15:00:18 #startmeeting kolla 15:00:19 Meeting started Wed Apr 22 15:00:18 2020 UTC and is due to finish in 60 minutes. The chair is mgoddard. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:20 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:23 The meeting name has been set to 'kolla' 15:01:13 #topic rollcall 15:01:16 \o 15:01:17 o/ 15:01:34 noob question, what's rollcall ? 15:01:48 ~o~ 15:01:54 Fl1nt: 'who is here' 15:02:03 https://en.wikipedia.org/wiki/Roll_call 15:02:05 oh ok 15:02:09 o/ 15:02:11 "A taking of attendance as part of a meeting agenda" 15:02:12 so ^^ 15:02:14 o/ 15:03:04 #topic announcements 15:03:17 #undo 15:03:18 o/ 15:03:18 Removing item from minutes: #topic announcements 15:03:21 #topic agenda 15:03:33 * Roll-call 15:03:35 * Announcements 15:03:37 ** Proposed cross-project goal for Victoria: https://review.opendev.org/720107 15:03:37 patch 720107 - governance - goals: add container-images - 3 patch sets 15:03:39 ** Kolla builder next generation proposal: http://lists.openstack.org/pipermail/openstack-discuss/2020-April/014255.html 15:03:41 * Review action items from last meeting 15:03:43 * CI status 15:03:45 * Kolla Klub https://etherpad.openstack.org/p/kolla-sig 15:03:47 ** Survey tools comparison https://ethercalc.openstack.org/46c7q4lnww6j 15:03:49 * Ussuri release planning (kayobe) 15:03:51 * Ussuri release planning (kolla & kolla ansible) 15:03:53 * Virtual PTG 15:03:55 * Train release planning (centos 8) 15:03:57 * Announcements 15:03:59 #topic announcements 15:04:09 #info Proposed cross-project goal for Victoria on container images 15:04:16 #link https://review.opendev.org/720107 15:04:16 patch 720107 - governance - goals: add container-images - 3 patch sets 15:05:10 Seems there could be some interesting outcomes from this 15:05:24 And some opening up to the potential of using kolla 15:05:25 so this is mnaser's proposal that we could participate in as we have some relevant experience 15:05:52 bindep is interesting to finally get nice 15:06:15 I decided to limit myself to read comments 15:06:21 right, although it's a long way from magically installing all your deps IMO 15:06:26 actually sean-k-mooney has just written up something relevant to bindep on the ml, regarding the next announcement 15:06:35 (as well) 15:06:48 #info Kolla builder next generation proposal 15:06:54 #link http://lists.openstack.org/pipermail/openstack-discuss/2020-April/014255.html 15:07:12 Thanks yoctozepto for starting this conversation. I encourage people to get involved 15:08:27 mgoddard: sure, I think we should do these kinds of things on ml from time to time, if not just to raise community awareness 15:08:40 #info OpenStack projects proposing RC1 releases this week 15:08:53 (not kolla!) 15:09:02 o/ 15:09:05 #info Kolla feature freeze next week 15:09:17 :O 15:09:19 yup 15:09:21 :D 15:09:26 let me check what I have open... 15:09:31 ok, that's enough #info 15:09:49 Review action items from last meeting 15:09:53 * Fl1nt reading 15:09:53 #topic Review action items from last meeting 15:10:10 mgoddard to prepare project update for kolla klub 15:10:13 osmanlicilegi to adapt popularity-contest for kolla needs 15:10:15 osmanlicilegi to look into kolla klub survey/poll 15:10:17 mgoddard to read PTG emails and request slots 15:10:19 hrw to switch kolla centos 8 train from delorean to centos-release-openstack-train 15:10:22 we will find out if I have done mine tomorrow 15:10:46 I haven't requested PTG slots yet, but did put a proposal on the mailing list 15:10:59 https://review.opendev.org/721329 does my part. We are waiting for centos-release-* packages (on a way). 15:10:59 patch 721329 - kolla (stable/train) - [WIP] switch to official CentOS 8 repositories - 3 patch sets 15:11:11 #link http://lists.openstack.org/pipermail/openstack-discuss/2020-April/014174.html 15:11:12 all images build fine. 15:11:25 thanks 15:11:42 osmanlicilegi did his, but can't attend the meeting 15:11:50 yeah, thanks osmanlicilegi! 