14:00:14 <lajoskatona> #startmeeting networking 14:00:14 <opendevmeet> Meeting started Tue Nov 16 14:00:14 2021 UTC and is due to finish in 60 minutes. The chair is lajoskatona. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:14 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:14 <opendevmeet> The meeting name has been set to 'networking' 14:00:18 <mlavalle> o/ 14:00:20 <lajoskatona> o/ 14:00:23 <obondarev> hi 14:00:45 <slaweq> o/ 14:01:59 <lajoskatona> ok, let's start 14:02:03 <lajoskatona> #topic Announcements 14:02:16 <lajoskatona> The usual Yoga cycle calendar https://releases.openstack.org/yoga/schedule.html 14:02:42 <lajoskatona> We are in Week R-19 ses the mail about : http://lists.openstack.org/pipermail/openstack-discuss/2021-November/025756.html 14:02:58 <ralonsoh> hi 14:03:02 <lajoskatona> neutron-lib 2.17.0 is out, we had issue with ovo hash change due to constant changes in neutron-lib 14:03:12 <lajoskatona> requirements patch is slowly moving in: https://review.opendev.org/c/openstack/requirements/+/817763 14:03:23 <lajoskatona> thanks ralonsoh for pushing 14:03:26 <ralonsoh> yw 14:03:31 <ralonsoh> and https://review.opendev.org/c/openstack/neutron/+/818067 14:04:00 <lajoskatona> that's a duplicate of https://review.opendev.org/c/openstack/neutron/+/817870 :-) 14:04:14 <lajoskatona> but we can sort it out which to merge finally :-) 14:04:17 <ralonsoh> not really 14:04:49 <lajoskatona> oh I see it is doing other things as well, ok I will check it 14:05:44 <slaweq> 818067 can be on top of 817870 if You want to have both of them :) 14:05:45 <lajoskatona> ok, thanks, so keep an ey on these if you have few mins for it please 14:05:58 <lajoskatona> I can leave without it 14:06:04 <slaweq> :) 14:06:36 <lajoskatona> so that's it for announcements from me, do you have anything else to announce? 14:07:31 <lajoskatona> I checked the doodles for meeting time slot, and I need more time to see which would be the best 14:07:53 <lajoskatona> ok we can move forward 14:07:57 <lajoskatona> # topic Community Goals 14:08:21 <lajoskatona> The new goal for RBAC is forming: https://review.opendev.org/c/openstack/governance/+/815158 14:09:12 <lajoskatona> if I understand well in this cycle not much is expected from Neutron, as Nova will be the first to adopt new personas and whatever 14:10:44 <lajoskatona> slightly related, but we discussed "community wide" topics during the PTG, one was edge and document somewhere how Neutron/designate works in edge environment 14:11:12 <lajoskatona> I discussed it with ildikov and I will send out a mail like this: https://etherpad.opendev.org/p/edge_documentation_mail 14:11:34 <lajoskatona> We will see where it will progress if it will 14:12:28 <lajoskatona> the other one is deprecating python-neutronclient python bindings, for that I prepared a mail as well: https://etherpad.opendev.org/p/deprecate_neutronclient_py_bindings 14:12:48 <lajoskatona> if you have any thoughts please share it with me, or comment to the etherpads 14:13:36 <slaweq> thx lajoskatona 14:13:41 <slaweq> looks good for me 14:14:23 <lajoskatona> slaweq: I tried not running to the china shop.... 14:14:54 <slaweq> :) 14:15:04 <lajoskatona> If no comments for community goals 14:15:12 <lajoskatona> #topic Bugs 14:15:24 <lajoskatona> slaweq was bug deputy last week: http://lists.openstack.org/pipermail/openstack-discuss/2021-November/025770.html 14:16:03 <slaweq> it was not very busy week for me :) 14:16:11 <lajoskatona> Would you like to highlight anything from the list ? 