Tuesday, 2018-04-17

*** jroll has joined #openstack-tc00:00
*** jroll has quit IRC00:16
*** jroll has joined #openstack-tc00:17
*** mriedem has quit IRC01:01
*** liujiong has joined #openstack-tc01:59
*** pabelanger_ has joined #openstack-tc02:02
*** pabelanger has quit IRC02:03
*** pabelang- has joined #openstack-tc02:03
*** pabelang- is now known as pabelanger02:05
*** pabelanger_ has quit IRC02:06
*** gcb has joined #openstack-tc02:33
*** annabelleB has quit IRC02:40
*** liujiong has quit IRC03:14
*** liujiong has joined #openstack-tc03:15
*** annabelleB has joined #openstack-tc03:32
*** liujiong_lj has joined #openstack-tc03:38
*** liujiong has quit IRC03:40
*** annabelleB has quit IRC03:51
*** dklyle has joined #openstack-tc03:57
*** jbryce_ has joined #openstack-tc04:03
*** ildikov has quit IRC04:03
*** jbryce has quit IRC04:03
*** mrhillsman has quit IRC04:03
*** jbryce_ is now known as jbryce04:03
*** mrhillsman has joined #openstack-tc04:04
*** gagehugo has quit IRC05:16
*** mrhillsman has quit IRC05:29
*** mrhillsman has joined #openstack-tc05:30
*** gagehugo has joined #openstack-tc05:46
*** gcb has quit IRC05:50
*** ianychoi has joined #openstack-tc06:38
*** gagehugo has quit IRC07:54
*** openstackgerrit has joined #openstack-tc08:01
openstackgerritMerged openstack/governance master: Create tripleo-ha-utils project  https://review.openstack.org/55583008:01
openstackgerritMerged openstack/governance master: Add puppet-senlin to Puppet OpenStack  https://review.openstack.org/55953908:01
openstackgerritMerged openstack/governance master: Add puppet-pacemaker to Tripleo  https://review.openstack.org/55966408:01
*** ianychoi has quit IRC08:04
openstackgerritMerged openstack/governance master: Update docs job info to match current PTI  https://review.openstack.org/55657608:04
*** jpich has joined #openstack-tc08:04
*** ianychoi has joined #openstack-tc08:10
*** gagehugo has joined #openstack-tc08:11
*** cdent has joined #openstack-tc09:04
ttxLooks like we have 9 candidates for 7 positions now09:04
ttxso we'll definitely have an election09:05
ttxAll good candidates too! Already hard for me to pick09:05
cmurphyme too :'(09:06
ttxIn other news, there is a question for us on the Adjutant review, regarding Python3-only projects09:06
ttxI think we said that is alright, but wanted to doublecheck09:06
ttxSee https://review.openstack.org/55364309:07
cdentYeah, I was going to bring that up. If it's not alright previously, we should make it alright09:07
cmurphyseems like it's time09:07
ttxI think it was part of dhellmann-s timeline09:07
ttxIIRC it included a "Python3-only projects are OK from that date on"09:08
ttxso probably better to let him answer that question09:08
cmurphyadrian sent an email to -dev about it as well09:08
ttxAh, nice09:09
ttxcdent: We said we'd poll the -dev list for topics we should bring to the board+TC meeting. What would be the ideal timeline to do that ? I planned to send something today but maybe waiting for the election to conclude would be better ? Or too late ?09:12
cdentI'd say start soon and repeat later?09:12
ttxOK. I can start a thread today and let the new chair pick it up after election early May09:13
cdentI'm not going to be able to make the board meeting, which is very disappointing (to me). I can't get flights to work with a wedding I need to attend the day prior. Very difficult to get late evening departures out of heathrow :(09:14
ttxah, sad09:14
ttxprivate jet it is then09:15
* cdent checks the bank09:15
cdentnot yet09:15
ttxMaybe you can get Mark Shuttleworth to pick you up on the way09:15
* cdent checks his rolodex09:16
cdentnot yet09:16
ttxNow we'll see if he monitors this channel09:17
ttxok, email sent09:24
ttxyet another thing that will be facilitated by moving election further away from summit09:25
andreafhello TC members :)09:35
andreafI've got a short talk (20min) accepted at FOSS Backstage in Berlin in June about OpenStack Four Opens https://foss-backstage.de/session/openstack-and-four-opens09:36
andreafI plan to talk about how OpenStack implemented the 4 opens practically with a mix of governance structures and technical solutions (e.g. IRC logs)09:37
andreafif anyone has a topic that they feel should be represented in that talk please ping me, I'll do my best to include it09:37
cmurphyawesome andreaf !09:43
cdentandreaf: nice!09:43
cdentandreaf: in that context I would think one of the most interesting things would be discussing some of the costs and benefits of the choices09:47
cdentfor example in goverance we have very limited executive power across the entire world of openstack09:47
cdentthat's good almost all the time, unless we want to make a serious course correction09:49
ttxI would mention two things that might be considered undesirable side-effects of following the 4 opens09:50
ttxOne is scope creep -- if you're truly open it's hard to say no to features09:50
ttxThe other is that you have to accept to go slower09:50
