16:00:26 <gthiemonge> #startmeeting Octavia
16:00:26 <opendevmeet> Meeting started Wed Oct 18 16:00:26 2023 UTC and is due to finish in 60 minutes.  The chair is gthiemonge. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:26 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:26 <opendevmeet> The meeting name has been set to 'octavia'
16:00:31 <johnsom> o/
16:00:34 <gthiemonge> Hi Folks
16:00:38 <QG> o/
16:00:38 <oschwart> o/
16:00:42 <tweining> o/
16:01:34 <gthiemonge> #topic Announcements
16:01:39 <gthiemonge> * 2024.1 Caracal vPTG
16:01:49 <gthiemonge> the PTG is next week!
16:02:01 <gthiemonge> on Tuesday 14:00 UTC - 18:00 UTC
16:02:07 <gthiemonge> (I don't think we really need 4 hours)
16:03:03 <gthiemonge> and of course, the etherpad:
16:03:08 <gthiemonge> https://etherpad.opendev.org/p/caracal-ptg-octavia
16:03:20 <QG> thanks !
16:03:49 <gthiemonge> QG: will you attend the PTG?
16:04:03 <gthiemonge> oh yeah I see your name now
16:04:04 <QG> i think so
16:04:05 <gthiemonge> cool
16:04:46 <gthiemonge> so Folks, if you have some time before the PTG, maybe you can read the specs proposals
16:04:55 <gthiemonge> LB resize
16:04:57 <gthiemonge> https://review.opendev.org/c/openstack/octavia/+/885490
16:05:01 <gthiemonge> SRIOV
16:05:03 <gthiemonge> https://review.opendev.org/c/openstack/octavia/+/891244
16:05:13 <gthiemonge> Active-Active BGP
16:05:15 <gthiemonge> https://review.opendev.org/c/openstack/octavia/+/888020
16:05:25 <gthiemonge> ACMEv2 Cert management
16:05:28 <gthiemonge> https://review.opendev.org/c/openstack/octavia/+/877281
16:06:43 <gthiemonge> do we have any other announcements?
16:07:28 <tweining> not from my side
16:07:39 <johnsom> I don't
16:07:50 <gthiemonge> ack
16:07:58 <gthiemonge> #topic CI Status
16:08:18 <gthiemonge> FYI we need to update the zuul config on stable branches
16:08:23 <gthiemonge> there are still some centos-8-stream jobs (non-voting)
16:08:48 <gthiemonge> they fail and one job had been stuck in the CI for 24h
16:09:06 <gthiemonge> (I just rebased the commit so it killed the stuck job)
16:09:54 <gthiemonge> if we find a volunteer, it's cool
16:09:57 <gthiemonge> or I'll do it
16:11:10 <tweining> if it's not urgent I can do it
16:11:15 <gthiemonge> +1
16:12:26 <gthiemonge> #topic Brief progress reports / bugs needing review
16:13:00 <gthiemonge> FYI we fixed an important bug on master, the amphora configure API call didn't work
16:13:05 <gthiemonge> https://review.opendev.org/c/openstack/octavia/+/898232
16:13:26 <gthiemonge> it needs to be backported to 2023.2 (and we will propose a bugfix release)
16:13:42 <gthiemonge> (the changes required by sqlaclhemy2 introduced the bug)
16:15:53 <QG> i have started working on https://bugs.launchpad.net/octavia/+bug/2039430 but i'm stuck on the functionnal tests in test_load_balancer.py
16:16:28 <QG> it broke 78 functional tests :-D
16:16:59 <gthiemonge> QG: if you need help, maybe you can upload the patch and we can take a look
16:17:09 <gthiemonge> but 78 is not bad
16:17:13 <gthiemonge> :D
16:17:19 <QG> gthiemonge: cool will do
16:17:58 <johnsom> Yeah, sometimes those come down to changing one thing fixes them all
16:19:31 <johnsom> I am making good progress on SRIOV. I have ports attaching to the amphora now. Most of my work over the last week was dealing with failure cases and nova bugs
16:19:54 <johnsom> I have opened a nova bug about some "API unexpected error" 500s I get
16:20:08 <johnsom> In the mean time, I am parsing the exception text to get something useful out of it.
16:21:07 <johnsom> My plan is to get the VIP path working, then do a push, then work on tests, then work on the nftables stuff in a different patch. Following that will be looking at the member path.
16:22:14 <gthiemonge> cool
16:22:46 <QG> you're used to splitting patches when they're too big ?
