Tuesday, 2017-06-13

*** mugsie has quit IRC06:35
*** mugsie has joined #openstack-tc06:39
*** mugsie has quit IRC06:39
*** mugsie has joined #openstack-tc06:39
*** jpich has joined #openstack-tc07:29
ttx(Office hours time)09:00
ttxWho is around ? flaper87 johnthetubaguy maybe?09:01
ttxI don't see cdent on the channel list09:02
*** mugsie has quit IRC09:08
*** mugsie has joined #openstack-tc09:09
*** mugsie has quit IRC09:09
*** mugsie has joined #openstack-tc09:09
*** cdent has joined #openstack-tc09:59
*** cdent has quit IRC10:51
johnthetubaguyttx: I had intended to be, but got distracted10:55
johnthetubaguyttx: hoping to have a job again soon-ish, apparently paperwork can take a long time10:56
*** mugsie has quit IRC11:16
*** mugsie has joined #openstack-tc11:18
*** mugsie has quit IRC11:18
*** mugsie has joined #openstack-tc11:18
ttxIt can indeed11:29
*** cdent has joined #openstack-tc11:38
*** cdent has quit IRC12:22
*** cdent has joined #openstack-tc12:34
ttxFor those interested, I refreshed the contribution drop data a couple of weeks ago after smcginnis asked for it12:35
ttxSee https://etherpad.openstack.org/p/9YSc50YgUb12:35
ttxA number of project exhibit a YoY drop > 30%: keystone, swift, glance, searchlight, designate, horizon, heat, trove, docs, oslo, sahara, chef, fuel12:38
cdentttx: do insight to how many different factors are involved in the drop. For example we know that one major factor is lack of people supply, but is there any extent to which "maturity" could be a factor?12:39
ttxMost of the drop is stable (i.e. not worse today than a couple of months ago), only a few projects seem to have recently dropped: swift, fuel, sahara, trove12:40
ttxcdent: maturity is a factor in the 2016 drop around oslo, for sure12:40
ttxsome are more of a long-term trend, like Horizon12:40
ttx(or docs)12:41
ttxOthers are just base projects that had an unusual concentration in "original" contributors from Rackspace / IBM / Intel, like Glance or Keystone12:41
ttxso when one of the oldies decide to readjust their contribution, it hurts those base projects more12:42
* cdent nods12:43
ttxKeystone, Glance, Horizon & Docs are slightly more worrisome, since lots of projects depend on them12:44
ttxFor Horizon & Docs I'd say that those teams should have migrated to more of a mentoring / base tooling team a long time ago12:44
ttxsince there is no way you can do that work for everyone anyway12:45
ttxDangerous levels (> 60% drop) are only reached for Fuel, Trove, Designate and Searchlight12:48
*** mugsie has quit IRC12:48
ttxTrove seems to be in better shape now that amrith found a new home12:48
ttxFuel is an outlier, which we should probably make a hosted project at that point12:48
ttxDesignate we identified early I something that needs support12:49
ttxSearchlight... more of an open question I guess12:50
*** mugsie has joined #openstack-tc12:50
notmynamettx: FWIW, here's my graph tracking swift active contribs http://d.not.mn/active_contribs.png12:51
cdentsearchlight will be interesting over the long run: nova wants to depend on it for listing instances from multiple cells12:51
ttxOn the bright side, those projects experience YoY growth: ironic, karbor, kuryr, zun, kolla, watcher, cloudkitty, dragonflow, freezer, monasca, senlin, watcher12:52
ttxnotmyname: would you see it tracks Rackspace's diminishing involvement, or there is more ?12:53
ttx(also why are you up at this hour ? traveling ?)12:54
ttxcdent: wasn't there a thread on using ELK stack (now ElasticStack) instead ?12:54
ttxnotmyname: err.. I mean "say", not "see"12:55
cdentit's unclear how any of that resolved12:55
cdentthe question of how to deal with parsing notifications still remains, and searchlight was going to fix that12:55
notmynamettx: rackspace (at least cloud files) hasn't been actively involved for years. maybe since 2015. maybe 2014?12:56
notmynamettx: other RAX activity has been largely from matt in australia (who was part of the OSIC fallout)12:56
notmynamettx: I've got a meeting now, but I'd be happy to discuss later12:57
ttxnotmyname: yeah, so maybe more of a "old project with old contributors" effect, where reduction of involvement from initial large shops affects old projects more (like Glance and Keystone)12:58
ttxnotmyname: ack, happy to12:58
notmynamettx: one factor is IBM (the increase starting mid 2015) and their subsequent 1.3 billion purchase on a swift competitor (spring 2016)12:59
ttxyeah, might just be a return to "normal levels", but still would prefer to see that curve start to flatten13:01
*** flaper87 has quit IRC13:06
smcginnisttx: Thanks for pulling all that data. I find it useful.13:07
ttxsmcginnis: also refreshed my YoY merged commits graph @ http://imgur.com/a/I52r013:10
ttxIn terms of official "OpenStack" activity June 2015-June 2016 is comparable to June 2016-June 201713:11
smcginnisttx: That actually doesn't look so bad, IMO.13:11
ttxsmcginnis: especially if you consider that we stopped adding that many projects13:11
ttx(which was fueling most of the growth)13:12
ttxNow ideally I would graph key projects with the same timeline... nomrlizing on 2014 contributions or so13:13
* smcginnis likes pictures :)13:13
cdent4 color glossies13:13
ttxbut then even that would be misleading, as it's "normal" that some projects drop activity when they get to "enough features"13:13
ttxone might even consider that healthy13:13
* cdent queues up alice's restaurant13:14
ttxI use that analysis to make sure there isn't a big issue flying under the radar, not really to dictate actions :)13:14
ttxfor example here, I did not notice until few weeks ago that swift contributors have been dropping for the last year or so13:15
smcginnisNumbers can be skewed in so many different ways, it's hard to just take anything at face value since there are so many factors.13:15
smcginnisBut I think you're right, it's good to see if anything stands out.13:15
ttxRight. I take those as "things are not completely horrible"13:16
smcginnisWhen I've been asked, I've been saying things are healthy and we're focusing more on things that are important.13:16
smcginnisThese numbers at least don't disprove that. :)13:17
ttxsmcginnis: still, I think we shouldn't ignore that the days where we could afford not to actively look for new contributors are over :)13:17
ttxespecially for critical functions and projects others depend on13:18
smcginnisttx: Times are certainly changing.13:18
*** cdent has quit IRC13:51
*** mugsie has quit IRC13:55
*** mugsie has joined #openstack-tc13:58
*** flaper87 has joined #openstack-tc14:14
*** flaper87 has quit IRC14:15
*** flaper87 has joined #openstack-tc14:15
*** emagana has joined #openstack-tc14:27
*** mugsie has quit IRC14:32
*** mugsie has joined #openstack-tc14:38
*** mugsie has quit IRC14:38
*** mugsie has joined #openstack-tc14:38
*** cdent has joined #openstack-tc15:04
*** cdent has quit IRC16:01
*** jpich has quit IRC16:05
fungioh, right, i should add our office hours to my schedule while i'm thinking about it (not that i was awake at 09:00 utc today)16:48
*** cdent has joined #openstack-tc18:25
*** sdague has joined #openstack-tc19:23
*** cdent has quit IRC19:56
*** dirk has joined #openstack-tc20:46
*** alex_xu has quit IRC20:56
*** alex_xu has joined #openstack-tc20:57
*** sdague has quit IRC22:51

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