ttx(like a dedicated start-up will go faster, but the open collaboration will go farther09:51
ttxandreaf: I'm actually giving a full talk on that at PyCon IT Saturday09:52
ttxYou should join us !09:52
*** cdent has quit IRC10:12
*** liujiong_lj has quit IRC10:29
*** cdent has joined #openstack-tc11:12
cdentFrom email (evrardjp) just now: "Our current way of working is not sustainable in the long run, as a lot of work (and therefore pressure) is concentrated on a few individuals." This is so true, across the board. We need to be actively working on this.11:18
fungii thought we already accepted a python3-only project a while back. now if i could only remember which one11:37
cdentI recall a conversation from smcginnis about such things11:38
andreafcdent ttx cmurphy thanks for the feedback!11:45
andreafttx unfortunately I won't be at pycon it this time :(11:46
andreafbest of luck for your talk!11:46
mugsiefor the py3 only projects we should update https://governance.openstack.org/tc/reference/pti/python.html11:56
openstackgerritGraham Hayes proposed openstack/governance master: Allow projects to drop py27 support in the PTI  https://review.openstack.org/56192212:12
*** mriedem has joined #openstack-tc12:59
*** ianychoi_ has joined #openstack-tc13:07
evrardjpcdent: I didn't expect this mail to be read! :p13:07
cdentevrardjp: a) with a subject like that, how could I resist? b) I like to maintain the illusion (it is definitely an illusion) that I read everything13:08
evrardjpcdent: :)13:08
mugsieevrardjp: you will always find yourself surprised at what people read :)13:09
*** ianychoi has quit IRC13:10
dhellmannmy suggested "timeline" for python-3-only was that we could start allowing that after it was reasonable for deployers to actually deploy on python 3 on the platforms we claim to support (Ubuntu and CentOS)13:38
dhellmannIIRC, the project where this came up in the past was able to make a rather small adjustment to their code and get it to work with py27 as well13:39
*** annabelleB has joined #openstack-tc13:50
dimsw00t @zaneb13:52
fungidhellmann: yeah, i couldn't figure out which one that was from digging back through past discussions13:54
fungiand as far as "python 3 on the platforms we claim to support" it's been possible with ubuntu for a while but centos doesn't yet have a system python3 so depending on how we define that...13:55
fungidoes python3 from software collections count? or from unofficial rpms? or locally-compiled?13:55
mugsieI would say py3 from software collections counts13:56
fungirumors are rhel8/centos8 may have python3 without needing software collections13:56
mugsiefungi: not rumours - it will13:56
fungiapparently deploying services to use software collections is far from straightforward because they're like little chroots and don't make the interpreter available system-wide13:57
mugsiethere was a note in the RHEL 7.5 release notes13:57
fungioh, nice13:57
mugsieand while it didn't say no py27, that was implied, and I have heard py27 won't be there from others13:58
fungiin the past there had been suggestions that red hat would refuse to support more than one python interpreter on the system, implying that rhel8 would have _either_ python2 or python3 but not both13:58
fungiso, right, that may make testing on centos interesting once we sunset centos 713:58
fungithat does seem to confirm it, thanks14:00
mugsieI did not realise that software collections were implmented as docker containers - that does complicate things14:01
mugsiethat does complicate things14:02
mugsiefungi:  would https://ius.io/ be acceptable?14:07
*** hongbin_ has joined #openstack-tc14:10
*** johnsom has quit IRC14:11
*** johnsom has joined #openstack-tc14:12
pabelangersoftware collections is also missing a lot of python libraries, as I understand it, it really is geared towards django apps14:12
mugsieI seem to remember tripleO / RDO / someone had a solution for python3 only CentOS, but I can't remember where I saw it14:19
pabelangermugsie: last I heard is to test on fedora-28 for tripleo / rdo14:20
pabelangerEmilienM: ^14:20
EmilienMpabelanger: in tripleo meeting14:20
dhellmannmugsie, fungi : I'm still working with the RH team to figure out the approach to python 3, but the current platform does *not* support deploying on py3 and I think it would be a mistake to push to drop 2 until we have one that does.14:20
EmilienMmugsie: on openstack-dev14:21
EmilienMmugsie: the idea is to build containers on fedora (that provides python3) and run them on centos7 (our testing platform for upstream)14:21
fungimugsie: it's hard for me to say whether https://ius.io/ is a suitable alternative given that i'm not really an openstack operator nor a user of rpm-based platforms, so i'd defer to the judgement of people who are14:21
dhellmannI think we're talking about 1-2 cycles before we can drop support for 214:21
dhellmannthat's after rocky, so T14:21
dhellmannat the earliest14:22
