15:00:15 <yoctozepto> #startmeeting kolla 15:00:16 <openstack> Meeting started Wed Jan 8 15:00:15 2020 UTC and is due to finish in 60 minutes. The chair is yoctozepto. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:17 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:19 <openstack> The meeting name has been set to 'kolla' 15:00:29 <yoctozepto> #topic rollcall 15:00:33 <hrw> patch ^^ has to wait for https://review.opendev.org/701552 one 15:00:33 <patchbot> patch 701552 - opendev/system-config - UCA: mirror Ussuri packages - 1 patch set 15:00:34 <hrw> o/ 15:00:38 <yoctozepto> \o 15:00:41 <chason> o/ 15:01:02 <cosmicsound> o/ 15:02:01 <yoctozepto> mgoddard and mnasiadka have limited availability for today 15:02:05 <yoctozepto> so let's continue 15:02:21 <mnasiadka> o/ 15:02:31 <yoctozepto> #topic agenda 15:02:39 <yoctozepto> * Roll-call 15:02:39 <yoctozepto> * Announcements 15:02:40 <yoctozepto> ** Welcome Dincer Celik (osmanlicilegi) to kolla core team 15:02:40 <yoctozepto> ** Ceph deployment support will be removed in Ussuri 15:02:40 <yoctozepto> * Review action items from last meeting 15:02:40 <yoctozepto> * CI status 15:02:40 <yoctozepto> * Train release planning 15:02:41 <yoctozepto> * Ussuri release planning 15:02:41 <yoctozepto> * So long, Ceph deployment 15:02:42 <yoctozepto> * CentOS 8 migration 15:02:42 <yoctozepto> * CI scenarios 15:02:43 <yoctozepto> * unpublishing images, e.g. https://hub.docker.com/r/kolla/ubuntu-source-qdrouterd/tags 15:02:51 <yoctozepto> #topic announcements 15:03:01 <yoctozepto> Welcome Dincer Celik (osmanlicilegi) to kolla core team! 15:03:17 <yoctozepto> our newest addition to the best team around 8-) 15:03:20 <hrw> +2 15:03:34 <yoctozepto> osmanlicilegi is probably swimming now 15:03:49 <yoctozepto> so might join us later 15:04:02 <yoctozepto> still, sincere congratulations! 15:04:04 <openstackgerrit> Merged openstack/kolla-ansible stable/stein: Fix nova and cinder handlers with no ceph https://review.opendev.org/698680 15:04:05 <chason> osmanlicilegi, congras 15:04:15 <cosmicsound> congrats osmanlicilegi 15:04:31 <yoctozepto> Ceph deployment support will be removed in Ussuri 15:04:44 <yoctozepto> we kind-of knew that already 15:05:02 <yoctozepto> but let's keep that in mind we are forced to remove it atm due to c8 15:05:09 <yoctozepto> c8=centos8 15:05:28 <hrw> and we wanted to remove it for quite a while 15:05:30 <yoctozepto> mnasiadka working on better/tighter ceph-ansible 15:05:32 <cosmicsound> rhel8 in same category? 15:05:33 <yoctozepto> integration 15:05:34 <hrw> now have more excuses 15:05:44 <yoctozepto> hrw: exactly 15:05:57 <yoctozepto> cosmicsound: yes, unless explicitly noted otherwise rhel8=centos8 for kolla purposes 15:06:11 <cosmicsound> thanks for clarification 15:06:12 <yoctozepto> this applies to kolla-ansible 15:06:21 <yoctozepto> we will no longer *deploy* ceph cluster 15:06:35 <yoctozepto> but integration with an existing cluster is supported 15:06:35 <hrw> cosmicsound: the only user of rhel is tripleo - they only build templates 15:06:46 <cosmicsound> good to know hrw 15:06:47 <hrw> cosmicsound: at least that's how I prefer to think 15:06:51 <mnasiadka> Did we deprecate Ceph images in Kolla as well? Or only K-A part? 15:06:51 <yoctozepto> hrw, cosmicsound: ^ that we know of 15:07:16 <yoctozepto> mnasiadka: not sure on that, c8 forces removal of centos parts 15:07:32 <yoctozepto> we could keep ubuntu/debian lying around for another release if need be 15:07:40 <yoctozepto> but there are not users outside k-a there 15:07:43 <yoctozepto> no known* 15:07:55 <kplant> does this mean ceph-ansible will not be a supported role in kolla-ansible? 