15:01:06 #startmeeting RDO meeting (2015-1-04) 15:01:06 Meeting started Wed Nov 4 15:01:06 2015 UTC. The chair is apevec. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:06 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:01:26 argh typo in title :) 15:01:31 #chair trown 15:01:31 Current chairs: apevec trown 15:01:43 #topic rollcall 15:01:45 o/ 15:01:49 o/ 15:01:51 !o! 15:01:53 #info agenda at https://etherpad.openstack.org/p/RDO-Packaging 15:01:55 \o 15:02:22 o/ 15:05:01 ok let's start 15:05:03 #topic Mitaka Summit reports 15:05:19 jpena, number80 - can you do quick summary? 15:05:32 More details can be rdo-list post/blog 15:05:53 sure. I was in the release mgmt session. The highlights were: 15:05:55 what should we take into account for RDO Mitaka? 15:06:06 Git tags will be used for versioning instead of setup.cfg 15:06:58 yep, ^ that applies to stable/liberty too 15:07:28 Some discussion on the "drop synced milestones" item. So there won't be a fixed day for Milestone 1,... instead there will be "suggested dates", which I understand as one week or so where projects will be releasing their versions 15:08:08 I remember us agreeing on figuring out a way to make it easy to pull in patches (either between milestones or if an operator/vendor wants to build a package with their own patches), say, perhaps a jinja template that iterates across the patches in a directory or something like that 15:08:14 There were some proposals not to tag betas at all. I stated that as a problem for distros (no way to have Beta X RDO Mitaka unless you chase each project individually) 15:09:04 Also, stable branches for Liberty will happen on demand, without a synchronized release date 15:09:11 jpena, yeah, no tag means no release 15:09:19 Note that in https://wiki.openstack.org/wiki/Mitaka_Release_Schedule there's now date ranges, not fixed dates. 15:09:38 which is fine with me 15:09:53 unless some project slips the release range 15:10:05 Do we want to do test days before that, or after that? 15:10:24 Or both, I suppose, is another option, if we do them more frequently. 15:10:28 rbowen, we should aim +1 week when tags are pushed 15:10:47 with moar automation it should be doable 15:11:32 dmsimard, we don't need a template, rdopkg could pull patches, we just need some tag as a base 15:11:56 apevec: aye but I doubt SUSE uses rdopkg :p 15:11:58 but let's first get automated distgit builds on tags 15:12:07 I think dmsimard is talking about the packaging-rpm session 15:12:08 and we're converging on upstream /openstack rpm repos now I think 15:12:21 ah that's next subtopic 15:12:32 yes, waiting for jpena to finish :) 15:12:32 doh, sorry 15:12:38 :) 15:12:39 still jetlagged to death 15:12:44 that was mostly it 15:12:53 thanks! 15:12:59 thanks for replacing me on such short notice! 15:13:26 in // of the stable mgmt session, we had the rpm packaging fishbowl followed by a workshop 15:13:58 highlights were: 1. we agreed to continue the common packaging goal but accelerate the push of new spec files 15:14:18 rbowen: btw, I have cracked the Japanese syllable stress rules! stress is on the third last syllable, so mitaka was correctly pronounced ;) 15:14:24 2. to reach that goal, we decided to add more core contributors (I nominated apevec and jruzicka) 15:14:38 oh yes, I need core 15:14:50 Good to know. 15:14:52 3. we managed to have a hallway agreement on pushing delorean for build automation upstream 15:14:58 to release pymod2pkg https://review.openstack.org/#/c/241221/ 15:15:20 number80: cool, I did not know about 3, that is awesome 15:15:43 trown: hallway discussion with Dirk and Igor about it 15:15:45 re 3.: it was unclear where that system would live I think, we discussed either ci centos infra or openstack-infra (for gating) 15:16:03 btw, thanks to Derek and trown for attending the RPM packaging session 15:16:17 and dmsimard 15:16:25 I don't count :p 15:16:52 both packaging groups (RPM and debian) attended each other sessions, and we found (not a surprise to me) that we have very similar issues and that we're at the same stage 15:17:12 dmsimard: you were compelled to attend! 