16:23:28 <johnsom> I am trying to break it down on logical boundaries. But, small patches is a skill I have yet to master. Grin
16:23:54 <tweining> it is good practice and make patches a lot easier to review
16:24:42 <QG> ah ok cool ! was thinking it is harder to test it before all are merged but if they are link by a relation chain then you can tests them all
16:25:09 <johnsom> Right
16:25:12 <gthiemonge> you need to have at least unit tests for all the small parts
16:25:29 <johnsom> Right, the patch must have unit tests to merge
16:26:18 <johnsom> For this one, tempest may not be fully possible as the distros don't have the virtual SR-IOV driver in qemu yet
16:28:51 <gthiemonge> ack
16:30:41 <johnsom> I have not check 23.10 yet to see if it's there.
16:32:02 <johnsom> Ah, it is in 23.10, cool
16:32:19 <gthiemonge> but will it work in the CI?
16:32:57 <johnsom> ... maybe ....  I don't know. I have a cookbook to get it working, but I have not tried it yet. I'm developing on real hardware
16:33:20 <johnsom> In theory it's a pure virtual driver that will allow allocating VFs
16:33:45 <johnsom> So, should work if the distro devstack is running on has qemu/libvirt support for it.
16:34:01 <johnsom> I don't know if devstack will support 23.10 or not, it's a short term release
16:36:52 <gthiemonge> ok
16:38:43 <gthiemonge> #topic Open Discussion
16:38:50 <gthiemonge> any other topics folks?
16:39:25 <johnsom> Is this meeting cancelled next week? Since we will have the PTG
16:40:15 <gthiemonge> right, I think we don't need a meeting
16:40:19 <gthiemonge> I'll send an email
16:40:29 <QG> on another subject, about the python-octaviaclient i have some questions : do you know why the python-octaviaclient is not integrated as a dependency in python-openstackclient, like neutron of swift ( the fact that we need to install it in addition to openstackclient )
16:40:48 <gthiemonge> hmm good question
16:40:50 <gthiemonge> I don't know
16:41:09 <tweining> +1 for that suggestion from me
16:41:36 <johnsom> Yeah, so OpenStack is (was?) moving to a plugin model for all of the clients. Neutron and swift were "legacy" in repo before the plugin model was created.
16:42:00 <johnsom> In theory, all of the projects were supposed to go to the plugin model.
16:42:27 <johnsom> It also makes getting patches merged easier as the core team can be Octavia people, as opposed to openstackclient folks, etc.
16:42:59 <gthiemonge> I think you're talking about adding octaviaclient to this list https://opendev.org/openstack/python-openstackclient/src/branch/master/requirements.txt
16:42:59 <johnsom> Now, the movement to plugins (many projects are plugins, not just Octavia) has slowed
16:43:03 <gthiemonge> right?
16:43:37 <johnsom> Ah, I thought i was just the plugin model issue since QG mentioned neutron and swift
16:43:42 <tweining> neutron is a core service of openstack (not sure about swift) so it makes sense to require that to be installed
16:44:30 <gthiemonge> I think the key is to not have to install it manually
16:44:38 <johnsom> #link https://opendev.org/openstack/python-openstackclient/src/branch/master/openstackclient/network
16:45:02 <johnsom> Well, not all clouds have all services, so installing everything may not make sense.
16:45:29 <QG> ok make sens thanks !
16:45:44 <johnsom> I would not be opposed to adding Octavia to the requirements.txt there, like keystone, etc.
16:46:02 <gthiemonge> someone can propose it
16:46:32 <johnsom> I mean, cinder (an optional service) is there, so...
16:48:00 <QG> i can do it if you want, no issue
16:48:37 <gthiemonge> QG: if you propose a patch, send the link here, I'll +1
16:49:08 <QG> gthiemonge: ok sure !
16:50:09 <QG> and do know if now that the new openstack release is out we can release https://review.opendev.org/c/openstack/python-octaviaclient/+/896255 on pypi ?
16:50:38 <gthiemonge> oops, I should have done that
16:50:53 <gthiemonge> I'll propose a bugfix release for python-octaviaclient
16:51:20 <QG> gthiemonge: cool thanks !
16:52:55 <gthiemonge> anything else Folks?
16:53:46 <QG> Nothing from me
16:54:23 <oschwart> Nothing from me
16:55:10 <johnsom> Nope, looking forward to the PTG!
16:55:14 <gthiemonge> ok
16:55:14 <gthiemonge> yeah
16:55:22 <gthiemonge> see you at the PTG!
16:55:30 <gthiemonge> #endmeeting