15:00:27 #startmeeting kolla 15:00:28 Meeting started Wed May 13 15:00:27 2020 UTC and is due to finish in 60 minutes. The chair is mgoddard. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:29 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:32 The meeting name has been set to 'kolla' 15:00:32 #topic rollcall 15:00:35 \o 15:01:56 o/ 15:02:10 ~o 15:02:19 mgoddard mnasiadka hrw egonzalez yoctozepto rafaelweingartne cosmicsound osmanlicilegi 15:03:28 o/ 15:03:48 #topic agenda 15:03:56 * Roll-call 15:03:57 * Announcements 15:03:59 ** OpenStack Ussuri released! https://www.openstack.org/software/ussuri 15:04:01 * Review action items from last meeting 15:04:03 * CI status 15:04:05 * Ussuri release planning (kayobe) 15:04:07 * Ussuri release planning (kolla & kolla ansible) 15:04:09 * Kolla Klub https://etherpad.openstack.org/p/kolla-sig 15:04:11 #topic announcements 15:04:22 #info OpenStack Ussuri released! (but not kolla yet...) 15:04:51 Kolla is a cycle-trailing project, so we now enter the time of confusion where Ussuri is released but Kolla is not 15:05:02 #link https://www.openstack.org/software/ussuri 15:05:14 congrats headphoneJames on making the front page ^ 15:05:22 (via backend TLS) 15:05:46 team effort ;) 15:05:54 o/ 15:06:19 #topic Review action items from last meeting 15:06:46 I hope that this time we will release sooner than later ;D 15:06:51 None 15:07:15 #topic CI status 15:07:35 Merged openstack/kolla-ansible master: Fix hacking min version to 3.0.1 https://review.opendev.org/727502 15:07:38 Ah yeah 15:07:58 #info Ansible 2.9.8 breaks kolla-ansible master branch 15:08:23 #info Ansible 2.9.9 released! 15:08:25 #link https://pypi.org/project/ansible/2.9.9/ 15:08:40 nice 15:08:45 with fix? 15:08:50 yes 15:09:02 My second ansible bug & fix to trigger a day 2 release :) 15:09:28 seems kolla-ansible is a better test of ansible than ansible's CI 15:09:55 ok CI 15:10:15 mgoddard: nice :-) 15:10:20 kolla & kolla-ansible lookin' green 15:10:40 Merged openstack/kolla master: Fix hacking min version to 3.0.1 https://review.opendev.org/727501 15:10:48 kolla-ansible is a test project for ansible. :-) 15:11:15 we are doing more ansible than ansible does ansible :-) 15:11:42 Merged openstack/kolla-cli master: Fix hacking min version to 3.0.1 https://review.opendev.org/727503 15:11:50 kayobe CI got hit by a zuul role name rename 15:12:01 backports in progress 15:12:48 #topic Ussuri release planning (kayobe) 15:13:00 priteau 15:14:18 https://tiny.cc/kayobe-review-dashboard2 15:14:47 o/ 15:15:02 hi all 15:15:20 hi 15:15:26 quite a few backports open 15:16:03 Any chance of getting mnasiadka's custom seed containers into ussuri? 15:16:03 Looks like I need to refresh the c8 docs patch 15:16:12 That depends on mnasiadka :) 15:16:27 I went over open patches recently and reviewed almost all of the ready ones 15:16:29 I've started using it in places you see :D 15:16:48 jovial[m1: I'll work on it today - specially for you ;) 15:17:01 I would like to see the seed container support too :) 15:17:29 high demand feature 15:17:35 thanks mnasiadka - It seemed to work well in my testing - very useful feature 15:17:41 Quick question: should we keep the six removal patch for just before branching, like the rest of py2 removal? 15:17:43 jovial[m1: I can't see LUKS. Did it land? 15:17:51 priteau: +1 15:17:56 my fix didn't work :( 15:18:08 LUKS doesn't pass CI 15:18:16 hmm 15:18:17 https://review.opendev.org/#/c/720040/ 15:18:18 patch 720040 - kayobe - Add support for encrypted block devices - 11 patch sets 15:18:22 I have to investigate what we are missing in the generic cloud image 15:18:37 works as expected in vagrant 15:18:37 misses criteria to make the dashboard 15:19:15 "Could not find the requested service systemd-cryptsetup@loopback-crypt.service: host" 15:19:49 jovial[m1: I'd suggest pointing requirements.yml to a branch of the role and iterating rather than releasing and hoping for the best :) 15:20:19 or (better) test locally. sausage has cloud images 15:20:21 good call - I was just so sure I squelched it :) 15:21:23 mgoddard: The reason I haven't approved "Constraint python-openstackclient install with requirements.txt" backport is because we discussed fixing it differently in master, by using a different venv 15:21:45 ok 15:22:44 jovial[m1: what about https://review.opendev.org/#/c/724818/ ? 