fungiwe may also need to expect that the last openstack coordinated release to support python 2.7 across the board may see a lot more stable maintenance interest for longer14:22
dhellmannyes, I expect so14:22
mugsiedhellmann: cool - I will link this backlog to the o/goverance patch14:23
jrollperhaps the guidance for new projects should be something about "it's okay to be py3 only, but note that adoption may be slow due to red hat things bla bla bla"14:23
dhellmannmugsie : that's a definite "at the earliest" and not a commitment that it will be T :-)14:23
*** dmsimard|off is now known as dmsimard14:23
jroll(and existing projects should wait for a go-ahead)14:24
mugsiemaybe having a "T" deadline might encourage distros :P14:24
mugsieand some projects14:24
dhellmannjroll : that approach may work. I'm not likely to vote in favor of any projects that don't support the platforms we actually deploy to, but if teams want to wait to become official I wouldn't object to that.14:24
jrolldhellmann: okay, that's different, then :)14:24
dhellmannI mean, I've written python-3-only projects myself lately (downpour) so I get the appeal14:24
dhellmannbut we do have to remember that "openstack is written for our users"14:25
dhellmannand a lot of our users do not have a platform for python 3, yet14:25
dhellmannit's coming14:25
dhellmannvery soon14:25
jrollright, but if a project applies to be official knowing it's likely to have a few cycles before they really gain adoption anyway, they might not mind so much14:25
dhellmannI am not allowed to make more specific statement than what we've said publicly in the 7.5 release notes14:25
mugsiewhat is the RHEL 8 date? because we will have the opposite problem when that lands14:25
EmilienMthere is no date14:25
dhellmannwe don't have a date yet14:26
dhellmannbelieve me, we are pushing the RHEL team hard to get one14:26
dhellmannthis is a huge issue for us, and we're all aware internally, and we're working through it14:26
dhellmannmugsie: the mitigation for the hard-switch is to have py2 and py3 support at the same time14:26
mugsieOK - we will have issues then if we have to block current projects who are py27 only now14:26
dhellmannthat's why, although for 4 years I've been urging us to move to 3, now I'm urging us to slow down a bit14:27
dhellmannyes, as far as I know we only have 1 project that hasn't successfully finished setting up integration jobs14:27
dhellmannwe have quite a few functional test jobs that we need to replicate or move to python 314:27
dhellmannsee https://wiki.openstack.org/wiki/Python3#Python_3_Status_of_OpenStack_projects for (fairly recent) details14:28
mugsieOK, things in some projects are better than I remebered14:30
dhellmannyes, a bunch of teams made progress when we weren't monitoring it closely14:31
cdentI think we need to be ahead of the curve, especially since most users are steps behind our release cycle. So allowing projects to be python 3 _now_ does not mean (all) users are required to python 3 _now_. It means we will be ready for them _then_.14:32
cdentAnd unless we start now nobody will be ready.14:32
dhellmannthe work we're doing in oslo now will form the basis of a goal proposal for next cycle to move everyone else over, but I need to understand what changes have to happen before I can propose that14:32
dhellmanncdent : it doesn't cost us much more to have 2 and 3 support simultaneously, and we already require support for 3.14:33
cdentdhellmann: sure, but it sounds like some projects would like to use some tools that they can't use because of that python2 requirement14:33
cdentAs people said above: adoption may be stymied for a new project that wants to use python3 only stuff, but that's a risk they can accept, perhaps?14:34
dhellmannSo we'll have different rules for new projects vs. existing projects?14:35
dhellmannbecause I don't think we want to say let glance drop support for python 214:35
dhellmannor oslo14:36
smcginnisI'm a little less concerned about sytem packaged python with the trend I've seen of operators moving to containerized deployments.14:36
cdentI'm not certain, just thinking out loud. But generally, I think it ought to be perfectly okay for us to go all python3, on master, today. Any anybody who wants to use old stuff just can. They always have that option.14:36
smcginnisI wouldn't want oslo to drop support, as that would have follow on effects of forcing others.14:36
smcginnisBut "end" projects like services, I wouldn't mind seeing them drop py2.14:37
dhellmanncdent : that ignores new users, and the significant portion of our contributor base who works for a platform that isn't quite ready for that move14:37
smcginnisNew users also have some leeway in choosing which platform to deploy these greenfield deployments on though too, right?14:38
cdentdhellmann: how does it ignore new users?14:38
dhellmannseriously, I'm not talking about waiting for years. and I hate that I'm saying we need to wait at all. but that's the reality I see14:38
dhellmanncdent : it would force them to use something old, no?14:38
