which have been fixed for a while but the distro never pulled in the patch.13:36
haleyb_cardoe: if it's an ubuntu package let me know, but not sure we have either of those13:38
haleyb_last bug i see unassigned is13:39
haleyb_#link https://bugs.launchpad.net/neutron/+bug/216775513:39
haleyb_rare neutron-functional job timeout when workers stopped13:39
mlavalleI've been watching the functional failures in zuul13:40
mlavalleit seems normal to me13:40
mlavalleso I want to ask ichen if we can close it13:40
ichenYes, please close.  I haven’t seen it since I filed the bug.13:41
mlavallecool13:41
haleyb_ack, could have just been zuul hiccup, as i saw it get stuck just today13:41
* mlavalle just marked it invalid13:42
haleyb_so that was all the unassigned bugs, any other bugs to discuss?13:42
haleyb_thanks for taking the busy week mlavalle  :)13:42
mlavalleNP13:42
haleyb_this week lajoskatona is the bug deputy, next week is ralonsoh13:43
haleyb_is that still good for you lajoskatona ?13:43
lajoskatonamy eyes on launchpad13:43
haleyb_great13:43
haleyb_and thanks everyone that picked-up and worked on a bug!13:44
haleyb_#topic specs13:44
haleyb_#link https://review.opendev.org/q/project:openstack/neutron-specs+status:open13:44
haleyb_there are a number of new ones from items discussed over the past couple of weeks13:45
haleyb_so any reviews would be great13:45
haleyb_and just a reminder to authors to move things to the 2027.1 folder if not there already13:46
haleyb_#topic community goals13:46
haleyb_we can go through our active series13:46
haleyb_#link https://review.opendev.org/q/topic:%22sdk_for_neutron%2213:47
haleyb_this is the set of nova patches to move to the sdk13:47
cardoeSo for bugs I've got https://bugs.launchpad.net/neutron/+bug/2167266 which is assigned to me. I've got patches written but no reviews.13:47
cardoeIt's a data loss issue and honestly I should have opened it as a security bug because it is and you can do bad things with it.13:47
haleyb_cardoe: we can discuss more in on-demand now that we've moved on agenda-wise13:48
haleyb_lajoskatona: i saw your ping in the nova meeting yesterday, hopefully we can get that series going13:49
haleyb_next series is13:50
haleyb_#link https://review.opendev.org/q/topic:%2522migrate_stadium_osc%252213:50
lajoskatonahaleyb_: yes, I hope as the release is out we can get reviews13:50
haleyb_lajoskatona: thanks13:50
haleyb_i might have a bad copy/paste on that osc link13:50
haleyb_#link https://review.opendev.org/q/topic:migrate_stadium_osc13:51
haleyb_just the one left13:51
haleyb_#link https://review.opendev.org/c/openstack/python-neutronclient/+/99599313:51
haleyb_we'll have to throw a party after that one13:52
lajoskatonaIf I have time I go back and check the last one for logging still hanging in neutronclient13:52
haleyb_great, thanks13:53
haleyb_next one13:53
haleyb_#link #link https://review.opendev.org/q/topic:%2522neutron_openstacksdk%252213:53
haleyb_#link https://review.opendev.org/q/topic:%2522neutron_openstacksdk%252213:53
haleyb_sigh13:53
lajoskatonauhh, I have merge conflicts13:53
haleyb_#link https://review.opendev.org/q/topic:neutron_openstacksdk13:53
lajoskatonaI have to check as I rebased and updated the first ones last week13:53
haleyb_heat series, all in merge conflict :(13:53
haleyb_will watch for updates13:54
haleyb_and final one13:54
haleyb_#link https://review.opendev.org/q/topic:%22neutron-project-id-migration%2213:54
haleyb_project_id migration series - i have not worked on this the past week13:55
haleyb_that was all for community13:55
haleyb_#topic on-demand13:55
haleyb_cardoe: you had a topic from above13:56
haleyb_#link https://bugs.launchpad.net/neutron/+bug/216726613:56
cardoeI'm happy to change direction to whatever folks want.13:56
ichenSorry to jump in.  I haven’t forgotten cardoe’s request from last week’s meeting to make RT configurable for EVPN.  It doesn’t look like there’s a problem.  I just need to test it before I update the spec.  I plan to finalize EVPN Type-2 spec first.13:57