15:17:41 Goal is to have some deliverables for Mitaka. 15:18:01 I'm glad that the SUSE folks are very open to collaborating on this 15:18:07 We also had an informal hackfest on friday morning to review spec files 15:18:56 same goes for Mirantis folks 15:19:24 ok, lemme try to summarize for the minutes 15:19:36 This is more interesting for RDO, but goal is at some point RPM packaging project replaces delorean 15:20:18 #info release management session highlights - desynchronized Mitaka milestones and stable/liberty point releases 15:20:49 #info upstream packaging-rpm get momentum - extend core, deliverables for Mitaka 15:21:21 #info move Delorean into upstream infra, replaced by packaging-RPM at some point 15:21:29 jpena, number80 ^ good enough? 15:21:49 apevec: excellent! 15:21:58 we're not moving delorean exactly, just reusing delorean tool for upstream packaging 15:22:11 ah then 15:22:13 #undo 15:22:14 Removing item from minutes: INFO by apevec at 15:21:21 : move Delorean into upstream infra, replaced by packaging-RPM at some point 15:22:21 number80, please correct me 15:23:06 #info use Delorean tool for upstream infra builds, packaging-RPM builds will replace RDO delorean at some point 15:23:32 that's something we tried to sell but it was not accepted by suse 15:23:35 what is "infra builds" ? 15:23:46 packaging-RPM spec files builds 15:24:25 all the work we're doing in delorean, fixing specs would happen upstream, and RDO stable builds would derive from then 15:24:34 apevec, number80: http://logs.openstack.org/58/239258/9/check/gate-puppet-openstack-integration-scenario001-dsvm-centos7/d0ca800/console.html#_2015-11-04_15_03_51_572 15:24:44 /usr/bin/yum -d 0 -e 0 -y list openstack-trove' returned 1: Error: No matching Packages to list 15:24:46 using GA repo 15:24:54 EmilienM, yeah, not pushed yet 15:25:00 EmilienM: ack, we'll review that post-meeting 15:25:00 also OT :) 15:26:07 let's move on 15:26:12 #topic RDO meetup 15:26:15 rbowen, ^ 15:26:16 We had roughly 70 people in attendance for the RDO meetup. I recorded everything, and am working on doing some transcribing. Please blog about your part if you presented, or attended and cared about a part of the conversation, as I'm sure I'll miss details. 15:26:35 The agenda of discussed topics is at https://etherpad.openstack.org/p/rdo-tokyo 15:27:03 We had some folks coming and going, so it was hard to get a count, but it was 70 at one point. 15:27:19 #info 70 people at RDO meetup, minutes coming, please blog your views if you were there! 15:27:39 rbowen: some people told me that they couldn't find the room so we could have gotten slightly more people ;) 15:27:45 rbowen, thanks for recording, I'm looking forward to review notes! 15:27:52 Yeah, the room wasn't exactly central. 15:28:10 rbowen, get flashing lights next time :) 15:28:23 In Austin, we'll all be in the same building, which will help. 15:28:40 "RDO meetup - we're OPEN" 15:28:47 I thought the lunch was a bit awkward with no tables, I expected some sort of round table or multiple tables. It was all chairs :) 15:28:51 At least, I assume we will - the convention center is pretty huge. 15:28:56 rbowen: two hours would be nice to have 15:29:11 ya I think having it on top of lunch was not ideal 15:29:12 So, for Austin we need to start earlier in our quest for a room. There we very slim pickings this time. 15:29:17 or at least, keep some time for new faces to present themselves 15:29:33 cloudkitty folks were too shy to speak up :) 15:29:37 We'll also have more time at FOSDEM, but likely a different audience. 15:29:38 I would have also liked to hear from users and operators more, their experiences - what they like, dislike, would like improved 15:29:50 (or we're too loud to let new faces to speak up) 15:30:09 dmsimard: good idea 15:30:21 Yeah, one hour is just not long enough. 15:30:47 I think that some CERN folks will be present at FOSDEM, they *should* submit a talk (and not just CERN btw) 15:30:50 Anyways, I'll try to get a blog post out this week, and others who spoke are encouraged to blog aobut their piece. 