14:16:15 <slaweq> but there is couple of bugs which aren't assigned to anyone on that list 14:16:17 <lajoskatona> that's always good :-) 14:16:54 <slaweq> there is one high bug related to the logging and ovn: https://bugs.launchpad.net/neutron/+bug/1950273 14:16:59 <slaweq> it has no assignment 14:17:07 <slaweq> maybe someone will want to take a look into that 14:17:15 <ralonsoh> I can 14:17:30 <slaweq> it don't happens always but when it happened there is stack trace in the neutron-server's log IIRC 14:17:47 <slaweq> so hopefully it will be easy to fix :) 14:17:51 <slaweq> thx ralonsoh 14:18:10 <ralonsoh> easy to fix... don't lie to me!! 14:18:15 <slaweq> there is also https://bugs.launchpad.net/neutron/+bug/1950686 and related to that docs bug https://bugs.launchpad.net/neutron/+bug/1899207 14:18:22 <slaweq> both aren't assigned yet also 14:18:41 <slaweq> other than that, all good last week 14:18:47 <ralonsoh> I can ping OVN folks 14:18:51 <ralonsoh> for those bugs 14:19:13 <slaweq> ralonsoh: yeah, maybe lucasagomes or jlibosva can take a look at them 14:19:25 <slaweq> both are actually related to each other 14:19:58 <lajoskatona> I think frickler propsed those for on demand agenda 14:20:12 <opendevreview> Terry Wilson proposed openstack/networking-ovn stable/train: Fix race condition retrieving logical router rows https://review.opendev.org/c/openstack/networking-ovn/+/810820 14:20:27 <frickler> yes, I also added https://bugs.launchpad.net/neutron/+bug/1951074 which I created today which is also related 14:20:57 <slaweq> thx frickler 14:21:05 <frickler> not sure how much is bug vs. undocumented design choice 14:21:39 <lajoskatona> frickler: thanks, we can check them during the on-demand part, if it is ok for You 14:21:45 <frickler> sure 14:22:21 <lajoskatona> ok thanks 14:22:43 <lajoskatona> back to bug deputy cycle: this week hongbin is the deputy, and next week haleyb will be. 14:23:07 <lajoskatona> I wrote mail to hongbin he follow this week's bugs 14:23:20 <slaweq> please email haleyb about it as I'm not sure how much available he will be 14:23:30 <lajoskatona> slaweq: ok, I will 14:23:50 <lajoskatona> #topic L3 subteam 14:24:04 <bcafarel> late o/ (with irc back) 14:24:15 <lajoskatona> liuyulong is not on IRC, so I think we can skip it this week 14:24:26 <lajoskatona> Hi bcafarel 14:24:52 <lajoskatona> #topic ryu and os-ken 14:25:11 <lajoskatona> there was nothing merged to ryu 14:25:27 <ralonsoh> no, no new patches 14:25:33 <lajoskatona> so we can jump to the last topic of today 14:25:33 <ralonsoh> there are some bugs under review 14:25:41 <lajoskatona> thanks ralonsoh 14:25:51 <lajoskatona> in os-ken or in ryu? 14:25:57 <ralonsoh> ryu 14:26:24 <lajoskatona> ok, I have no notification for that :-) 14:26:47 <mlavalle> can I ask a silly question? 14:27:02 <lajoskatona> ralonsoh: Do you think that we have to follow some of those? 14:27:05 <lajoskatona> malavalle: sure 14:27:19 <lajoskatona> there's no silly question 14:27:21 <mlavalle> why not just go back to ryu? 14:27:28 <ralonsoh> lajoskatona, yes, I'm doing it 14:27:45 <mlavalle> instead of moving all these patches around 14:28:03 <ralonsoh> ryu stopped providing library compatibility with openstack 14:28:05 <lajoskatona> ralonsoh: thanks 14:28:25 <lajoskatona> mlavalle: we discussed it during the PTG: https://etherpad.opendev.org/p/neutron-yoga-ptg#L232 14:28:46 <lajoskatona> and yeah it was among the reasons that ryu removed some openstack compatiblity code 14:29:12 <mlavalle> thanks 14:29:16 <lajoskatona> and of course other things like different process, no control on what and when to merge, release.... 