15:11:54 #action mgoddard to sign up for PTG slots 15:11:56 nice summary there 15:12:10 16:02 < osmanlicilegi> mgoddard: I have to skip today's meeting cause there will a lockdown here for the next 4 days. I need to visit pharmacy and market. You can find my survey tools comparison at https://ethercalc.openstack.org/46c7q4lnww6j. If there's anything missing for this task, just ping me. 15:12:16 #link https://ethercalc.openstack.org/46c7q4lnww6j 15:12:38 #topic CI status 15:13:02 c8/aarch64 is RED due to lack of erlang 15:13:11 James Kirsch proposed openstack/kolla master: Add stunnel container for Glance service https://review.opendev.org/720850 15:13:16 both train and ussuri. we have to wait for CentOS 8.2 ;( 15:14:00 otherwise pretty green 15:14:05 hrw: can you mark the rabbitmq image as unbuildable until then so we're green? 15:14:19 then at least we can catch regressions 15:14:22 mgoddard: kolla-toolbox is red too :( 15:14:24 mgoddard: ok 15:15:02 #action hrw mark rabbitmq and kolla-toolbox unbuildable for c8/aarch64 15:15:20 #topic Kolla Klub https://etherpad.openstack.org/p/kolla-sig 15:15:39 Meeting tomorrow, 1500 UTC 15:16:13 I'll give a project update, and we can feed back about surveys 15:16:25 #topic Survey tools comparison https://ethercalc.openstack.org/46c7q4lnww6j 15:16:29 We should also 15:16:37 it's interesting to learn details on kolla usage too 15:17:05 indeed. I think some case studies would be nice 15:17:06 have a little agenda for what we as a community expect from this klub 15:17:13 ideally I won't be presenting every session :) 15:17:27 mgoddard: that would be regression! 15:18:06 Fl1nt: yup, that discussion should be part of agenda 15:18:14 the homework was to think about it 15:18:17 Fl1nt: I did try to ask last time, but didn't get too much response 15:18:41 Marcin Juszkiewicz proposed openstack/kolla master: No kolla-toolbox & rabbitmq for aarch64/centos8 https://review.opendev.org/721999 15:18:42 Marcin Juszkiewicz proposed openstack/kolla master: Revert "No kolla-toolbox & rabbitmq for aarch64/centos8" https://review.opendev.org/722000 15:18:45 feedback seemed to be 'not too much naval gazing - just get on with it' 15:19:05 yep, unfortunately, I think, on my own part, that many operators wait for a more vivid community in terms of support channels. 15:19:48 Agenda is here, feel free to add something: https://docs.google.com/document/d/1EwQs2GXF-EvJZamEx9vQAOSDB5tCjsDCJyHQN5_4_Sw/edit#heading=h.u784zrqsv2k1 15:20:08 ok, I'll think about it. 15:20:25 if it's something you want to drive, that might encourage others to get involved 15:20:44 ok, let's look at osmanlicilegi's results, since he put time into it 15:20:57 https://ethercalc.openstack.org/46c7q4lnww6j 15:21:42 at the end of the day, google forms look the best 15:21:49 kolla klub agena item 1: 15:21:52 Survey tools survey results 15:21:55 but I guess China cannot access them 15:23:04 yeah that's a shame 15:23:39 yeah, such constraints definitely slowdown collaboration. 15:23:56 typeform seems nice, apart from the 10 question limit 15:24:22 Can we achieve the survey in 10 questions? 15:24:58 It seems a bit low 15:25:26 but maybe it fits 15:25:37 Maybe the trick is to create multiple survey and explain that multiple strategy has we want to include and ease the discussion with everyone? So no one will complain about having multiple poll. 15:25:38 we also need to consider multianswer and freeform parts 15:25:49 +1 15:25:59 osmanlicilegi has some multi answer 15:26:06 let's extend the calc for that 15:26:10 I don't see freeform though 15:26:17 IMO that would be one of the most valuable parts 15:26:19 Fl1nt: the problem is analizing results 15:27:14 hum... 45€ for the professional plan, can't we pay for it ? 15:27:28 One thing I worry about with these surveys is how discoverable the data will be 15:27:49 that's the killer feature of Google 15:27:56 I think at a minimum we need data export 15:28:04 but also we need to be able to host that somewhere 15:28:33 it would be nice to learn about osf's solution 15:28:42 +1 15:29:16 I see a few uses for the results 15:29:30 1. give the team a better picture of what people do and don't use 15:29:30 you can connect typeform anserws to google sheets 15:29:54 2. allow to connect up questions to sources of knowledge 15:30:14 a survey is probably quite good at 1., I don't know how well it serves 2 15:30:31 mgoddard: could you elaborate on 2.? 