15:30:53 #chair jschlueter 15:31:01 #chair jschlueter 15:31:14 #chair rbowen number80 15:31:14 Current chairs: apevec number80 rbowen trown 15:31:20 #chair jschlueter 15:31:20 Current chairs: apevec jschlueter number80 rbowen trown 15:31:21 #chair jschlueter 15:31:21 Current chairs: apevec jschlueter number80 rbowen trown 15:31:35 #chair jpena dmsimard 15:31:35 Current chairs: apevec dmsimard jpena jschlueter number80 rbowen trown 15:31:46 rbowen: I'll try to write at least a short one 15:32:03 #action number80 write blog post about summit 15:32:19 #action trown blog about summit from rdo-manager perspective 15:32:42 #action rbowen start earlier our quest for RDO meetup room 15:33:09 anything else? 15:33:25 No, I think that's done, until I get through the recording. Thanks. 15:33:36 thank you! 15:33:39 #topic RDO Mitaka themes 15:34:02 this one should derive from the Summit themes 15:34:23 I've notice few on upstream list and started opening tasks in trello 15:34:44 there are also unfinished from previous cycle 15:34:52 e.g. 15:34:55 #info python3 15:35:24 we have stopped that b/c it python guidelines changed late in Liberty cycle 15:35:32 but now should start on that 15:35:42 yeah, right time to reboot that but at a slower rate so that we could review them properly 15:35:44 one issue is switching db driver 15:35:49 what else? 15:36:15 Fedora 22 not having python 3.5? 15:36:27 #info switch to PyMySQL https://trello.com/c/q0VoAYJq/89-migrate-mysql-python-to-pymysql 15:36:30 yeah, forgot that 15:36:49 jpena, yeah, we'll need 2 delorean instances, 15:36:55 during cross-projects session, everyone agreed to support python35 as default python3 stack 15:36:57 one for Fedora stable with py2 only 15:37:06 and Rawhide/f24 with py3.5 15:37:23 number80, is f24 going to have _only_ py3.5 ? 15:37:29 or there'll be still python2.7 ? 15:37:37 f24 will have both 15:37:54 too much still depends on py27 methinks 15:37:54 bah, that's silly, so we'll need all that python2-* mess 15:37:57 but it's time to consider having python35 in CBS 15:38:22 number80, let's make it work on Rawhide first 15:38:25 ack 15:38:32 imho python35 from epel will be pain 15:38:34 ok, I'll try to include the rawhide worker in tomorrow's delorean rebuild 15:38:42 b/c you'd have to rebuild every single python lib 15:38:54 jpena, ack 15:39:04 action yourself :) 15:39:17 #action jpena to include fedora rawhide worker in delorean rebuild 15:39:57 number80, do we have python3 trello card ? 15:40:13 apevec: nope 15:40:24 I was starting one :_ 15:41:34 ack, please add links to other cards like pymysql one 15:42:02 moving on, another unfinished from L 15:42:19 #info -tests subpackages and testdeps / enable %check 15:42:35 not much to say, we just need to do it now 15:42:50 I'll add card with checklists for tracking that 15:43:13 #action apevec add card for tracking -tests subpackages / testdeps / enable %check progress 15:43:47 btw, tagging these card with Next Release label 15:44:21 dmsimard, just one CI question - functional tests are now in projects, how will jobs include them? 15:44:49 is it just matter of installing them and tempest picks them up? 15:45:06 apevec, number80: same for Aodh: '/usr/bin/yum -d 0 -e 0 -y list openstack-aodh-common' returned 1: Error: No matching Packages to list -> in GA repo 15:45:15 ack 15:45:28 it's blocking our CI to bump to GA and provide stable/liberty 15:45:33 I thought GA repo would be ready 15:45:38 any timeframe? 15:45:48 apevec: what kind of functional tests ? can you link example ? 15:46:21 Are they atomic in the sense that they only require their own package to run the tests ? 15:46:51 e.g. https://github.com/openstack/neutron/tree/master/neutron/tests/functional 15:47:11 iiuc those are moved out of tempest source tree ? 15:47:30 but let's take this offline or list 15:47:38 and manila is broken too: http://logs.openstack.org/42/241642/1/check/gate-puppet-manila-puppet-beaker-rspec-dsvm-centos7/ff9e318/logs/manila/api.txt.gz#_2015-11-04_15_33_58_303 15:47:54 sorry, am I in the middle of a meeting? 