haleyb_cardoe: i see there are 2 open reviews?13:58
haleyb_i will put them on the priority list and if people have cycles please look13:59
lajoskatona+114:00
haleyb_any other on-demand topics?14:00
haleyb_we are at end of time14:00
haleyb_thanks for attending, have a good week everyone!14:01
haleyb_#endmeeting14:01
opendevmeetMeeting ended Tue Sep 22 14:01:09 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)14:01
opendevmeetMinutes:        https://meetings.opendev.org/meetings/networking/2026/networking.2026-09-22-13.01.html14:01
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/networking/2026/networking.2026-09-22-13.01.txt14:01
opendevmeetLog:            https://meetings.opendev.org/meetings/networking/2026/networking.2026-09-22-13.01.log.html14:01
lajoskatonao/14:01
jamesdentonHi everyone. I don't have a spec, yet, but was wondering how you feel about a new RBAC rule "access_as_readonly" that would help satisfy Nova's need to see a network's properties when booting a VM on a port who's owner != the network owner. Would rather avoid making the network shared=true. Still looking into this route but figured i'd feel it out here first14:01
cardoehaleyb_: I don't see any open ones? Rodolfo gave me a comment and I completely rewrote the patch into the alternate approach I had originally said there were 2 ways to go. So the current implementation is totally different.14:05
cardoejamesdenton: That's be great for not just VMs. But for RDMA networks for bare metal boxes that TheJulia is looking at.14:07
jamesdenton<thumbs up>14:08
TheJuliaIt could be super useful in theory, not sure of the practice, since I would always expect for RDMA networks it all to be the same user... but I guess I could see in some ops heavy shops the tenant/operator boundrary blurring and that becoming more useful/reasonable.14:15
opendevreviewMerged openstack/neutron master: ovn: Advertise the RBAC address group and subnet pool extensions  https://review.opendev.org/c/openstack/neutron/+/100644914:17
jamesdentonwell, the use-case here is to have a network owned by one project and a port created in that network on behalf of another project. This is currently allowed. However, Nova can't see the network and fails to attach that port to the instance. The goal is to satisfy Nova's requirement to read the network properties. We need to avoid shared=true because that gives the project the ability to create ports on that network. We could modify14:19
jamesdenton policy.json to disallow that, but it's an all-or-nothing at that point (from what i can see)14:19
cardoeTheJulia: provider data sources and networks defined but devices plugged up to it14:22
haleyb_cardoe: isn't there https://review.opendev.org/c/openstack/neutron/+/1005615 and https://review.opendev.org/c/openstack/neutron/+/1005616 ?14:25
cardoeyes14:25
TheJuliajamesdenton: yeah, that makes sense but then the question is how do you know it is authorized. Admin made the request, and how much weight do you then apply to that. Dunno.14:25
cardoehaleyb_: the 1st one had a comment from Rodolfo and I then rewrote it using the alternate approach that I had originally said there's 2 ways to approach this... no comments after the fact14:26
cardoehaleyb_: the 2nd one never had any comments14:26
TheJuliacardoe: oh yeah, but I'd love to get folks out of the provider pattern especially with baremetal. Similar to a point I very blunty had to get across to someone last week, your not going to have two different tenants attach to the same provider network you just whipped up because then each tenant has a basis to find each other and begin the hackening. :)14:26
jamesdentonTheJulia thats a good point. Are you saying that one random project might create a port for another project, and that one unknowingly consumes it? Something like that?14:26
cardoeTheJulia: Yeah unfortunately I need neutron to understand that there's two different fabrics in the same neutron for that to really work.14:27
