14:00:23 <ralonsoh> #startmeeting networking
14:00:23 <opendevmeet> Meeting started Tue Mar  7 14:00:23 2023 UTC and is due to finish in 60 minutes.  The chair is ralonsoh. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:23 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:23 <opendevmeet> The meeting name has been set to 'networking'
14:00:24 <ralonsoh> hello all
14:00:33 <slaweq> o/
14:00:37 <haleyb> o/
14:00:39 <obondarev> o/
14:00:39 <elvira> hi!
14:00:44 <rubasov> o/
14:00:59 <bcafarel> o/
14:01:03 <lajoskatona> o/
14:01:09 <ralonsoh> let's start!
14:01:13 <ralonsoh> #topic announcements
14:01:27 <ralonsoh> as usual, during this important weeks
14:01:29 <ralonsoh> #link https://releases.openstack.org/antelope/schedule.html
14:01:44 <ralonsoh> this week is the final release for the non-client libraries
14:01:46 <tobias-urdin> o/
14:02:01 <ralonsoh> sorry, no!
14:02:12 <ralonsoh> (what I'm reading?)
14:02:29 <ralonsoh> next week is the final week for the RC
14:02:38 <frickler> \o
14:02:39 <ralonsoh> so far, nothing went wrong during the last week
14:03:00 <ralonsoh> when we merged the releases for all the Neutron related projects
14:03:20 <lajoskatona> Cool
14:03:21 <frickler> reno generation is still a bit broken, but a fix is in the works
14:03:27 <ralonsoh> but please, keep an eye on the CI and launchpad
14:03:30 <ralonsoh> frickler, which one?
14:03:55 <frickler> 2023.1 sorts before zed and everything else, leading to duplicated entries
14:04:02 <ralonsoh> ah yes yes
14:04:11 <ralonsoh> but seems to be addressed now
14:04:13 <frickler> https://review.opendev.org/c/openstack/reno/+/876581 is the proposed fix
14:04:29 <ralonsoh> yeah and almost merged, so cool
14:05:03 <ralonsoh> ah, I almost forgot, last week we had the "election" polling (there wasn't for Neutron as I was the only candidate)
14:05:13 <ralonsoh> so you'll have me again for the next 6 months
14:05:16 <ralonsoh> sorry for you!
14:05:28 <bcafarel> congratulations here's to 6 more months :)
14:05:41 <tobias-urdin> lucky us! congrats :)
14:05:49 <lajoskatona> +2
14:05:54 <obondarev> thank you, congrats!
14:06:07 <ralonsoh> and please, remember the PTG is very very close
14:06:17 <ralonsoh> and we have a not-so-active etherpad
14:06:21 <ralonsoh> #link https://etherpad.opendev.org/p/neutron-bobcat-ptg
14:06:41 <ralonsoh> next week I'll add all related to sqlalchemy, neutronclient and CI
14:07:05 <ralonsoh> but if you have new features/RFE/bugs, please add them in this etherpad
14:07:23 <ralonsoh> (I'll send another reminder today)
14:07:30 <ralonsoh> something else here??
14:07:50 <bcafarel> just a stable service announcement, now that we have antelope branched
14:08:05 <bcafarel> remember to also cherry-pick to this branch when doing a series of backports :)
14:08:08 <ralonsoh> right, 2023.1
14:08:31 <ralonsoh> (^^^ please don't make my mistake again: antelope is officially called 2023.1 in the code)
14:08:50 <lajoskatona> yeah be careful as this breaks alphabetical order :-)
14:09:30 <ralonsoh> btw, about this, I tested the patches renaming the DB branches
14:09:57 <ralonsoh> luckily the db migration tool only checks the migration IDs, not the directories
14:10:25 <ralonsoh> ok, let's move to the next topic
14:10:27 <ralonsoh> #topic bugs
14:10:35 <lajoskatona> It was my week
14:10:42 <ralonsoh> exactly
14:10:44 <ralonsoh> there is the report
14:10:46 <ralonsoh> #link https://lists.openstack.org/pipermail/openstack-discuss/2023-March/032577.html
14:10:52 <lajoskatona> thanks
14:11:04 <ralonsoh> many bugs but most of them addressed or assigned under investigation
14:11:12 <ralonsoh> (thanks for this active community!)
