15:00:29 #startmeeting RDO meeting - 2019-06-12 15:00:31 Meeting started Wed Jun 12 15:00:29 2019 UTC and is due to finish in 60 minutes. The chair is jpena. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:32 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:33 #topic roll call 15:00:36 The meeting name has been set to 'rdo_meeting___2019_06_12' 15:00:43 Remember to add any last-minute topic to the agenda at https://etherpad.openstack.org/p/RDO-Meeting 15:00:45 o/ 15:00:46 o/ 15:00:55 #chair amoralej ykarel 15:00:55 Current chairs: amoralej jpena ykarel 15:02:39 o/ 15:03:02 o/ 15:03:09 #chair baha mjturek 15:03:10 Current chairs: amoralej baha jpena mjturek ykarel 15:03:51 alright, let's start with the agenda 15:03:58 #topic Pike is now extended maintenance, time for EOL in RDO 15:04:16 amoralej, this is yours ^^ 15:04:37 pike was moved to extended maintenance what means that there will be no new releases 15:04:59 although branches can remain opened in some projects 15:05:07 it's up to each project to maintain 15:05:25 so i'd propose to just take it as an EOL from RDO side 15:05:28 as we did for ocata 15:05:40 just, keep the DLRN builder running 15:05:54 I'm fine with that 15:05:58 but remove jobs, and references to cloudsig repos 15:06:18 and iirc, when centos-7.7 is published we can retire it 15:06:31 amoralej, with jobs means u mean just cloudsig jobs or also tripleo promotion and upstream jobs also? 15:07:00 well, it's up to them, tbh 15:07:13 as the consume RDO Trunk 15:07:27 yes i asked because i saw some recent changes to stable/pike 15:07:27 i'll send a mail for users 15:07:56 yes, some projects may decide to maintain it with some gating 15:08:19 and if tripleo considers useful to keep promotion pipeline working for it 15:08:42 no problem from my side, i'm afraid that at some point some projects may break 15:09:32 just one to alert other users or operators cloudsig repo is no longer maintained 15:09:54 #action amoralej to send a mail about Pike CloudSIG repos being EOL 15:10:08 yes, for TripleO side, it would be TripleO says pike is unmaintained, then next step would be clearing upstream and promotion jobs 15:10:19 cloudsig can be cleared as there will be no more releases 15:12:04 shall we move on to the next topic? 15:12:09 we can also remove pike from all the bots, updating from u-c, automatic promotions to -testing, etc... 15:12:13 ykarel, ^ 15:12:37 amoralej, yes 15:13:09 #info tracking pike removal in https://trello.com/c/6e0bVGzK/712-pike-is-em-removal-from-rdo 15:13:28 let's add what we need to do there 15:13:34 ack 15:13:42 jpena, yeah, we can mov on 15:13:58 good 15:14:02 #topic altarch container discussion 15:14:09 hey! 15:14:22 a few things to bring up 15:14:49 1) a patch is up that tags the containers for registry.rdoproject.org https://review.rdoproject.org/r/#/c/20974/ 15:15:08 reviews are appreciated 15:15:12 the intent is to have a tag that is _ppc64le 15:15:33 these would be in the same namespace as containers that are 15:15:49 (the ones that are being x86_64 containers) 15:16:18 upon promotion, the arch portion of the tag would be removed, a manifest would be created, and we'd upload to dockerhub 15:16:22 all sounding sane? 15:16:27 jpena, may be we may need some storage for ^^ new containres, is that already considered 15:16:52 ykarel: yes, we took that into account last time we upgraded storage for registry.rdo 15:17:00 okk cool 15:17:08 * jpena crosses fingers 15:17:34 mjturek: that sounds ok to me 15:17:39 2) jpena has a patch up for adding registry credentials to cico https://review.rdoproject.org/r/#/c/20974/ 15:17:50 hasn't moved in a bit so reviews would be appreciated 15:18:04 however, jpena mentioned that there may be a path forward in the meantime 15:18:13 jpena could you elaborate on that? 15:18:32 mjturek: yes. While it gets merged, you can work as if it was already. I've manually added that env var to the different jenkins machines 15:18:49 ahhhhhh understood 15:18:52 so merging the review will just make things "right" :) 15:18:53 thank you jpena 15:19:08 that makes sense :) 15:19:14 okay I'm good! baha you have anything here? 15:19:29 mjturek, so these containers would be pushed as part of TripleO promotion? will u be adding some jobs as well to test these built containers? 15:19:42 mjturek: Nope! I'm good 15:19:50 ykarel - we're unclear if that's in our hands or our partners at redhat's 15:19:57 but there will be some minimal testing to start 15:20:09 as i see u running those in ci.centos while tripleo promotion jobs are running in zuul, so u would need some sort of sync 15:20:24 yeah we need to sync up with weshay on that 15:20:28 building containers with same hash 15:20:32 okk good 15:21:29 in rdoinfo we launch jobs in ci.centos.org from jobs in zuul 15:21:38 using a jenkins api token 15:22:12 i'm not sure if the same could be done from a job in the periodic pipeline to create containers for ppc64le 15:22:57 it's kind of hack, but could be useful 15:23:06 amoralej: what about just building daily for now? 15:23:08 is that too often? 