15:31:21 i.e. user comes in asking about storlets + kolla, who do we connect them to? 15:31:54 maybe I'm just thinking too siloed, and we should send them to #openstack-storlets 15:32:14 almost what I want is a social network :) 15:32:19 linkedin for openstack 15:32:30 mgoddard: I was about to say "linkedin for kolla" 15:32:32 but yeah 15:32:43 maybe it's a pipe dream 15:33:04 you already know I'm a dreamer too, so you have my full support 15:33:29 I'd just like to avoid the problem that questions often go unanswered because not all experts read IRC, and when we send people to $project-channel it's often to /dev/null 15:33:49 yeah, especially those least used ones 15:33:59 yeah, exactly my point of why I suggested having discord, let's face it, nobody except old school devs/operators like us will use IRC, new comers to Openstack tends to be rather younger and have different tools/usage to communicate. 15:34:08 * osmanlicilegi is back after a quick shopping 15:34:18 osmanlicilegi: quick, huh? ;D 15:34:26 probably after cilegi ;-) 15:34:43 Fl1nt: it's not just about the medium though, most people just don't want to consume the firehose 15:34:54 honestly, kolla channel is for sure the most active one. 15:35:15 yoctozepto: ;] 15:35:18 sure, it was an exemple to support the fact that this community can't be based on either IRC or List. 15:35:19 maybe the answer is we should direct people to the mailing list more often 15:35:35 at least there people filter out threads 15:35:36 just let's not talk ask.oo, it s... is not decent enough :-) 15:35:51 indeed 15:36:06 anyways, in terms of surveys 15:36:16 google forms is best, but limited 15:36:17 if you ask you'll get 10000 answers of the form to use, if you build it and propose it, you'll get adoption if it fit. 15:36:36 is there a runner up to forms? 15:37:01 typeform? apart from 10 question limit 15:37:10 yep typeform 15:37:19 osmanlicilegi: did any surveys offer freeform text? 15:37:27 it feels tedious to fill in 15:37:36 and has 100 answers limit monthly 15:37:47 typeform and surveymonkey.com are the both big names in that industry 15:37:55 outside of google form I mean 15:38:00 mgoddard: iirc all has freeform text support 15:38:03 I think 100 answers per month would be plenty yoctozepto! 15:38:23 mgoddard: be a dreamer! ;p 15:38:28 btw, typeform seems best after google forms if 10 questions is ok 15:39:05 if 10 questions is your blocker, I'll subvention the typeform if necessary. 15:39:06 seems so 15:39:13 I don't like one q per page though 15:39:31 I think we should move on :) 15:39:35 ++ 15:39:49 #topic Ussuri release planning (kayobe) 15:39:50 yep 15:39:56 priteau, dougsz, jovial[m] 15:39:58 yoctozepto: typeform is customizable. one question per page can be my fault 15:40:11 osmanlicilegi: would you drop a question to ml and osf about their framework for surveys? 15:40:15 If you didn't catch it earlier, feature freeze next week 15:40:17 maybe someone responds 15:40:19 osmanlicilegi: ack 15:40:25 (now onto kayobe) 15:40:25 yoctozepto: I will, today. 15:40:40 https://review.opendev.org/#/q/project:openstack/kayobe+branch:master+status:open 15:40:47 mgoddard: thanks for the reminder 15:40:57 Lots of open patches :o 15:41:08 yep 15:41:19 quite a few +2'd or approved though 15:41:52 mnasiadka: are you planning to update https://review.opendev.org/#/c/709689/ to answer Mark's comments? 15:41:52 patch 709689 - kayobe - Add framework to deploy user-defined containers on... - 11 patch sets 15:42:13 if anyone has time then adding a second +2 would help clear some 15:42:54 would be nice to get https://review.opendev.org/#/c/658815/ in but I don't know if I have time to test the migration 15:42:54 patch 658815 - kayobe - WIP: Switch default docker storage driver to overlay2 - 2 patch sets 15:43:10 For Kayboe, should cloud_vip_address be set for localhost? I'm receiving the error that kolla_internal_vip_address is not set for localhost. And if so, where's the right place to set that? 15:43:14 if someone else has time feel free to pick it up 15:43:25 dking: we're just in a meeting at the moment 15:43:37 dking: but you can set it in networks.yml 15:43:51 Okay. Thank you. I'll probably try back in an hour or so, then. 15:43:51 dking: which is global for all hosts 15:44:26 any more for kayobe 15:44:29 ? 