15:08:20 <kplant> or only referring to legacy ceph deployment via k-a 15:08:35 <yoctozepto> kplant: legacy 15:08:48 <mnasiadka> We are removing legacy ceph in k-a 15:08:49 <yoctozepto> k-a will stop deploying ceph cluster in ussuri release 15:08:53 <yoctozepto> it has been deprecated 15:08:57 <yoctozepto> and is being removed 15:09:02 <mnasiadka> Only external ceph will be supported 15:09:05 <yoctozepto> ceph external worked and will work 15:09:10 <yoctozepto> and we support that 15:09:16 <yoctozepto> and we <3 ceph 15:09:19 <mnasiadka> No ceph-ansible or other deployer integration 15:09:20 <yoctozepto> etc. 15:09:25 <kplant> right, what i mean specifically is that there was an effort to integrate ceph-ansible 15:09:28 <kplant> that's been abandoned? 15:09:32 <yoctozepto> just stop deploying/upgrading cluster part 15:09:44 <yoctozepto> kplant: mnasiadka working on that exactly 15:09:47 <yoctozepto> I mentioned that above 15:09:49 <cosmicsound> seen also discussion about replacing k-a recipe 15:09:52 <mnasiadka> kplant: there was, but ceph-ansible doesn’t have a future 15:10:07 <cosmicsound> makes sense mnasiadka then 15:10:10 <kplant> ouch 15:10:11 <yoctozepto> yes, the integration will not be tight 15:10:20 <yoctozepto> just enough to follow instructions 15:10:56 <yoctozepto> let's move on 15:11:06 <yoctozepto> #topic Review action items from last meeting 15:11:17 <yoctozepto> mnasiadka to remove priority from inactive blueprints 15:11:24 <yoctozepto> that's the only one 15:11:29 <yoctozepto> mnasiadka: done? 15:11:42 <mnasiadka> Partly, will finish this week 15:11:58 <yoctozepto> #action mnasiadka to continue removing priority from inactive blueprints 15:12:22 <yoctozepto> #topic CI status 15:12:38 <yoctozepto> we had this horizon failure on all the branches 15:12:46 <hrw> done 15:12:48 <yoctozepto> but it has been fixed 15:13:02 <yoctozepto> thanks hrw, mnasiadka (and myself) for handling 15:13:04 <hrw> aarch64 publishing waits in queue 15:13:23 <hrw> this time done properly 15:13:43 <hrw> can not be tested on gerrit so we need to merge it, observe and revert if broken 15:14:21 <yoctozepto> k green, k-a green, k-cli no recent jobs 15:14:35 <hrw> kayboe? 15:14:43 <yoctozepto> hrw: yes, I saw that, flaky but no better way for now 15:14:49 <yoctozepto> kayobe relatively green 15:15:25 <hrw> https://review.opendev.org/#/c/701011/ waits for fresh votes 15:15:25 <patchbot> patch 701011 - kolla - Add AArch64 publisher for Debian/source - 12 patch sets 15:15:41 <yoctozepto> hrw: after meeting :-) 15:15:54 <yoctozepto> #topic Train release planning 15:16:07 <yoctozepto> so we still have a few pending issue to fix on our recent release 15:16:18 <yoctozepto> most notably the check for ansible being installed properly along k-a 15:16:38 <yoctozepto> also, recently people again use too new k-a to deploy older k 15:16:51 <yoctozepto> (due to pip install kolla-ansible giving them train now) 15:17:01 <yoctozepto> so it might be worth looking into 15:17:36 <yoctozepto> I would include the mariadb deployment fix pack, and possibly these two, and then release 15:17:49 <yoctozepto> recently we got ipv6 fixed for centos7 courtesy of 99cloud 15:17:56 <yoctozepto> (incl. Jeffrey4l) 15:18:11 <yoctozepto> again, great job, folks 15:18:29 <yoctozepto> anyone anything regarding train? 15:18:32 <yoctozepto> hrw, mnasiadka 15:19:02 <hrw> not that I know 15:19:13 <mnasiadka> Still working on the fernet fix, but it doesn’t show up often 15:19:31 <yoctozepto> mnasiadka: and yeah, fernet if we get that reviewed, but we probably will 15:19:55 <yoctozepto> #topic Ussuri release planning 15:20:03 <yoctozepto> how's ussuri? 15:20:27 <hrw> c8 is blocked but there are patches for reviews 15:20:50 <yoctozepto> ah yes, many pending for me 15:20:54 <hrw> centos team says that SIGs may get buildroots after 8.1 release finally done 15:21:10 <yoctozepto> nice, hope it happens soon enough 15:21:17 <hrw> ubuntu got patch to move to UCA/ussuri 15:21:25 <yoctozepto> I see mnasiadka having progress on ovn 15:21:27 <hrw> infra got patch to start mirroring UCA/ussuri as well 15:21:47 <yoctozepto> there is also some moves in the TLS zone, in need of reviewing 15:22:00 <yoctozepto> hrw: uca +1 15:22:08 <mnasiadka> yoctozepto: yeah, need to polish testing (core-openstack) 15:22:35 <yoctozepto> mnasiadka: does it break the normal procedures? 