15:22:45 patch 724818 - kayobe - Add ability to run playbooks before and after a ka... - 7 patch sets 15:22:46 will it land? 15:23:21 I noticed the six patch conflicts with the centos 8 support one - which one should we merge first? 15:23:54 mgoddard: I put a note about it being experimental and subject to change in the README 15:24:13 does that give use clearance to merge and improve on it later? 15:24:56 ok. looks like it needs another review 15:25:04 on my list 15:25:38 I haven't looked at hooks patch yet, I was excluding patches that didn't pass CI 15:25:50 only docs fail 15:25:55 I've just seen that it's only docs failure 15:26:23 I can review it probably only on Friday 15:26:34 ok 15:27:14 I'll mark some patches with RP+1 15:27:25 I think we need another push 15:27:33 Anything else for Kayobe? 15:28:06 Not from me 15:28:40 Except this patches that presumably fix CI on stable branches: https://review.opendev.org/#/c/727684/ 15:28:40 patch 727684 - kayobe (stable/train) - Use ensure-docker role instead of install-docker - 1 patch set 15:28:48 s/this/these/ 15:29:07 yes, approvals for those please 15:29:17 We haven't run any CI job on stable branches recently but I expect it is broken 15:29:34 #topic Ussuri release planning (kolla & kolla ansible) 15:29:58 I suppose we ought to start working through the release process 15:31:12 #link https://docs.openstack.org/kolla/latest/contributor/release-management.html 15:31:43 e.g. switch to released tarballs 15:32:27 Does anyone want to pick that up? 15:32:28 mgoddard: before or after branching? 15:32:33 hrw: before 15:32:42 will prepare patch 15:32:45 all in the doc above :) 15:33:55 #action hrw to switch source tarballs to ussuri final releases 15:34:29 What do we have as FFEs? 15:34:33 Backend TLS? 15:34:46 hopefully :) 15:35:24 we may consider drop of gnocchi btw 15:35:29 may we? 15:35:45 I think our ceilometer users may be a bit annoyed 15:35:49 last release July 2019 15:35:59 AFAIK it still works fine 15:36:04 \o hi 15:36:10 and nearly no trace of activity 15:36:13 sorry was in a job meeting 15:36:27 hi Fl1nt 15:36:45 I think we can keep gnocchi and reconsider if it starts breaking on us 15:36:54 ceilometer may benefit from it but it can use another backend 15:37:32 not in an existing cloud 15:37:35 will write comments to tarballs patch 15:38:00 ok 15:38:10 It's possible to use Monasca as Ceilometer backend now, but that's still fairly new. I don't see why Gnocchi should be dropped as long as it is the recommended backend for Ceilometer. 15:39:06 yep definitely, that's not because there are no activity that it means it doesn't work anymore 15:39:07 At the very least, deprecate it in one release and drop it only in the next one 15:39:53 let's see how it evolves. There still seem to be people interested in using it. Maybe they will also maintain it 15:41:05 interesting (old but relevant) article from Julien that would explain a lot: https://julien.danjou.info/lessons-from-openstack-telemetry-deflation/ seems like people are transitioning from ceilo to prometheus etc 15:41:43 Fl1nt: no wonder as all telemetry components seem to be low-activity 15:41:52 a bit sad though 15:42:06 At this point in the release people should be switching to testing the new release and fixing bugs 15:42:08 yep, Julien explain it pretty well on its article and especially why 15:42:12 still, maybe this aspect is better handled at more general level 15:42:34 fixing bugs is indeed what is on my radar :-) 15:43:01 I'm on sync with you @priteau and @mgoddard we should keep it for compatibility until it really break something. 15:43:41 Do we have anything else to discuss for Ussuri? 15:44:09 Fl1nt: ++ 15:44:19 mgoddard: rather not, maybe only tls 15:44:25 are we getting tls for neutron? :D 15:44:35 it seems it's the major lacking component 15:44:51 it doesn't have apache 15:45:11 mgoddard: ah, so it counts as pitb 15:45:14 I can take a look shortly 15:45:16 yes 15:45:29 headphoneJames: I would prefer to focus on finishing what we have 15:45:34 headphoneJames: thanks, don't sweat it though 15:45:39 indeed ^^ 15:45:40 i.e. https://etherpad.opendev.org/p/tls_tasks 15:45:57 ~curse rally 15:46:53 @hrw, why? It work pretty well. 15:46:57 would the world end if we used native neutron-server & TLS? 15:47:02 https://docs.openstack.org/neutron/latest/configuration/neutron.html 15:47:09 certfile/keyfile 15:47:26 could always switch to apache later, but would be a simple change to complete the core set 15:47:28 what do you mean by native neutron server? 