Friday, 2026-03-20

sean-k-mooneyhaleyb: for what its worht in most project i have interacted with pep8 has alwasy been enforced for tests too00:15
haleybsean-k-mooney: right, and i think i fixed that in neutron last cycle, just seeing some oddities with indentation, and not just in tests, but no pep8 failures00:16
sean-k-mooneythe only thing that is generaly skip for it should be the security checks and even then that just so it dose not warn ofn passwords and such00:16
sean-k-mooneyso ye ar thinking its because of https://github.com/openstack/neutron/blob/master/.pylintrc00:16
sean-k-mooneyi have not seen that in other repos00:16
haleybi have no idea what's causing it at the moment, kuba was just seeing things fail downstream and pushed up some changes00:17
haleybi mean, i could have messed something up00:17
sean-k-mooneyim not seeing any filtering in the pre-commit def https://github.com/openstack/neutron/blob/master/.pre-commit-config.yaml#L47-L56 and your pyproject.toml looks pretty normal00:18
haleybsean-k-mooney: i basically copied it from nova00:18
sean-k-mooneyhaleyb: im not seein ganytin obvious 00:18
sean-k-mooneywell byond .pylintrc00:19
sean-k-mooneyye also have a https://github.com/openstack/neutron/blob/master/.coveragerc00:19
sean-k-mooneythose are the tow things that look diffent form nova at a glance00:19
haleybsean-k-mooney: thanks for looking, what it is with .pylintrc that you see?00:19
sean-k-mooneythe fact it exists00:20
sean-k-mooneynova does not have one00:20
sean-k-mooneyhttps://github.com/openstack/neutron/blob/master/.pylintrc#L800:20
sean-k-mooneyspecificly im wondering ifn the ignore paths has any effect00:20
haleybwe do run pylint separately from a tools script00:21
sean-k-mooneyhum well if i remove those fiels and run pre-commit nothing changes in the repo00:25
sean-k-mooneyso that does not appre to be related00:26
sean-k-mooneysuper odd i can confim that pep8 issues dont appear to be caught in the test modules00:32
sean-k-mooneyoh actully that not in the test it in neutron/service.py00:33
haleybwe borked something up00:36
sean-k-mooney[00:40:39]> uvx autopep8 -d  -r neutron -v00:41
sean-k-mooneyread config path: /home/smooney/repos/neutron/setup.cfg00:41
sean-k-mooneyread config path: /home/smooney/repos/neutron/tox.ini00:41
sean-k-mooneyenable config: section=flake8, key=select, value=H,N00:41
sean-k-mooneyenable config: section=flake8, key=ignore, value=H405,N53000:41
sean-k-mooneyenable config: section=flake8, key=exclude, value=./.*,build,dist,doc00:41
sean-k-mooneywhy do you have exclude = ./.*,build,dist,doc00:42
sean-k-mooneyin tox.ini00:42
haleybit's always been there so ?00:43
sean-k-mooneyoh no that not the issue00:43
sean-k-mooneyso the issue si 00:44
* haleyb waits for the bomb to drop :)00:44
sean-k-mooney[flake8]00:44
sean-k-mooney# We only enable the hacking (H) and neutron (N) checks00:44
sean-k-mooneyselect = H,N00:44
sean-k-mooneyyou only neable the hacking and N which i guess is neutron check00:44
sean-k-mooneythat also contols autopep800:44
sean-k-mooneyyour flake8 section is disbaling everything else which is fine if your using ruff for formating00:45
sean-k-mooneybut neutron currently is not00:45
haleybi thought it was00:45
sean-k-mooneyruff check is enabled00:45
sean-k-mooneybut not ruff format00:45
sean-k-mooneyif i comment out the select line autopep8 starts fixing a bunch of stuff00:46
haleybwtf00:46
sean-k-mooneyautopep8 reads the flake8 secotion of tox ini to know what rules to enforce00:46
sean-k-mooneynova does not use select https://github.com/openstack/nova/blob/master/tox.ini#L318-L344 because nova is still using autopep8 to format things00:48
sean-k-mooneyso the fix is either to adopt ruff for formating 00:48
sean-k-mooneyor to remove the select and keep autopep8 for now00:48
haleybso that change was in be77ed3a813709e by $someoneelse, so might have been an oversight00:49