15:23:20 no, no 15:23:46 but i understood you need to sync somehow to the jobs running in promotion pipeline 15:23:56 ahhhh right 15:24:17 if just using any of the hashes marked for testing by periodic pipeline is fine for you 15:24:35 then probably you can just use the tripleo-ci-testing symlink 15:24:35 got it 15:24:38 at any time 15:25:13 okay - baha and I will talk to tripleo folks this week and see what's up 15:25:18 sure 15:27:10 next topic? 15:27:15 +1 15:27:40 #topic New release of python-sqlalchemy-collectd requires collectd subpackage which is included in opstools 15:28:06 yeah, i was discussing this with mrunge before the meeting 15:28:53 so, while trying to update python-sqlalchemy-collectd to 0.0.4 we hit an issue because it depends on a collectd subpackage which is in opstools, not in rdo 15:29:18 so we were thinking where this package should be, RDO or OpsTools 15:29:39 in any case it will make one to depend on the other 15:29:55 RDO on OpsTools for collectd 15:30:12 which packages depend on python-sqlalchemy-collectd? 15:30:13 what needs python-sqlalchemy-collectd 15:30:13 or OpsTools on RDO for python-sqlalchemy, as the version of sqlalchemy in base is too low 15:30:32 ykarel, functional dependency for kolla 15:30:40 to send metrics to collectd 15:30:45 let me look for it 15:30:51 kolla requires both 15:31:22 at least for the collectd container, it used to require centos-opstools 15:31:25 https://review.opendev.org/#/c/659678/ 15:31:29 not sure how that is today 15:31:58 mrunge, yes, by today, i'd say that kolla is adding both rdo and opstools in every container build 15:32:09 so i'd say we are safe in that side 15:32:45 i'd say that the main topic here is if python-sqlalchemy-collectd is a pure openstack thing or can be used in others places where opstools are used 15:33:36 it is useful in other places as well, making it probably an opstools thing 15:33:52 I hate to say it, as that will make us add sqlalchemy as well 15:34:20 mrunge, you can crosstag from RDO's build 15:34:24 yes 15:34:39 * amoralej crosses fingers to not break anything 15:34:50 same as for qpid? 15:34:57 yep 15:35:09 we now have the messaging sig 15:35:09 last time for qpid we didn't break :) 15:35:10 so if it's mainly needed by collectd containers, so good to start with adding it to opstools repo 15:35:14 (a year after) 15:35:16 as that seems the primary consumer 15:35:29 ykarel: we have collectd-mysql 15:35:48 no real need for python-sqlalchemy-collectd 15:35:54 ykarel, look at https://review.opendev.org/#/c/659678/ will be added in all of them 15:36:04 * ykarel looks 15:36:13 mrunge, explanation from https://review.opendev.org/#/c/629916/ is 15:36:20 "This allows monitoring of SQLAlchemy connection pool usage." 15:36:47 not sure if collectd-mysql or others can do an equivalent thing 15:36:49 zzzeek, ^ 15:36:53 ohhk it's for python-sqlalchemy, so openstack need 15:37:09 collectd-mysql can not do that 15:37:35 amoralej: hi 15:37:43 mrunge, do you want me to take any action to move on with this? 15:37:48 mrunge: what are you looking to accomplish 15:37:51 mail somewhere? bz? 15:38:05 oh....OK 15:38:15 zzzeek, we are thinking in moving python-sqlalchemy-collectd from RDO to OpsTools repo 15:38:31 python-sqlalchemy-collectd is a thing anywhere somethign that uses SQLAlchemy which is also a rhel package AFAIK might be used 15:38:33 both are consumed by kolla and tripleo 15:39:04 basically anything that has both sqlalhcemy and collectd available in its scope woudl probably want this too 15:39:08 zzzeek, sqlalchemy in rhel base is 0.9.4 while python-sqlalchemy-collectd requires >= 1.1.0 iirc 15:39:24 amoralej: rhel8 is 0.9.4 ? 15:39:29 that woudl be...weird? 15:39:33 zzzeek, we are in centos7 :) 15:39:38 no centos8 yet 15:39:42 amoralej: ah 15:39:50 amoralej: then no, you don't need this yet :) 15:40:08 amoralej: id say it's a centos8 thing 15:40:41 https://review.opendev.org/#/c/659678/ is running with centos7 so we need to provide it 15:40:59 btw, zzzeek version 0.0.3 could be good? 15:41:14 amoralej: for sqlalchemy-collectd? 0.0.4 is way better :) 15:41:18 0.0.4 was the one that introduced the requirement on collectd-python 15:41:19 ok 15:41:21 it's all alpha so it's in deep flux 15:41:26 then we need to manage this 15:44:08 I'll follow up with mrunge about next steps 15:45:03 #action amoralej to follow-up with mrunge about the topic 15:45:58 #topic Next week's chair 15:46:06 Do we have any volunteer? 15:47:08 i can take it 15:47:29 #action ykarel to chair the next meeting 15:47:31 thanks ykarel! 15:47:36 #topic open floor 15:47:44 If there's anything else to discuss, this is the right time 15:48:29 we'll need to start syncing with other sigs about CentOS8 stuff 15:48:46 gfidente, ^ do you have any plan for CentOS8? 15:49:03 plan to rebuild nautilus for centos8 once is available? 15:49:23 amoralej yeah I think we should do that 15:49:30 :D 15:49:37 ack, those are good news for us :) 15:50:11 gfidente, fyi we plan to release RDO Train on CentOS8, so we'll depend on StorageSIG nautilus builds somehow 15:50:32 amoralej yeah got it 15:51:03 ok 15:51:47 this was it from me jpena 15:52:26 anything else? 15:52:37 If not, we'll get 5 minutes back for coffee (or tea) 15:54:05 or Irish coffee 15:54:37 in my case it will be horchata 15:54:41 but I digress :) 15:54:45 horchata lol 15:54:56 #endmeeting