14:11:17 <lajoskatona> There was only one which has no owner: https://bugs.launchpad.net/neutron/+bug/2009043
14:11:20 <ralonsoh> there are still 3 bugs to be discussed
14:11:24 <ralonsoh> yeah, this is the first one
14:11:42 <lajoskatona> actually the reporter just come back with some result what is the suspicion from their side
14:11:51 <ralonsoh> maybe slaweq could check that
14:11:54 <ralonsoh> if you have time
14:12:13 <lajoskatona> and I would like to as slaweq (or anybody more HA router knowledge) to check
14:12:33 <slaweq> sure, I will try to check it this week
14:12:37 <ralonsoh> thanks a lot
14:12:51 <lajoskatona> yes the reporter wrote that for them this patch is suspicios: https://review.opendev.org/c/openstack/neutron/+/776423
14:12:57 <lajoskatona> slaweq: thanks
14:13:19 <ralonsoh> the next one is
14:13:20 <lajoskatona> thats all from the bugs of last week, all other bugs are active
14:13:26 <ralonsoh> #link https://bugs.launchpad.net/neutron/+bug/2008858
14:13:50 <ralonsoh> I would like you to check this one because the cause (and the fix proposed) are not solving the reported issue
14:14:04 <ralonsoh> what is failing there is the Neutron DB access, not the OVN connectivity
14:14:05 <lajoskatona> ohh, I set it to inactive as there was no answer, but it is active again, thanks
14:14:58 <lajoskatona> ok, thanks ralonsoh
14:15:17 <ralonsoh> I'll check it today but my opinion is the same: in this bug the problem is the DB access, not the OVN connection. This timeout will make things worst
14:15:39 <ralonsoh> the last one in this list is
14:15:44 <ralonsoh> #link https://bugs.launchpad.net/neutron/+bug/2008808
14:15:54 <ralonsoh> but seems inactive since the report
14:16:05 <ralonsoh> to be honest, I don't know how to replicate this
14:16:14 <ralonsoh> did you have this problem before?
14:17:31 <ralonsoh> ok, we'll wait for more info
14:17:45 <ralonsoh> that's all from this list
14:17:50 <ralonsoh> This week slaweq is the deputy, next week will be haleyb
14:17:55 <slaweq> sure
14:18:11 <haleyb> that works
14:18:24 <ralonsoh> cool, thanks both
14:18:44 <ralonsoh> let's move to the next topic
14:18:50 <ralonsoh> #community_goals
14:18:57 <ralonsoh> 1) Consistent and Secure Default RBAC
14:19:11 <ralonsoh> we have an active n-t-p patch
14:19:13 <ralonsoh> #link https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/874709
14:19:18 <ralonsoh> with all the dependencies merged
14:19:33 <ralonsoh> but seems that this job "neutron-tempest-plugin-openvswitch-enforce-scope-new-defaults" is failing
14:19:36 <slaweq> I need to check why it's still failing
14:19:47 <ralonsoh> no rush and thanks
14:19:48 <slaweq> maybe we have missing something in neutron-lib, idk for now
14:20:18 <slaweq> I will ping You once it will be ready to review
14:20:28 <ralonsoh> sure, remember that we'll need that too in Zed
14:21:04 <ralonsoh> actually what is failing is zed job
14:21:06 <slaweq> yeah, in master it works fine, there is something missing in Zed just
14:21:07 <ralonsoh> not master one
14:21:37 <slaweq> in master I will work in this new cycle on the service role for communication between services
14:21:55 <slaweq> and on making new policies to be enabled by default
14:22:16 <lajoskatona> that is already agreed with nova for example?
14:22:26 <ralonsoh> qq, the service role will be available only in Bobcat, right?
14:22:26 <slaweq> lajoskatona what?
14:22:42 <slaweq> ralonsoh yes, I don't plan to backport it
14:22:48 <lajoskatona> I mean how the service role will be used for service-to-service comm
14:23:00 <slaweq> it's agreed generally
14:23:28 <slaweq> we need to identify what calls may be only for service2service and will be available by this new role
14:23:54 <lajoskatona> ok, thanks
14:24:05 <amotoki> admin role is too much and IIRC we agreed to use the service role for such communicatins.
14:24:26 <ralonsoh> yeah, this is not trivial to make a list of all S2S calls
14:24:37 <lajoskatona> do we need to sit down with nova team to clarify the deatils or it will be enough to do the reviews?
14:24:54 <slaweq> I think it will be enough to do reviews
14:25:00 <ralonsoh> because Nova is our client, they mostly should provide this info
14:25:13 <ralonsoh> but we can have a cross session during the next ptg
14:25:14 <slaweq> or if needed, I will add topic to the PTG to talk about that
14:25:15 <ralonsoh> to sync that
14:25:19 <slaweq> but need to check it first
14:25:19 <ralonsoh> correct
14:25:38 <ralonsoh> slaweq, will you ping Nova folks about this?