14:29:36 <lajoskatona> mlavalle: yw 14:29:44 <slaweq> also we discussed that it would be better to have os-ken tested together with Neutron in the OpenStack CI 14:30:47 <lajoskatona> ok, our last topic for today 14:30:50 <lajoskatona> #topic On Demand Agenda 14:31:20 <lajoskatona> so the topic from frickler: Bugs/Gaps with OVN and DNS resolution 14:31:59 <frickler> it all started out when a colleague came to me with OVN deployed via kolla and DNS resolution wasn't working in instances 14:32:31 <frickler> and I saw that the instance had "nameserver 127.0.0.53" in /etc/resolv.conf 14:32:45 <frickler> which is normal for ubuntu and others, but not for cirros 14:32:56 <frickler> and certainly doesn't work there 14:33:54 <frickler> devstack sets the ovn.dns_servers variable, so the issue isn't seen there 14:34:46 <frickler> that's https://bugs.launchpad.net/neutron/+bug/1951074 which is potentially a security issue, if nameserver setup is leaked from the host to instances without the operator being aware 14:35:20 <frickler> and then I found the other issues that njohnston had created like a year ago, but haven't seen much attention yet 14:36:02 <frickler> like the whole documentation at https://docs.openstack.org/neutron/latest/admin/config-dns-res.html doesn't work with OVN 14:36:47 <frickler> because the variables are different and also ovn cannot use host networking 14:37:36 <frickler> so I wonder what the plan with all this is, because for me as an operator, OVN isn't useable with the current state 14:38:48 <slaweq> we will need to work to close those gaps for sure :) 14:39:01 <lajoskatona> so few of these issues are more documentation issues am I right? 14:40:16 <frickler> some certainly are, the question is whether e.g. the issue with nameserver=0.0.0.0 would just be documented as known, permanent gap, or whether it can be fixed 14:40:49 <ralonsoh> I'll raise this issue in out internal bug triage meetings 14:40:54 <ralonsoh> our* 14:41:02 <slaweq> that probably has to be checked with jlibosva or lucasagomes as they are our ovn experts 14:41:09 <slaweq> maybe also dalvarez could help 14:41:11 <jlibosva> o/ 14:41:14 <lajoskatona> thanks ralonsoh & slaweq 14:42:31 <frickler> o.k., so let me know if there's any progress or some questions arise 14:43:04 <lajoskatona> frickler: thanks for the headsup 14:43:08 <jlibosva> frickler: is that bug just about documentation or you want to change the defaults? 14:43:36 <frickler> jlibosva: which bug? there's four of them currently 14:44:02 <jlibosva> ah, sorry - I just read the few lines above since the topic has changed - I looked at https://bugs.launchpad.net/neutron/+bug/1951074 14:44:02 <lajoskatona> list is here: https://wiki.openstack.org/wiki/Network/Meetings#On_Demand_Agenda 14:44:20 <jlibosva> gotcha, I'll walk over those 14:44:39 <lajoskatona> jlibosva: thanks for checking 14:44:54 <frickler> jlibosva: for that, I would suggest to change the way ovn is operating, because telling instances to use 127.0.0.53 as nameserver just doesn't make sense 14:48:08 <frickler> anyway, we can discuss later, don't need to hold up the meeting now, I guess 14:48:17 <lajoskatona> frickler: agree 14:48:36 <lajoskatona> thanks for bringing this to the meeting 14:48:53 <lajoskatona> if nothing more for the on demand agenda, we can close for today 14:49:12 <slaweq> ++ 14:49:34 <jlibosva> ack, thanks :) 14:50:23 <lajoskatona> ok, let's finish the meeting for today, thanks for attending 14:50:27 <ralonsoh> bye 14:50:28 <lajoskatona> #endmeeting