Thursday, 2017-07-06

*** johnthetubaguy has quit IRC06:21
*** ttx has quit IRC06:21
*** johnthetubaguy has joined #openstack-tc06:27
*** ttx has joined #openstack-tc06:34
*** ttx has quit IRC06:40
*** ttx has joined #openstack-tc06:40
*** jpich has joined #openstack-tc07:19
*** RuiChen has joined #openstack-tc08:42
*** jpich_ has joined #openstack-tc10:05
*** jpich has quit IRC10:08
*** cdent has joined #openstack-tc10:11
*** jpich_ is now known as jpich10:25
ttxSome refreshed contribution data. Between July 2015 and June 2016 (arguably our peak year) we merged 262 commits per day. Between July 2016 and June 2017, we are at 259 commits per day. Obviously each team is following a different trajectory (up and coming projects growing, stable projects steadily declining), but overall it's more a plateau than a decline.11:42
ttxTeams struggling the most recently are Designate, Searchlight, Trove, Murano. Keystone also shows some decline11:44
dimsttx : not too bad11:44
ttxlbragstad: would you say the current activity levels in Keystone are sustainable ?11:44
ttxdims: also worth noting that growth on the "overall" graph was generally driven by new team addition11:45
RuiChenhi dhellmann, I want to add the release-model of project into the release page, do you think it make sense?11:46
RuiChennow I can get it from yaml file of each project, but I think it's convenient for guys to show project release model in web page11:50
ttxRuiChen: more a discussion for #openstack-release11:50
ttxRuiChen: I answered to you there11:54
RuiChenttx: thank you, I saw your reply11:55
cdentttx: it might be interesting to model the commit graph without added projects12:01
cdentthat is pick a point in the past with N existing projects and compare that with those same projects, now12:01
ttxyes, would be good to limit that graph to a set of projects12:03
ttxI'll try to do that12:03
ttxlike the compute:starter-kit for example12:03
ttxcdent: such a graph would likely peak and decline. Most projects saw a -15% to -25% YoY decline12:07
cdentyeah, that would be my expectation too, but seeing the slope of the graph would be illumunating12:07
ttxSee refreshed data @
cdentthat data looks like a job for ethercalc12:08
*** hongbin has joined #openstack-tc14:01
lbragstadttx: we certainly took a hit - yes14:03
lbragstadttx: we have several initiatives that are well thought out - but we don't have the bandwidth to implement them14:04
lbragstadi'm not sure if that will change after more of the dust settles from the fallout this last spring14:04
ttxlbragstad: ok, that's good info re: bandwidth14:19
ttxSo not really a survival issue, more of a problem to make progress on new features14:20
lbragstadttx: the biggest bottle neck i noticed was at the forum during the API keys discussion14:20
lbragstadmost everyone in the room agreed on the need for API keys - but we don't have anyone with time to do the work.14:21
lbragstadwe have a few initiatives in that state at the moment14:21
* smcginnis turns on the office light15:00
fungiooh, in the office now!15:00
* cdent sits in a comfy chair15:01
ttxwanted to quickly discuss next steps for goals15:01
ttxLooks like we have champions for both, although I don't personally know Chandan Kumar15:02
ttxshuld we poke sdague or mtreinish to make sure they shadow him?15:02
smcginnisShadows are good.15:03
ttxIn other news, in the above analysis searchlight appears to be struggling a bit -- does anyone have a clear view of where it's headed ? It feels like it's not really on track to become a base service, in which case it might be a distraction ?15:07
cdentI seem to recall that some of the main participants evaporated in the last crisis15:09
cdentand there were concerns in nova-land for building a reliance on it15:09
ttxIt's the issue with base projects like searchlight or barbican. You need to push them to widespread adoption, or drop them, there is no try15:11
cdentwell, if we insist on having projects instead of "openstack", we'll have that... ?15:12
smcginniscdent: Fair point.15:14
cdentit may be fair, but itsn't particular helpful or useful, sadly15:14
dtroyerHowever we are organized, when we lose the people who intended to do a bit of work we lose that capability.  If the searchlight folk had been part of Nova directly, we'd be in the same boat.15:15
dtroyer(ignoring the other reasons to split projects up)15:15
dtroyerWhat it seems we are really lacking is the ability to depend for a semi-long period on (new) projects.15:16
fungii honestly don't quite see how one openstack vs disconnected projects solves that particular dilemma15:16
cdentit wasn't about solving15:16
cdentit was about the forces that lead to it happening15:16
cdentmany of the people I talk to think they work on a project15:17
cdentand many of the employers i talk to hire people to work on a project15:17
cdentnot on openstack15:17
cdentas someone hired to work on nova, it would be challenging for me to say "I need to work on searchlight because we need it in nova"15:17
fungiforces that led to splitting searchlight out of horizon, or the forces that led to rackspace laying off most of their openstack upstream devs?15:17
dtroyerin the last two years I've seen firsthand the difference between a startup mentality (we work on openstack) and the corporate mentality (bodies working on specific goals)15:18
smcginnisdtroyer: +1 - there's a big difference in attitude and willingness to even look at other things.15:19
cdentmentality is a good word for it15:19
cdentwell that seems to have depressed the mood a bit15:22
cdentmy point was that we need to come around to some positive ways of adjusting the mentality15:23
cdentthe top 5 thing is a probably a start15:23
ttxcdent: I hope championing will help too15:23
dtroyerthat attitude is pushed down from managers trying to meet their weekly/monthly goals.  to be fair, they also sometimes want to move devs from one project to another as the corporate strategy changes, so the idea of one project only isn't there, it's just "go do this and nothing else" for any given time15:23
ttxif we spin it as very positive15:23