14:25:49 <ralonsoh> to know if they will be ready for this during the PTG
14:25:52 <slaweq> ralonsoh sure, if needed I will :)
14:25:58 <ralonsoh> perfect
14:26:14 <ralonsoh> thanks again for taking care of this
14:26:42 <ralonsoh> let's move then
14:26:46 <ralonsoh> 2) Neutron client deprecation
14:26:52 <ralonsoh> the agenda: https://etherpad.opendev.org/p/python-neutronclient_deprecation
14:27:04 <ralonsoh> lajoskatona, something new?
14:27:17 <lajoskatona> I think no news
14:27:23 <lajoskatona> the usual etherpad: https://etherpad.opendev.org/p/python-neutronclient_deprecation
14:27:45 <lajoskatona> I have to come back and push my patches as the last ones are still WIPs
14:28:15 <ralonsoh> during the next PTG I'll add a topic for this, mostly to make it public, see what is the progress and to try to involve more people on this
14:28:26 <lajoskatona> I already added :-)
14:28:36 <ralonsoh> right, i see it
14:28:39 <lajoskatona> I will add some details, I just added a line for it
14:28:41 <ralonsoh> perfect
14:29:17 <ralonsoh> we don't have a hard schedule, but let's see if during the next cycle we can deprecate nclient
14:29:24 <ralonsoh> the whole project
14:29:31 <lajoskatona> ok
14:30:05 <ralonsoh> and that's all in this topic
14:30:09 <ralonsoh> let's move to the last one
14:30:11 <frickler> didn't we want to keep it as osc-plugin?
14:30:30 <ralonsoh> the plan is to move everything to sdk
14:30:38 <ralonsoh> and use the OSC, instead of the nclient
14:30:39 <frickler> sdk is only one half
14:30:39 <lajoskatona> we want, but we can remove the python binding code
14:31:10 <frickler> but we can discuss at the ptg. just invite gtema to that session
14:31:22 <ralonsoh> perfect
14:31:25 <amotoki> I think we need moree discussion on osc-plugin from neutronclient
14:31:31 <amotoki> frickler: right
14:31:45 <lajoskatona> frickler: good idea, I ask gtema f he can join
14:31:53 <amotoki> our current target is python bindings
14:32:23 <frickler> right, that is undisputed afaict
14:33:18 <lajoskatona> in the patches which are up for this topic I kept the CLI code in neutronclient, but change to use sdk and not neutronclient binding code
14:33:35 <ralonsoh> I think we can have this conversation in the PTG, but is there any limitation to move all the bindings to SDK?
14:33:40 <lajoskatona> +1
14:34:07 <frickler> even if that is finished, you still cannot deprecate the OSC-plugin part
14:34:24 <ralonsoh> no of course
14:34:49 <frickler> o.k., maybe I misunderstood "deprecate nclient" then
14:34:51 <ralonsoh> but we'll stop doing anything in nclient, that at least is a good goal
14:35:36 <amotoki> frickler: no worries. neutronclient has two parts: CLI and python bindings, and it is always confusing :p
14:36:11 <frickler> well if there happened to be a new feature in n-d-r for example that would require client support, that still would be added to nclient, wouldn't it?
14:36:16 <ralonsoh> actually what was supposedly deprecated (the CLI) can't be yet removed
14:36:48 <ralonsoh> well that's the point lajoskatona job here
14:36:52 <amotoki> ralonsoh: "neutron" CLI was deprecated but OSC plugin is still active I think
14:37:02 <ralonsoh> amotoki, right
14:37:38 <ralonsoh> if n-d-r is still using nclient, it should move to OSC
14:37:54 <ralonsoh> we should avoid any new develpment in nclient
14:38:20 <frickler> all the "openstack bgp ..." commands are in nclient as osc-plugin
14:38:32 <amotoki> n-d-r support is now implemented as OSC plugin (not in OSC).
14:38:33 <frickler> and there are no current plans to change that, or are there?
14:38:59 <lajoskatona> I think we said that no plan for that
14:39:11 <lajoskatona> and let's keep the repos as OSC plugin
14:39:15 <amotoki> perhaps what we first need to do is that OSC plugin for n-d-r consume openstacksdk instead of neutronclient python bindings
14:39:19 <lajoskatona> I mean neutronclient repo
14:39:42 <lajoskatona> amotoki: exactly, that's what I started
14:39:46 <frickler> amotoki: that is what lajoskatona is working on I think, first moving the sdk code to openstacksdk repo
14:39:59 <amotoki> yes
14:40:38 <amotoki> at least we still need to add CLI commands to neutronclient OSC plugin
14:41:05 <amotoki> if we would like to add new commands related to n-d-r, right?