cdentNew users are welcome to choose whatever platform they want14:38
dhellmannof course14:39
cdentso how are we forcing anything?14:39
dhellmannwell, they could either have a platform of their choice or a tool of their choice, but not both14:39
*** cdent has quit IRC15:15
*** cdent has joined #openstack-tc15:16
* dims catches up16:14
dimssmcginnis dhellmann : for oslo, is there a proposal on the table?16:16
* dims checks his py3/nova scars ... 16:17
dhellmanndims : not that I'm aware of16:17
smcginnisNone I've heard.16:17
dhellmannI was using that as an example of why we need to keep the rules the same for all projects until we can safely drop support for 216:17
dimsk so business as usual for now16:17
dhellmannyeah, although we need to continue to monitor it16:18
dimsack. i like the idea of the services dropping py2 first and work backwards16:18
dimscould be "leaf" projects like magnum that use everything that get the ball rolling16:19
*** harlowja has joined #openstack-tc16:23
*** jpich has quit IRC16:25
dimsharlowja : making it to vancouver?16:42
*** kfox1111 has joined #openstack-tc16:42
harlowjadeciding; gotta update my passport (noticed that this weekend)16:42
harlowjathough canada is like the US; imho shouldn't need passport :-P16:43
dimssssshhhhh! don't say that out loud16:44
clarkbharlowja: you don't if you have one of those fancy drivers licenses and drive16:44
clarkboh looks like its only NEXUS card and passport cards now16:46
*** harlowja has quit IRC16:52
*** harlowja has joined #openstack-tc16:57
openstackgerritDoug Hellmann proposed openstack/project-team-guide master: add some process instructions for updating dependencies  https://review.openstack.org/56200617:03
*** dklyle has quit IRC17:07
*** harlowja has quit IRC17:20
fungiyeah, they dropped the drivers license thing for canada and mexico a while back17:43
fungiin happier news, a ran across this collection of articles which had some good nuggets of free/libre open source governance experience https://changeset.nyc/resources.html17:45
fungier, i ran across17:45
fungia tech writer for wmf, and so unsurprisingly adept at getting points across in writing17:59
*** annabelleB has quit IRC18:03
*** mriedem is now known as mriedem_shamwow18:16
*** dklyle has joined #openstack-tc18:30
*** hongbin_ has quit IRC18:41
*** hongbin_ has joined #openstack-tc18:41
dhellmannfungi : that looks really interesting; I've added a bunch of them to my reading queue18:42
*** dklyle has quit IRC18:46
*** dklyle has joined #openstack-tc18:51
* persia thought the Western Hemisphere Travel Initiative allowed new "enhanced" drivers licenses to be used as acceptable proof of citizenship and identity to travel between any of the three north american states, with the necessary access controls being available to Canadian Border Patrol Officers since December 201718:54
clarkbpersia: ah maybe the thing I found is out of date if december was when this changed again18:55
persiaclarkb: Between 2008 and 2017, US EDL couldn't be accepted for entry to Canada because the canadian officers didn't have access to the database to confirm the cards were valid (but Mexican EDL was accepted).18:56
*** dklyle has quit IRC18:57
persiaOn the other hand, deciding to issue EDL is a local thing: Quebec stopped issuing them in 2014, as one example.18:57
clarkbwe are going in the opposite direction due to tsa demands aiui18:57
clarkb(so states that refused to offer them are now sorting out how to do it so that you don't need a passport to travel domestically)18:57
persiaDon't confuse "REAL ID" and "EDL": one can comply witht he former without complying with the latter.18:58
clarkbI thought real id was a subset?18:58
clarkbyou ahve to do one to do the other18:59
persiaMy understanding is that all US EDLs are "REAL ID" compliant, but not all documents that are "REAL ID" compliant are sufficient for travel uneder WHTI.  On the other hand, if one *can* qualify for "REAL ID", then one can likely qualify for an APEC travel card, which gives priority entry to all north american countries.19:00
persiaI may not have a complete understanding of these things.19:00
*** cdent has quit IRC19:04
dimsfungi : cool (links!)19:34
*** mriedem_shamwow is now known as mriedem19:49
*** harlowja has joined #openstack-tc19:50
*** annabelleB has joined #openstack-tc19:50
*** dklyle has joined #openstack-tc19:52
*** annabelleB has quit IRC20:14
*** dklyle has quit IRC20:21
*** annabelleB has joined #openstack-tc20:23
*** dklyle has joined #openstack-tc20:31
*** annabelleB has quit IRC20:57
*** annabelleB has joined #openstack-tc21:13
*** dklyle has quit IRC21:42
*** dklyle has joined #openstack-tc21:46
*** mriedem is now known as mriedem_away22:17
*** dklyle has quit IRC22:27
*** annabelleB has quit IRC22:34
*** hongbin_ has quit IRC22:58
*** annabelleB has joined #openstack-tc23:06
*** mriedem_away has quit IRC23:44

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!