TheJuliajamesdenton: that is sort of how I read that case14:27
TheJuliacardoe: oh yeah, absolutely14:27
cardoeTheJulia: its even more real now with Helen's OVN work because at scale, you're gonna want separate OVN fabrics potentially14:29
haleyb_cardoe: well, we just don't always get to reviews immediately, i sometimes wait for them to hit our priority board14:29
cardoehaleyb_: I was just trying to bring it up during the bugs section of the meeting.14:29
TheJuliaI mean, this is exactly why I intentionally documented in the baremetal l2vni stuffs, split on physnets. Its not perfect, but its a starting point. The annoying thing is you have to then think about that.14:30
cardoehaleyb_: I've also been asked by more than one person if that bug could allow one tenant to access the network of another tenant without being aware and the answer is yes.14:30
cardoeSo it should have been a security bug but unfortunately I didn't catch that when I filed it at first.14:31
cardoeSo it feels like one that should be prioritized is all. I'm not trying to be annoying.14:32
opendevreviewMiro Tomaska proposed x/whitebox-neutron-tempest-plugin master: [DNM] POC whitebox test which will allow inspecting state of chassis  https://review.opendev.org/c/x/whitebox-neutron-tempest-plugin/+/100036615:20
opendevreviewMerged openstack/neutron stable/2026.2: qos: Handle missing port binding in resource request  https://review.opendev.org/c/openstack/neutron/+/100551815:21
opendevreviewMerged openstack/neutron stable/2026.2: ovn: Add ip_substring_filtering API extension support  https://review.opendev.org/c/openstack/neutron/+/100654815:21
opendevreviewEduardo Olivares proposed x/whitebox-neutron-tempest-plugin master: WIP - new test_evpn_vrrp_vip_failover test  https://review.opendev.org/c/x/whitebox-neutron-tempest-plugin/+/100675315:36
opendevreviewLajos Katona proposed openstack/neutron stable/2025.2: Avoid stale flow deletion at ovs-agent startup  https://review.opendev.org/c/openstack/neutron/+/100676116:11
opendevreviewJakub Libosvar proposed openstack/neutron master: evpn: Schedule VRF hosting  https://review.opendev.org/c/openstack/neutron/+/100634816:36
opendevreviewMerged openstack/neutron stable/2026.2: dhcp: only use local subnets for dnsmasq --dhcp-range  https://review.opendev.org/c/openstack/neutron/+/100622417:10
opendevreviewJakub Libosvar proposed openstack/neutron master: evpn: Schedule VRF hosting  https://review.opendev.org/c/openstack/neutron/+/100634818:38
opendevreviewChris Breu proposed openstack/neutron-specs master: Add flavor QoS service plugin spec  https://review.opendev.org/c/openstack/neutron-specs/+/100605418:40
opendevreviewMerged openstack/ovn-octavia-provider master: Remove hacking check for assertItemsEqual  https://review.opendev.org/c/openstack/ovn-octavia-provider/+/100640019:10
opendevreviewHelen Chen proposed openstack/neutron master: EVPN router always redistributes static routes  https://review.opendev.org/c/openstack/neutron/+/100613320:26
opendevreviewMerged openstack/neutron stable/2026.1: ovo: Fix ``qos_network_policy_id`` dirty flag after DB load  https://review.opendev.org/c/openstack/neutron/+/100622920:38
opendevreviewJakub Libosvar proposed openstack/neutron master: evpn: Schedule VRF hosting  https://review.opendev.org/c/openstack/neutron/+/100634820:50
-opendevstatus- NOTICE: The Gerrit service on review.opendev.org will be offline momentarily while we upgrade to a new patch release, but should return within a few minutes23:02
opendevreviewKyuyeong Lee proposed openstack/neutron master: db: Add port group models  https://review.opendev.org/c/openstack/neutron/+/100617123:42
opendevreviewKyuyeong Lee proposed openstack/neutron master: objects: Add port group objects  https://review.opendev.org/c/openstack/neutron/+/100617223:42
opendevreviewKyuyeong Lee proposed openstack/neutron master: api: Add port group CRUD  https://review.opendev.org/c/openstack/neutron/+/100617323:42

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