sean-k-mooneyya at this point i woudl proably jsut add ruff here https://github.com/openstack/neutron/blob/master/.pre-commit-config.yaml#L47-L5100:49
sean-k-mooneyhttps://github.com/openstack/cyborg/blob/master/.pre-commit-config.yaml#L33-L3800:50
sean-k-mooneyi did that in cyborg this cyle00:50
sean-k-mooneythe select thing is correct if you have adoptied ruff but when you do that you also remove autopep800:50
haleybso i don't see ruff in nova, guess it does things differently?00:50
sean-k-mooneynova has not adopted ruff for formatting as some peopel stongly dislike its style00:51
sean-k-mooneyso its still entirly pep800:51
sean-k-mooneythere is a series up to adopt it but it not a battle im currently planning to fight00:52
sean-k-mooneywatcher is in the transation like neutron00:52
sean-k-mooneywith ruff-check for extra linters and autopep8 for formating https://github.com/openstack/watcher/blob/master/.pre-commit-config.yaml#L44-L5200:53
haleybsean-k-mooney: well thanks for finding the issue, luckily i don't think it's caused any issues00:53
haleyboh boy, ruff is pretty aggressive00:53
sean-k-mooneybut again importally no select line https://github.com/openstack/watcher/blob/master/tox.ini#L134-L14100:53
sean-k-mooneyruff has diffent defaults00:54
haleybi think i'll just remove the select line00:54
sean-k-mooneyif you want to be concervitive i woudl say ya remove the select line and let autopep8 fix things for now00:54
sean-k-mooneyand then consider moving to ruff next release00:54
haleybit is the next release :)00:55
sean-k-mooneywell true00:55
haleybbut yeah, i'll wait a while for that as it touched over 400 files00:55
sean-k-mooneyi was thinking you might want to remove the select on the stabel branch and just apply autopep8 on the affected stabel branches00:55
sean-k-mooneyya the cyborg diff was pretty massive00:56
sean-k-mooneyhaleyb: my advice is done try to cherry pick the fix00:56
sean-k-mooneyjsut do a stabel only change to remove the select on the affected stable branches 00:57
sean-k-mooneyand let auto pep8 adress the relevent fixes for each branch00:57
haleybwell i don't think the changes impact anything, so i wouldn't touch stable, mostly is missing newlines and multi-line wrap stuff00:58
sean-k-mooneywell yes00:59
haleybbut kuba will be happy he doesn't have to look into it further00:59
sean-k-mooneyyou coudl leave it as is but it will be anoying when you backprot00:59
sean-k-mooneystephen skiped a sepc in https://github.com/openstack/neutron/commit/be77ed3a813709eebb2659b5473b890b5978c743#diff-ef2cef9f88b4fe09ca3082140e67f5ad34fb65fb6e228f119d3812261ae5144900:59
sean-k-mooneymy guess is an elerive version also enabeld ruff formating01:00
sean-k-mooneyand when swaping to autopep8 that select didnt get removed01:00
sean-k-mooneyah https://review.opendev.org/c/openstack/neutron/+/950640/2 was abandoned01:00
sean-k-mooneyhttps://review.opendev.org/c/openstack/neutron/+/950640/2/.pre-commit-config.yaml01:01
sean-k-mooneyso yes there orgial plan was to also apply ruff format but that never happend01:01
sean-k-mooney"show all 1440 files" ... ya im not going to click that01:02
haleybyeah, he was the $someoneelse, i didn't want to blame him as i probably approved it01:02
sean-k-mooneywell its not really blaming htem it has 4 +2s :)01:02
sean-k-mooneybut i know they have doen the conversion on seveal other repo but few have doen it in mulitple steps where they ran with auto pep8 + ruff check for protacted peroids01:03
sean-k-mooneyso this was just an oversight01:03
sean-k-mooneyanyway i better go to sleep if you need any help fixing it let me know01:04
haleybno, i've been done for a while, just saw the ping, have a good one and thanks for finding it!01:05
opendevreviewRodolfo Alonso proposed openstack/neutron master: [OVN] Fix an issue in the L3 OVN scheduler (3)  https://review.opendev.org/c/openstack/neutron/+/98121108:23