cdentbut probably doesn't do a lot for long term embedded contributors15:23
smcginnisThe top 5 does have a positive impact I think. I know of some vendors that look at that as a way to justify reallocating resources. At least temporarily.15:24
dtroyerthe greater problem is ensuring the long-term health and stability of those things we (now) need to be around15:24
* dims reads scroll back15:25
ttxthingee will post a superuser article soon, now that the glance one merged15:25
dtroyerI do think the champions will help, especially if we grow a class of experienced champions that get to know the decision makers at our sponsoring companies and can help them understand priorities15:25
cdenthow should one counter the argument of "well, if nobody wants to do the work, maybe it doesn't matter?"?15:26
dtroyerwith a reference to
cdenti don't mean it quite in that way15:27
smcginnisMaybe by making sure we spell out what it brings to OpenStack that would be missing otherwise?15:27
cdentfor instance openstack has got along in a mostly working state for a few years with no barbican15:27
cdentso someone could say "why fix that?"15:27
ttxcdent: arguably one could say openstack is not fully secure as a result15:28
ttxso it's missing a feature, in several places15:28
dtroyerhow is that different from asking why we still support Xen when the deployment on the survey is <10%  (5% even?)15:28
ttxsometimes nobody does the work because everyone waits for someone else to do it15:29
dtroyerttx: that is what I was getting at15:29
cdentcan barbican be marketed as a security fix (instead of a feature)?15:30
ttxLarge service providers like IBm or Red Hat are used to throwing resources at strategic issues15:30
cdent(barbican contribution, I mean)15:30
ttxbut as we transition to a more user-driven contribution base, contributing tactically will become the norm15:30
ttxwe need to actively work on fixing that15:31
ttxBack in 2011-2012 we had the same prblem, with the startup crowd15:31
ttxwe had golden years in 2013-2016 where the orgs sustaining our contribution base happened to understand the value of contributing strategicallyt15:32
ttxNow we need to go back at explaining again15:32
dtroyerlike it is the old "stop taking the meds because I feel better" and wondering why you don't feel better later15:32
dtroyerwe should probably always keep explaining some things at a low level all of the time15:33
fungii agree, though remembering to explain them when they're not in your face due to a problem can be challenging15:34
cdentMaybe we've already had the mental shift, but if not we probably need to have it: the golden years were not normal and not common for open source-like things15:34
ttx wow I feel young again15:34
ttxIt's the famous video where markmc and I showed up with the same shirt15:34
smcginnisWho is that young looking guy. :P15:35
fungia fashion tragedy to be sure. it was the talk of the conference15:35
smcginnisttx: Actually... I think you look the same.15:35
ttxsmcginnis: I was always old15:35
dtroyercdent: right, and that's a lesson the k8s and others on the upswing of the hype train crowd might want to pay attention to sooner than later (they may be already, I don't know)15:35
cdentthe takeaway from that is that _voluntary_ strategic orientation may be hard to come by15:36
fungiwhich brings us back around to finding ways to encourage more of it (top five, community goals, et cetera)15:40
cdentyeah, I guess the word "voluntary" is not what I meant. Perhaps "spontaneous"15:41
cdentmore directed guidance a good thing15:41
fungii wonder if the new contributor portal work could be leveraged to help steer contributors down less specialized paths15:43
dimsfungi : ttx : cdent : have you seen this? (checkout the video)15:44
cdentfungi: that's a good idea, it's certainly seemed in the past that onboarding was oriented towards choosing a project to join15:44
cdentdims: that's interesting, perhaps, but I'm not a good judge of such things. I _hate_ being gamified, incentivized, ranked, etc.15:49
cdentthe challenges idea is perhaps a useful way to get enaged15:49
dimscdent : not advocating, just sharing :)15:49
*** jpich has quit IRC15:49
cdentbut I think, really, if we want engaged devs what we really need is for openstack to be something that's more accessible and useful to devs15:49
ttxdims: it feels a bit too much between on-boarding, training and actually triggering contributions15:50
ttxLike they give a badge for a number of Pull requests. There is only so much typo patches I can take on a calm day15:50
cdentwe've got plenty of devs (like me) who write a lot of in the guts code but rarely use openstack, and plenty of people who use openstack but rarely get the chance to write lots of code15:51
ttxdims: but there are good things in there. Some of the work on the contributor portal replicates that15:52
ttxlike setting up your first steps15:52
ttxThe interesting part is the Rewards area15:53
ttxwe kind of do that with contributors discounts15:54
ttxBut then I'm with cdent, I must be too old to appreciate the gamification systems15:55
* cdent is definitely old15:56
ttxMy experience of those is that they work for a time and then fail completely15:56
ttxusually way before they pay up the time invested on them15:56
ttx -- who remembers "Ubuntu Accomplishments" ?15:58
fungiit's not that i'm tool old to appreciate gamification (and i do play plenty of video games in my limited spare time), but i've seen first hand what gamification (intentional or emergent) does to communities over time16:13
fungis/tool/too/ (though i am a tool sometimes too)16:13
*** cdent has quit IRC19:11
*** lbragstad has quit IRC19:58
*** lbragstad has joined #openstack-tc20:03
mordredlbragstad: btw - api keys is totes on my plate - it's just down a few in the queue20:46
lbragstadmordred: ack - thanks for the update :)20:48
*** cdent has joined #openstack-tc21:02
*** cdent has quit IRC21:13
thingeettx: ack glance goal merged22:08
*** lbragstad has quit IRC23:03
*** lbragstad has joined #openstack-tc23:36
*** hongbin has quit IRC23:38

Generated by 2.15.3 by Marius Gedminas - find it at!