15:47:56 apevec: I have no clue - if these are standalone to the packages, I'd probably include them in the delorean ci gate 15:48:00 EmilienM: yes :) 15:48:13 number80: I feel really dumb. sorry for that guys /quit 15:48:30 EmilienM: you're jetlagged like most of us 15:48:37 related to ci will we have ci run more than just tempest smoke tests? 15:49:11 jschlueter, let's have smoke first :) 15:49:13 https://trello.com/c/pFBmc3rk/80-bump-rdo-liberty-ci-tests-from-minimal-to-smoke-tests 15:49:28 jschlueter: I want to set the bar pretty high for CI - the first target is to have stable smoke tests. We're really close to that. 15:49:31 apevec: noted ... good goal! 15:49:31 rdo-manager is running smoke 15:49:40 trown: are you still skipping tests though ? 15:49:51 (that's cheating :P) 15:50:15 dmsimard: skipped tests why is that cheating? 15:50:15 lol, I think there are a couple skips with BZs 15:50:18 half-smoke? 15:50:44 apevec: ping 15:50:45 olap: Ping with data, please: https://fedoraproject.org/wiki/No_naked_pings 15:50:51 but yeah anyway, I eventually want to run full tempest and defcore 15:50:54 olap, after meeting please 15:51:27 apevec: could you please approve me in th erdo mailing list? 15:51:30 apevec: kk 15:51:39 RDO 15:51:47 let's move on, we have few more topics 15:51:47 +1 on full tempest run 15:51:54 #info DLM support 15:52:11 I've posted on rdo-list, so this is just FYI - please reply there 15:52:26 full tempest is not very useful for you I think 15:52:29 for rdo-manager, I brought it up in tripleo meeting yesterday, so it is on the radar 15:52:34 in my experience, smoke + some other tests are useful 15:52:38 I noticed it on upstream list as important for Mitaka 15:52:41 they test the whole stack 15:52:51 EmilienM: not for gating, but before a release 15:52:58 you don't really need to test negative tests for Nova Aggregate tests 15:53:00 smoke is fine for gating 15:53:19 EmilienM, so there should be extened-smoke tag in tempest? 15:53:33 apevec: not sure they would do that 15:53:52 sorry, what are we testing? Packages? or code ? 15:53:53 EmilienM, how to document " + some other tests" ? 15:54:09 if you have specific patches in RDO packages that are not in OpenStack, +1 for Full Tempest run 15:54:33 if you just want to test packages that are based on upstream only (like does delorean), Smoke + some tests (dashboard because not in smoke), etc 15:54:48 o/ 15:55:13 ok, let's move that discussion to rdo-list, dmsimard ? 15:55:22 sure 15:55:42 moving quickly thru next few: 15:55:44 #info more automation 15:56:00 jruzicka, ^ we'll need more automagic :) 15:56:09 CI Automation: More and better CI and also auto-promote is on the radar 15:56:20 and auto-build 15:56:28 the auto promote job is already there but I want a smoke job before it 15:56:37 cool 15:56:56 always ready for mor automation ;) 15:57:04 *m04r 15:57:05 jruzicka, I'll file some rdopkg RFEs 15:57:16 apevec, please do ;) 15:57:28 #info rdo-manager quickstart 15:57:34 trown, ^ ? 15:58:21 ya I have a working playbook to create an "undercloud" virtual appliance that fits into the khaleesi workflow 15:59:19 I would like to have that published with a successful promote, and use that for an rdo-manager quickstart 15:59:38 still quite a bit to do for it, but it is one of my main mitaka goals to polish that up 15:59:49 sounds good! 16:00:07 #topic FOSEDM event 16:00:09 RDO community day, January 29th, at the IBM building in Brussels, prior to FOSDEM 16:00:09 CFP at http://goo.gl/forms/oDjI2BpCtm and will be announced on rdo-list today 16:00:09 Distros and Virt/IaaS devrooms available. CFPs mentioned on rdo-list 16:00:25 That's really the whole story. I'll be emailing rdo-list with the details I have in just a moment. 16:00:28 ack 16:00:42 and last FYI: 16:00:46 #topic Delorean instance rebuild on Nov 5 16:00:59 rbowen, ^ jpena will need DNS change 16:01:11 supposedly I should have access now? 16:01:14 yes, just a reminder that tomorrow I will be rebuilding the instance. 16:01:23 ok. just send the info and I'll make it happen. 16:01:54 #topic open floor 16:02:03 overtime, so be quick :) 16:02:26 * number80 has none 16:02:31 ok, thanks everyone! 16:02:33 #endmeeting