opendevreviewRodolfo Alonso proposed openstack/neutron master: WIP == DNM == OVN L3 scheduler test (3)  https://review.opendev.org/c/openstack/neutron/+/98121208:23
opendevreviewHarald Jensås proposed openstack/neutron stable/2026.1: Include dynamic segments in NetworkContext  https://review.opendev.org/c/openstack/neutron/+/98145608:24
opendevreviewHarald Jensås proposed openstack/neutron stable/2026.1: Add docstrings for filter_dynamic parameter  https://review.opendev.org/c/openstack/neutron/+/98145708:25
opendevreviewStephen Finucane proposed openstack/neutron master: Enable additional ruff rules  https://review.opendev.org/c/openstack/neutron/+/98147110:04
opendevreviewMerged openstack/neutron master: bgp: Fix more pep8 errors  https://review.opendev.org/c/openstack/neutron/+/98121510:52
opendevreviewRodolfo Alonso proposed openstack/neutron master: [OVN] Fix an issue in the L3 OVN scheduler (3)  https://review.opendev.org/c/openstack/neutron/+/98121111:58
opendevreviewRodolfo Alonso proposed openstack/neutron master: WIP == DNM == OVN L3 scheduler test (3)  https://review.opendev.org/c/openstack/neutron/+/98121211:58
opendevreviewMerged openstack/neutron stable/2026.1: Include dynamic segments in NetworkContext  https://review.opendev.org/c/openstack/neutron/+/98145613:04
haleybralonsoh, lajoskatona: just an fyi that stable/2026.1 still isn't running the full set of zuul tests. i have a patch up for tempest which is blocked but can't remember why things like functional aren't running there13:10
lajoskatonahaleyb: ack, thanks for checking13:12
haleybthere must be something over in the openstack-zuul-jobs repo or something ?13:12
haleybi.e. a stable/2026.1 regex13:12
lajoskatonahaleyb: https://review.opendev.org/c/openstack/neutron/+/981131 , ykarel wrote that not all branches were cut13:13
haleyblajoskatona: right, but what about functional?13:14
lajoskatonahaleyb: +1, and I hope we have the branches now13:15
elodillesdevstack and requirements repositories not branched yet, but all the others had13:18
lajoskatonafor functional that is really strange13:18
haleybi just don't remember doing anything special for 2025.2, maybe it's just devstack then13:18
elodilleshaleyb: btw, do you plan to update neutron-tempest-plugin latest release patch?13:19
lajoskatonaelodilles: but than we cant run tempest jobs without devstack, am I right?13:19
lajoskatonahaleyb: true, for functional we also neew devstack, so that can be the culprit13:19
lajoskatonas/neew/need13:19
haleybelodilles: hey, might have just forgot, let me find it and i can ping people while they're online13:20
elodilles+113:21
elodilleslajoskatona: i think the bigger problem would be if devstack + requirements branched, but the others are not. though it is possible that the not yet existing stable/2026.1 of devstack causing issue on neutron's stable/2026.1 branch13:23
haleybwell that isn't happening since we're only running pep8, unit and docs jobs :-o13:23
elodillesthe general template for python unittest jobs should run at least: https://review.opendev.org/c/openstack/openstack-zuul-jobs/+/97661913:25
elodillesi mean, the template is updated, so those jobs should run13:26
haleybright, i think that's why we see unit tests run, functional would rely on devstack :(13:30
lajoskatonastrange because nothing is in the job definition that hardcodes devstack branch13:31
* haleyb shrugs13:32
haleybelodilles: i updated to the latest hash for the n-t-p release, think it should be good just con't confirm since jobs are not running there13:37
elodilleshaleyb: ACK13:47
haleybnot sure if we'll have quorum for drivers meeting but will see14:00
haleyb#startmeeting neutron_drivers14:00
opendevmeetMeeting started Fri Mar 20 14:00:33 2026 UTC and is due to finish in 60 minutes.  The chair is haleyb. Information about MeetBot at http://wiki.debian.org/MeetBot.14:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.14:00