15:44:34 Merged openstack/kolla-ansible master: Fix that cyborg agent failed to start privsep daemon. https://review.opendev.org/721139 15:44:48 #topic Ussuri release planning (kolla & kolla ansible) 15:45:21 all good 15:45:48 Debian packages partially available 15:46:00 added RP+1 to https://review.opendev.org/#/c/721632/1 15:46:01 patch 721632 - kolla-ansible - Add docs and release note for CentOS 8 - 1 patch set 15:46:15 for CentOS I wait for centos-release-* packages and will add some of them to cover RDO needs 15:46:51 rabbitmq/erlang are now in queue: https://review.opendev.org/721800 then advanced-virtualization and ceph-nautilus will follow 15:46:51 patch 721800 - kolla - CentOS: use RabbitMQ and Erlang from official repos - 1 patch set 15:46:52 hrw: is that for train? 15:47:00 mgoddard: both train and ussuri 15:47:03 ok 15:47:28 mgoddard: only centos-release-openstack-{train,ussuri} will differ 15:48:28 we have a few tls backend patches in flight. Hopefully we can get them in 15:49:06 would be nice to do it for all services already using httpd, but getting late 15:49:08 I will add relnote to kolla/rmq patch for aarch64 to make it visible that c8/aarch64 is not part of initial release 15:49:44 ok, thanks 15:49:49 https://review.opendev.org/707800 15:49:49 patch 707800 - kolla - [community goal]: Add contributor and PTL guide - 2 patch sets 15:49:54 this needs some love 15:50:14 +1 15:50:29 although not affected by freeze 15:50:56 priteau: yes, tomorrow 15:51:50 #topic Virtual PTG 15:52:03 Do we have much to say on it? 15:52:27 We can gather topics in a few weeks I think 15:52:53 #topic Train release planning (centos 8) 15:52:57 as usual: this cycle summary, c8 summary, next cycle goals 15:53:11 train for C8 is splendid 15:53:27 we have all images buildable 15:53:28 I think this was just a request to make a release when train/c8 is ready? 15:53:33 yes 15:54:17 ok. If I backport the k-a docs/reno patch to train then I think we're ready for a release 15:54:49 Marcin Juszkiewicz proposed openstack/kolla master: No kolla-toolbox & rabbitmq for aarch64/centos8 https://review.opendev.org/721999 15:54:52 Merged openstack/kolla-ansible stable/train: Fix that cyborg conductor failed to communicate with placement https://review.opendev.org/721770 15:55:01 added relnote 15:55:02 it should at least work for greenfield deploys. I'll do 7-8 migration docs separately to not block a releaase 15:55:09 ok 15:55:11 thanks 15:55:17 #topic Open discussion 15:55:29 Documentation question 15:55:43 is rocky release considered as EOL ? 15:56:10 Fl1nt: extended maintenance 15:56:22 Fl1nt: we accept patches as lons as CI works (it works afaik) 15:56:33 but do not really propose ourselves etc. 15:56:40 because the k-a with external ceph needs some fixes, I'll be happy to write them, but it also need some fixes on the nova role as it doesn't work at this moment. 15:56:49 and I'd like to discuss https://bugs.launchpad.net/kolla-ansible/+bug/1874218 a bit 15:56:49 Launchpad bug 1874218 in kolla-ansible "swift sanity checks (smoke tests) broken" [Low,Triaged] 15:56:56 are we into sanity checks / smoke tests? 15:57:06 they look pooey now 15:57:18 James Kirsch proposed openstack/kolla master: Modify Heat and Nova containers to execute httpd setup https://review.opendev.org/722028 15:57:23 Fl1nt: only rocky? 15:57:42 Fl1nt: I am curious, what doesn't work with nova and external ceph? 15:57:46 maybe stein, but I didn't have HW/CI to test that release for now 15:57:55 @priteau, the keyring 15:57:58 because 15:58:06 with external ceph 15:58:19 you don't handle which ceph client/release you're using 15:58:21 and so 15:58:53 when using ceph.client.nova.keyring filename with rbd_user = cinder and inner key = ceph.client.cinder.keyring content 15:58:59 ceph client saw the trick 15:59:16 and complain about it and raise an error 16:00:22 mgoddard: no thoughts on smoke tests / sanity checks? 16:00:40 the only way to overcome that, is to upload the ceph.client.cinder.keyring on nova_compute and nova_libvirt container instead of ceph.client.nova.keyring. 16:00:48 or 16:01:05 to add a client directive within your ceph.conf like 16:01:13 [client.nova] 16:01:23 keyring = 16:01:35 which is... ugly for the least ^^ 16:01:48 exemple: https://opendev.org/openstack/kolla-ansible/src/tag/7.2.1/ansible/roles/nova/tasks/external_ceph.yml 16:02:14 but on train: https://opendev.org/openstack/kolla-ansible/commit/a9fecc41a31749c97e89ba6b66c9c46714fa6da8 16:02:25 Fl1nt: I would suggest talking to mnasiadka. He made quite a few changes to external ceph config in master 16:02:38 cool 16:02:55 we're out of time I'm afraid 16:02:59 Thanks all 16:03:02 #endmeeting