15:23:04 <hrw> Polish testing... start to be afraid 15:23:37 <yoctozepto> hrw: shibboleth? 15:24:11 <mnasiadka> yoctozepto: no, have to adapt init-runonce to use geneve instead vzlan only on ovn scenario 15:24:18 <yoctozepto> chrząszcz brzmi w trzcinie w Szczebrzeszynie, i Szczebrzeszyn z tego słynie :-) 15:24:34 <mnasiadka> yoctozepto: something like that :) 15:24:36 <yoctozepto> mnasiadka: ah, it's geneve finally somewhere 15:24:49 <hrw> mnasiadka: init-runonce... it waits for years to be playbooked... 15:25:00 <yoctozepto> hrw: it's optional 15:26:16 <mgoddard> o/ 15:26:28 <yoctozepto> we welcome our PTL, mr mgoddard 15:26:29 <mgoddard> sorry for lateness, busy with a client 15:26:49 <mgoddard> (still busy :)) 15:27:10 <yoctozepto> well, let's move on 15:27:12 <yoctozepto> #topic So long, Ceph deployment 15:27:20 <yoctozepto> it has been discussed already above 15:27:50 <yoctozepto> mgoddard: did you remember deprecating something cephish on kolla side? 15:27:58 <yoctozepto> do* you 15:28:16 <yoctozepto> I seem unable to find traces of it 15:28:26 <mgoddard> yoctozepto: I thought I had 15:29:13 <yoctozepto> mgoddard: no signs of it 15:29:16 <mgoddard> perhaps it was just k-a 15:29:27 <yoctozepto> yeah, that was probably what it was 15:29:40 <cosmicsound> So ceph deployment still works on all previous releases beside ussuri? 15:29:52 <cosmicsound> Or will be completely deprecated from k-a 15:30:02 <yoctozepto> cosmicsound: yeah, you can essentially deploy still with Train 15:30:10 <yoctozepto> and then care for such deployment yourself 15:30:20 <cosmicsound> Was just getting ready to do so 15:30:21 <yoctozepto> older branches are still supported 15:30:25 <openstackgerrit> Viktor Michalek proposed openstack/kolla-ansible master: Support for custom API paste files https://review.opendev.org/695054 15:30:50 <yoctozepto> if someone is to pick it up, we could externalize the code for it 15:30:59 <yoctozepto> but that's not the topic for today I guess 15:31:09 <yoctozepto> all righty 15:31:12 <yoctozepto> #topic CentOS 8 migration 15:31:17 <yoctozepto> the huge on 15:31:32 <yoctozepto> I hope you guys did homework and read mgoddard's notes - hrw, mnasiadka 15:31:41 <yoctozepto> or rather a "spec" 15:31:46 <hrw> iirc +2 it 15:31:55 <yoctozepto> yeah, I think you did 15:32:56 <yoctozepto> due to your time constraints, mgoddard and mnasiadka, we might have to postpone this and remaining topics for the next meeting? 15:32:58 <mgoddard> any issues with the proposed approach? 15:33:08 <yoctozepto> ok, you here 15:33:29 <yoctozepto> I think it's quite well-covering now 15:33:52 <yoctozepto> I guess this is as much as one can get without actually doing it 15:34:21 <mgoddard> anyone know the latest on CentOS 8 + ceph? 15:34:31 <mnasiadka> I think it’s good, do want to merge it and add stuff later, or leave it dangling? 15:34:33 <mgoddard> I think that's the main gap in the support we have 15:34:50 <mnasiadka> mgoddard: in progress 15:34:58 <mnasiadka> Slowly... 15:35:05 <hrw> mgoddard: still same: no packages 15:35:09 <mgoddard> if everyone's happy then let's merge 15:35:28 <yoctozepto> yup, let's merge 15:35:29 <hrw> https://koji.mbox.centos.org/koji/packageinfo?packageID=220 still lists only old build of ceph 12 15:36:41 <openstackgerrit> Merged openstack/kolla-ansible master: Change Manila cephfs driver https://review.opendev.org/701515 15:36:44 <yoctozepto> #topic CI scenarios 15:36:51 <yoctozepto> my one finally 15:37:11 <yoctozepto> because someone has been prioritizing theirs :-) 15:37:18 <yoctozepto> so, folks 15:37:26 <yoctozepto> I made the extra scenarios 15:37:35 <yoctozepto> we agreed to and merged the swift one 15:37:37 <hrw> yoctozepto: who cares about CI? 15:37:41 <yoctozepto> plus the great refactoring 15:37:53 <yoctozepto> hrw: I don't know 15:38:08 <yoctozepto> senlin and qinling are dangling atm 15:38:35 <yoctozepto> it surely looks bad to indefinitely grow the list of scenarios for almost each single service 15:38:58 <yoctozepto> especially since they use a lot of the same infra services 15:39:39 <yoctozepto> e.g. one could save cycles on all preparation, bootstrapping, common, rabbitmq, mariadb, memcached etc. 15:40:00 <hrw> yoctozepto: group them, check do they fit in 3h time? 15:40:23 <yoctozepto> might be as easy as that, surely would work senlin 15:40:27 <yoctozepto> and qinling 15:40:49 <mgoddard> I think we just need to come up with some sensible groupings 15:40:58 <yoctozepto> we could get all, let's say, "less impacting" services into one extra job 15:41:06 <hrw> just not stretch time to much so changes will be checked in sane time 15:41:11 <yoctozepto> mgoddard: any proposals of such? 15:41:32 <mgoddard> no 15:41:46 <mgoddard> impact is one helpful dimension though 15:42:02 <yoctozepto> then we might be best throwing in those extra little devils in that one job 15:42:08 <yoctozepto> keeping current scenarios for now 15:42:08 <mgoddard> as is dependencies - group things that depend on nova etc 15:42:40 <yoctozepto> ah, yeah, this is a good hint 15:42:51 <mgoddard> and similaritity 15:42:59 <mgoddard> monitoring, storage, etc. 15:43:16 <mgoddard> although overlapping roles might cause problems 15:43:53 <mgoddard> magnum & octavia could go together 15:44:10 <mgoddard> maybe qinling too 15:44:22 <mgoddard> containery stuff 15:44:35 <yoctozepto> mhm, that makes sense 15:44:38 <cosmicsound> kuryr/zun 15:44:53 <hrw> cosmicsound: +2 15:45:08 <mgoddard> I guess we need a list of services on an etherpad, and we can propose some groupings 15:45:20 <mgoddard> anyone want to sign up to starting the list? 15:45:20 <yoctozepto> though zun does not do nova 15:46:18 <yoctozepto> mgoddard: you know who the CI guy is, why do you ask? 15:46:19 <cosmicsound> then kuryr added next to magnum 15:47:11 <mgoddard> #action CI guy to start an etherpad with a list of services for CI testing 15:49:02 <yoctozepto> works for me 15:49:11 <mnasiadka> mgoddard: maybe we could create the promised doc?:) 15:49:24 <mgoddard> mnasiadka: that would be nice 15:49:51 <mnasiadka> csv-table and go :) 15:50:00 <yoctozepto> ah, the promised doc 15:50:09 <yoctozepto> mnasiadka is right 15:50:22 <yoctozepto> it was on me and I forgot, sorry 15:51:12 <mnasiadka> yoctozepto: no need to be sorry, I forget what I do on daily basis :) 15:51:43 <yoctozepto> mnasiadka: if you say so 15:51:46 <yoctozepto> let's move on 15:52:04 <yoctozepto> #topic unpublishing images, e.g. https://hub.docker.com/r/kolla/ubuntu-source-qdrouterd/tags 15:52:11 <yoctozepto> so we have some images dangling 15:52:20 <yoctozepto> preteding to be the latest, greatest in the world 15:52:34 <yoctozepto> yet not so much in reality 15:52:49 <yoctozepto> a policy on unpublishing is needed 15:52:53 <cosmicsound> i hit this issue in train 15:53:17 <hrw> we can create unpublish job 15:53:35 <yoctozepto> to clean up master tags at least 15:53:36 <hrw> with some list of images to unpublish maintained in repo 15:55:48 <yoctozepto> could work 15:55:59 <hrw> quite easy stuff 15:56:00 <yoctozepto> better to download manifest and compare against current list 15:56:20 <yoctozepto> anyone volunteering? 15:57:22 <mjturek> hrw sorry for the delay but in response to yesterday's question, I'm fine with it as is! 15:57:38 <mjturek> referring to https://review.opendev.org/#/c/699361/ 15:57:38 <patchbot> patch 699361 - kolla - build: remove not buildable templates for --templa... - 2 patch sets 15:57:40 <hrw> mjturek: thanks 15:57:55 <yoctozepto> I guess not, so I will sketch it as blueprint and leave it be for now 15:58:07 <yoctozepto> #action yoctozepto to create bp for unpublishing 15:58:53 <yoctozepto> we don't really have time for free-form discussion in today's session I'm afraid 15:59:03 <hrw> happens 15:59:11 <yoctozepto> thank you, folks, for attending :-) 15:59:20 <yoctozepto> #endmeeting