15:48:22 at the moment we use the python web server 15:48:25 no httpd 15:48:25 Marcin Juszkiewicz proposed openstack/kolla master: Switch to Ussuri release tarballs https://review.opendev.org/727810 15:48:41 not sure which library it uses under the hood 15:48:44 oh ok 15:48:53 anyways, it's not top of the list 15:49:16 I added it to the bottom 15:49:17 As for the docker philosophy, I would use neutron-server native as it's already a wsgi app 15:49:18 Fl1nt: rally or rally-openstack? ;D 15:49:33 rally and rally-openstack 15:50:06 adding another process within the container would make it a good candidate for PODs rather than docker actually 15:50:23 Fl1nt: we use httpd for many of our APIs 15:50:34 even if I know that there are many containers already usign httpd ^^ 15:50:54 better performance & reliability 15:51:18 Fl1nt: rally does not use openstack/releases at all 15:51:21 #topic Kolla Klub https://etherpad.openstack.org/p/kolla-sig 15:51:26 @mgoddard, yeah, and that's actually the best way to decide, does it add value by improving performance and stability? Then it's OK ^^ 15:51:43 #link https://docs.google.com/document/d/1EwQs2GXF-EvJZamEx9vQAOSDB5tCjsDCJyHQN5_4_Sw/edit# 15:51:51 We have no topic for the next meeting 15:52:04 (21st May) 15:52:09 What should we discuss? 15:52:25 I've had a quick side chat 15:52:31 Marcin Juszkiewicz proposed openstack/kolla master: Switch to Ussuri release tarballs https://review.opendev.org/727810 15:52:33 with an attendee of the last meeting 15:53:31 and it seems folks are interested by how is kolla/kolla-ansible/kayobe team can actually simplify code contribution as it's an easier step to climb. 15:53:42 like 15:54:14 they asked a lot about something like a clean and easily discoverable onboarding page. 15:54:24 and of course, communication channels 15:54:44 we have a community goal this cycle to improve the contributor page 15:55:06 https://review.opendev.org/707800 15:55:06 patch 707800 - kolla - [community goal]: Add contributor and PTL guide - 2 patch sets 15:55:14 Reviews appreciated 15:55:38 Also, the first topic proposed for kolla klub is 'project onboarding' 15:56:03 yep, I think that the process is pretty straight forward if you carefully read all openstack related documentation, but it yet lack a lot of discoverability 15:56:28 and the fact that the way to contribute to openstack projects differs a lot from any other GIT hosted projects is kind of a pain point for them. 15:56:39 "a lot" 15:56:46 we can't change that 15:56:53 yep I know ^^ 15:57:06 we can improve our docs and discoverability though 15:57:18 it's not that difficult honestly and maybe that entry level kind of filter a lot of bad things too. 15:57:22 if you have ideas, please propose or draft up 15:57:30 yep 15:57:55 (after reviewing the proposed patch, since that is what the future looks like) 15:58:30 mgoddard: I would be glad if somebody could take the burden of improving contributor and PTL guide :) 15:58:38 (from my hands) 15:58:42 I'll have a look at that doc topic as I'll have to write a bunch of. 15:58:48 I think often though the problem is 'I simply don't know where to start or what to search for' 15:59:09 yep, that what I call discoverability 15:59:33 Yes, what I am saying is that if you don't know what to ask for, I don't know how to help :) 15:59:43 I think that starting with 'I have a problem with service XY' is good way 16:00:07 but that point is coming from two factors too, 1 is the openstack doc site being a pain of UI/UX and Search engine such as google still referencing really old docs as first answers. 16:00:33 yes, SEO is a problem. It was discussed on ML recently 16:00:45 it is discussed again and again 16:00:54 and will be 16:00:54 we're out of time 16:01:07 Shall I propose an onboarding session for the klub next week? 16:01:18 yep, that would be cool actually 16:01:39 ok 16:01:58 I have slides. Anyone want to help present? 16:02:16 I can 16:02:20 thanks 16:02:23 I did onboard 16:02:36 #action mgoddard and yoctozepto to sync up on onboarding for kolla klub 16:02:45 I'll send an email with the topic 16:02:51 ok 16:02:51 That's all folks 16:02:59 #endmeeting