opendevmeetThe meeting name has been set to 'neutron_drivers'14:00
haleybPing list: ykarel, mlavalle, mtomaska, slaweq, tobias-urdin, lajoskatona, haleyb, ralonsoh14:00
mtomaskahi14:01
ichenhello14:01
slaweqhi14:01
lajoskatonao/14:02
haleybhi everyone, and i guess with 3 we have quorum? i can't remember what we reset that to14:02
haleybthere was only one item on the agenda today from ichen 14:04
haleyb#link https://bugs.launchpad.net/neutron/+bug/214461714:04
slaweqaccording to https://launchpad.net/~neutron-drivers/+members#active there are 5 members in total, so 3 is quorum14:04
lajoskatona+114:04
haleybthanks for looking14:05
ichenShould I start?14:05
lajoskatonaso this is an extension for the feature on which jlibosva is working/ just finished?14:05
haleybichen: yes, please14:05
ichenYes, this RFE is to extend the BGP service plugin that jlibosva started.14:05
ichenIt’s to support EVPN type-5 route.14:06
lajoskatonaack14:06
ichenI forgot to mention it in the RFE, but after some prototyping work, the first phase will only support centralized routing.  OVN currently doesn’t make it easy to support DVR.14:07
lajoskatonaand for that you need FRR am I right?14:08
ichenYes.  Jlibosva’s BGP service plugin uses FRR, and configures one BGP instance.  EVPN extension will reuse the same FRR/BGP setup, except that that it’ll use l2vpn evpn address family.14:09
ichenMore precisely, *add* l2vpn evpn address family.14:09
lajoskatonaack, thanks for clarification14:10
haleyband a type-5 route is for a prefix?14:10
ichenAdditionally, each VNI will have a corresponding VRF and BGP vrf instance.14:10
ralonsohhello14:11
haleybo/14:11
ichenYes, type-5 route is only IP prefix, including /32.  The plan is to advertise the VM host routes.14:11
haleybichen: and an entire private subnet i guess?14:12
ichenhaleyb: do you mean the subnet the hosts are on?  I haven’t checked if OVN supports both subnet and host routes simultaneously.14:13
haleybichen: i guess i thought that would just be possible, instead of a /32 a type-5 could be a /28 or /24 or...14:14
ichenYes, that is possible.14:15
ichenI meant that I haven’t tried to advertise both the network and the /32.  I personally think that advertising just the network makes sense without type-2, but I tried the host routes because I got the question explicitly.14:16
haleybis that part of this RFE, or just changing to use a type-5 for /3214:16
ichenI consider it part of this RFE, and whether we advertise /32 or the network is a design decision.14:18
lajoskatonado we need to set it in a design doc/spec ?14:18
ralonsohyes, I think so14:19
ichenThe way to advertise a route is for Neutron to set an option in the NB ovsdb, and I was planning to call out what option Neutron will set.14:19
ichenHere are the possible options: https://github.com/ovn-org/ovn/blob/4bab00d57e5cc3d8d7a6670660fc8f6d63c2bdf9/ovn-nb.xml#L331914:20
ichenThis is the host route: https://github.com/ovn-org/ovn/blob/4bab00d57e5cc3d8d7a6670660fc8f6d63c2bdf9/ovn-nb.xml#L332714:20
ichenThere are more following those two.14:21
ichenSorry, yes, what ralonsho said.  The design doc/spec should call out the options.14:22
opendevreviewStephen Finucane proposed openstack/neutron master: Enable additional ruff rules  https://review.opendev.org/c/openstack/neutron/+/98147114:22
opendevreviewStephen Finucane proposed openstack/neutron master: Re-enable flake8  https://review.opendev.org/c/openstack/neutron/+/98152114:22
opendevreviewStephen Finucane proposed openstack/neutron master: ruff: Enable preview rules  https://review.opendev.org/c/openstack/neutron/+/98152214:22
lajoskatonaichen, ralonsoh: +1 for spec14:23
stephenfinhaleyb: sean-k-mooney: ^ I didn't know so many of the ruff rules were behind the preview flag. Sorry about that :(14:23