14:41:28 <frickler> that's my understanding, yes
14:41:40 <ralonsoh> and why not spend time first on moving all the bindings to SDK and use the OSC in n-d-r?
14:41:58 <ralonsoh> we'll never finish the migration doing that
14:42:01 <lajoskatona> yes but we have only one SDK for all these when it is finished
14:42:12 <amotoki> +1
14:42:19 <frickler> the commands are implemented in neutronclient, not in n-d-r
14:43:01 <frickler> I can double check if n-d-r uses bindings from neutronclient somewhere
14:43:26 <amotoki> anyway the ptg session would be helpful to share the current situation and futures :-)
14:43:39 <amotoki> the discussion here is a good example
14:44:00 <frickler> from a quick look n-d-r only uses rpc client
14:44:35 <frickler> yes, let's go on and defer to PTG
14:45:35 <ralonsoh> ok, let's move on
14:45:54 <ralonsoh> #topic on_demand_agenda
14:46:13 <ralonsoh> I've been asked to release new versions for stable releases
14:46:19 <ralonsoh> X, Y and Z
14:46:31 <haleyb> U as well?
14:46:37 <ralonsoh> and that makes sense as we haven't done it for a long time
14:46:39 <ralonsoh> not U
14:46:55 <ralonsoh> let me check the state of U
14:47:23 <slaweq> Isn't U in EM phase already?
14:47:32 <haleyb> there have been some merges recently, one of which i'm interested in releasing
14:47:40 <haleyb> yes i do see the -em tag there
14:47:41 <ralonsoh> yeah, is in EM mode
14:47:56 <frickler> could that also become a regular topic? weekly is likely too often, but e.g. in kolla we do a monthly check whether stable releases should be done
14:48:10 <ralonsoh> I'll add it to the agenda
14:48:24 <ralonsoh> I'll check if we can push a new hash for EM branches
14:48:29 <ralonsoh> and how
14:48:40 <frickler> em gets no new tags
14:49:03 <ralonsoh> so that could be a problem then
14:49:04 <frickler> it is consumed only from stable/ussuri or whatever head
14:49:16 <ralonsoh> yes
14:49:24 <haleyb> why? i'm probably just forgetting
14:49:26 <ralonsoh> what I mean is if we can update the EM hash?
14:49:49 <frickler> what is "the EM hash"? move the ussuri-em tag: nope
14:50:22 <ralonsoh> yeah so the only next tag that a EM branch can have is EOL
14:50:25 <ralonsoh> right?
14:50:29 <frickler> yes
14:50:43 <lajoskatona> thats my undertsanding also, the tag is set to some hash and it cant be moved
14:50:44 <frickler> and that also drops the branch
14:50:46 <ralonsoh> haleyb, so no, we can't release any new version for W and older
14:51:11 <opendevreview> Merged openstack/neutron master: Add full support for OVN NB "Gateway_Chassis" table  https://review.opendev.org/c/openstack/neutron/+/874767
14:51:32 <ralonsoh> about Z, Y and X, I think you are ok with this, right?
14:51:34 <haleyb> ralonsoh: ack, i'll ask our downstream team about it
14:52:10 <haleyb> i'm fine with releasing everything else, yes
14:52:31 <lajoskatona> agree for new release
14:52:54 <ralonsoh> ok, I'll push the patches this week and I'll add all of you as reviewers
14:53:20 <lajoskatona> thanks
14:53:22 <frickler> semi-related: is someone working on creating n-t-p zuul.d/2023.1_jobs.yaml ?
14:53:41 <ralonsoh> not yet, I think (let me check)
14:53:58 <frickler> I didn't see a review at least
14:54:12 <ralonsoh> no, I'll do it too this week
14:54:17 <lajoskatona> I can check it
14:54:54 <ralonsoh> btw, I almost forget
14:55:06 <ralonsoh> related to the stable releases
14:55:07 <opendevreview> Merged openstack/neutron stable/yoga: Reintroduce agent bridge resync test  https://review.opendev.org/c/openstack/neutron/+/876464
14:55:14 <ralonsoh> I'll add the rest of the projects too
14:56:11 <ralonsoh> ok, so that's all for today, do you have something else?
14:56:36 <ralonsoh> thank you all and remember that today there isn't CI meeting
14:56:45 <ralonsoh> #endmeeting