haleybichen: yes a spec would be good since there are options on what to support14:24
haleybstephenfin: thanks, will look in a bit14:25
sean-k-mooneystephenfin: oh cool ill take a look at those when i can14:25
haleybanyway, are there more questions about the RFE?14:25
ichenThe spec is almost ready, just need to incorporate some corrections based on the prototype.14:25
sean-k-mooneythanks for spliting thos up it will make the review eaiser14:25
ralonsohichen, will that be an independent extension? Or just an update of the ovn-bgp feature? 14:26
slaweqno questions from me about this RFE14:26
ichenJust an update to what jlibosva created.14:26
ralonsohcan both bgp and evpn work together? (if that is even necessary)14:26
mtomaskaBGP is a mechanism to EVPN working, they have to work together for the overall solution to work. 14:28
ichenYes.  The plan is to reuse the same BGP instance and the same peer(s), with additional EVPN configurations.  (The additional EVPN configurations are in the form of BGP vrf instance, so it sounds like another instance, but everything will be advertised through the same BGP peer(s).)14:28
ralonsoh^ that was the answer I was looking for14:29
ichenThe same BGP peer can advertise regular routes and evpn routes.14:29
ralonsohcool14:29
jlibosvaI thought it was more independent and OVN steers the traffic to the tunnel already from the Neutron router14:29
jlibosvaso so none of the underlaying BGP resources was gonna be used in such case14:30
ichenI assume that OVN would respect the VRF routing.  The regular routes will be routing in Linux 42.  The EVPN routes will be routing based on their respective VRF.14:31
ichenSorry, not “Linux 42”, but “Linux VRF table 42”, which is what the existing BGP service plugin uses.14:32
ralonsohok, so that needs to be clarified in the spec too14:33
ichenI will incorporate a discussion of regular address family and EVPN address family in the spec.14:34
ralonsohbecause how the architecture jlibosva designed, I don't see how the traffic will sent to the bgp peers14:34
jlibosvaI think ichen means if we have a remote workload running in the remote cloud then it's gonna use tunnel created by evpn. If it's some east-west or provider north-south it's gonna be using the underlaying core BGP14:36
ichenAdditionally, a workload is either within an EVPN VRF or not.  If the workload is within an EVPN VRF, then the forwarding is based on the EVPN VRF.14:38
haleybok, so some things to clarify in the spec14:38
haleybany other questions or should we vote?14:38
haleybok, i am a +114:39
slaweq+1 from me14:39
lajoskatona+1 from me with spec14:40
ralonsoh+114:40
haleybok, i will approve it, thanks everyone14:41
ichen Thanks, everyone!14:41
haleybichen: i think we have to create a 2026.2 directory in the specs repo, will check14:41
ichenack14:42
haleybthat was the only thing on the agenda, anything else?14:42
haleybichen: actually looks like it's there14:43
lajoskatonanope14:43
haleybok, thanks everyone for attending, have a nice weekend!14:43
haleybremember the neutron and ci meetings next week have moved an hour for DST14:43
slaweqhave a nice one14:43
haleyb#endmeeting14:43
opendevmeetMeeting ended Fri Mar 20 14:43:38 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)14:43
opendevmeetMinutes:        https://meetings.opendev.org/meetings/neutron_drivers/2026/neutron_drivers.2026-03-20-14.00.html14:43
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/neutron_drivers/2026/neutron_drivers.2026-03-20-14.00.txt14:43
opendevmeetLog:            https://meetings.opendev.org/meetings/neutron_drivers/2026/neutron_drivers.2026-03-20-14.00.log.html14:43
slaweqo/14:43
lajoskatonao/14:43
mtomaskao/14:43
ralonsohbye14:43
ralonsohichen, https://review.opendev.org/c/openstack/neutron-specs/+/98030514:44
ralonsohthat was merged 1 week ago14:44
zigoHas this been implemented in Gazpacho, or not yet ?14:51
zigohttps://opendev.org/openstack/neutron-specs/src/branch/master/specs/2025.2/ovn-bgp-integration.rst14:51
ralonsohjlibosva, ^14:52
ralonsohthere are some missing bits that are almost merged14:52
zigoralonsoh: Is there hope to have them in a Gazpacho point release then?15:12
haleybzigo: that is the plan, we (canonical) want them too15:13
zigoSounds gr8 ! :)15:13
ralonsohhaleyb, hi! So I have the 3rd patch for the L3 OVN scheduler. And the upper patch is testing both affected UTs15:39
ralonsohand I've also increased the coverage of one of these tests15:40
ralonsohhttps://review.opendev.org/c/openstack/neutron/+/981211/15:40
ralonsohthat needs to be backported to 2026.115:40
haleybralonsoh: ack, will look15:41
opendevreviewRodolfo Alonso proposed openstack/neutron master: [OVN] Update the parent hostname in the virtual ports  https://review.opendev.org/c/openstack/neutron/+/98154016:41
ildikov_Hi Neutron team16:57
ildikov_I'm reaching out with a quick question, favor to ask16:58
ildikov_On the last OpenStack Ops Radio Hour folks brought up OVN as a topic area they would like to learn and talk more about16:58
ildikov_The next call is on March 27, next Friday, at 1300 UTC16:59
ildikov_Would any of y'all be available to join the meeting?16:59
opendevreviewChris Buggy proposed openstack/neutron master: Add logging decorator for OVN maintenance methods  https://review.opendev.org/c/openstack/neutron/+/98155617:22
opendevreviewElvira García Ruiz proposed openstack/neutron-lib master: Add PVLAN api extension  https://review.opendev.org/c/openstack/neutron-lib/+/97593117:58
elodilleshaleyb: it seems that now that devstack, grenade and requirements have branched, the gate is back to how it should look like. see the job list: https://zuul.opendev.org/t/openstack/status?change=98113118:02
haleybelodilles: great, jobs are good :)18:03
elodilles:]18:03
haleybhopefully we didn't merge a bad one18:04
elodillesit will turn out soon :S if needed (let's hope not :S fingers crossed) then a DNM empty patch can be proposed to execercise a bit the gate and see that every job is still passing18:06
opendevreviewMerged openstack/neutron unmaintained/2024.1: [OVN] Add a maintenace method to create the AS for each AG  https://review.opendev.org/c/openstack/neutron/+/97790720:07
ichenralonsoh: thanks for the reference.  I’ll take a look.20:19
opendevreviewRodolfo Alonso proposed openstack/neutron master: All service plugins should inherit from ``ServicePluginBase``  https://review.opendev.org/c/openstack/neutron/+/97889820:57
opendevreviewRodolfo Alonso proposed openstack/neutron master: [OVN] Fix an issue in the L3 OVN scheduler (3)  https://review.opendev.org/c/openstack/neutron/+/98121121:16
opendevreviewRodolfo Alonso proposed openstack/neutron master: WIP == DNM == OVN L3 scheduler test (3)  https://review.opendev.org/c/openstack/neutron/+/98121221:17
opendevreviewRodolfo Alonso proposed openstack/neutron master: Revert "Mark OVN L3 test unstable"  https://review.opendev.org/c/openstack/neutron/+/98039721:20
opendevreviewRodolfo Alonso proposed openstack/neutron master: Revert "Mark OVN L3 test unstable"  https://review.opendev.org/c/openstack/neutron/+/98039721:20
opendevreviewBrian Haley proposed openstack/neutron master: Remove sort_ips_by_version()  https://review.opendev.org/c/openstack/neutron/+/98161422:48
opendevreviewBrian Haley proposed openstack/neutron master: Remove remove_interface_suffix() from ip_lib  https://review.opendev.org/c/openstack/neutron/+/98161522:50
opendevreviewTerry Wilson proposed openstack/neutron master: Don't process BGP events without the lock  https://review.opendev.org/c/openstack/neutron/+/98161623:04
opendevreviewMerged openstack/neutron master: Use ovn-db-sync-util related constants from neutron-lib  https://review.opendev.org/c/openstack/neutron/+/97806023:33

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