| *** diablo_rojo has quit IRC | 00:55 | |
| *** jamesmcarthur has joined #openstack-tc | 01:03 | |
| *** jamesmcarthur has quit IRC | 01:05 | |
| *** jamesmcarthur has joined #openstack-tc | 01:05 | |
| *** fanzhang has left #openstack-tc | 01:07 | |
| *** jamesmcarthur has quit IRC | 01:35 | |
| *** lbragstad has quit IRC | 01:49 | |
| *** lbragstad has joined #openstack-tc | 01:49 | |
| *** ricolin_ has joined #openstack-tc | 01:59 | |
| *** jamesmcarthur has joined #openstack-tc | 02:05 | |
| *** jamesmcarthur has quit IRC | 02:09 | |
| *** evrardjp_ has joined #openstack-tc | 05:51 | |
| *** ianw_ has joined #openstack-tc | 05:54 | |
| *** dims_ has joined #openstack-tc | 05:54 | |
| *** dtruong has quit IRC | 05:55 | |
| *** dims has quit IRC | 05:55 | |
| *** evrardjp has quit IRC | 05:55 | |
| *** ianw has quit IRC | 05:55 | |
| *** ianw_ is now known as ianw | 05:55 | |
| *** dangtrinhnt has quit IRC | 07:00 | |
| *** bauzas is now known as bauwser | 07:35 | |
| *** witek has quit IRC | 07:35 | |
| *** evrardjp_ is now known as evrardjp | 07:37 | |
| *** tosky has joined #openstack-tc | 07:46 | |
| *** jpich has joined #openstack-tc | 08:03 | |
| *** cdent has joined #openstack-tc | 08:52 | |
| *** e0ne has joined #openstack-tc | 09:16 | |
| ttx | I think if affiliation appears as part of the bio it makes it less look like company-reserved seating | 09:18 |
|---|---|---|
| cdent | because there's a mix? | 09:20 |
| cdent | When I go to openstack events I put "OpenStack" and "OpenStack" as my company and job on my badge. | 09:21 |
| ttx | At https://www.openstack.org/foundation/tech-committee/ company and bio are shown, but makes "company" the top info below name | 09:29 |
| openstackgerrit | Thierry Carrez proposed openstack/governance master: Indicate relmgt style for each deliverable https://review.openstack.org/613268 | 09:52 |
| *** ricolin_ has quit IRC | 10:02 | |
| *** jamesmcarthur has joined #openstack-tc | 10:12 | |
| *** jamesmcarthur has quit IRC | 10:16 | |
| *** dtantsur|afk is now known as dtantsur | 10:48 | |
| *** e0ne has quit IRC | 10:51 | |
| *** e0ne_ has joined #openstack-tc | 10:52 | |
| *** EmilienM is now known as EvilienM | 11:24 | |
| *** fanzhang has joined #openstack-tc | 11:26 | |
| smcginnis | As much as we want to think everyone here is coming in unbiased by employer obligations (which I do think we do a pretty good job of) there still is someone paying our bills and "suggesting" what we should be doing. | 11:27 |
| smcginnis | To varying degrees, I'm sure. | 11:28 |
| cdent | destroy the corporate hegemons smcginnis | 11:35 |
| smcginnis | :) | 11:36 |
| ttx | On that topic, I found that an interesting translation of the "openstack first, project second, company third" principle: https://twitter.com/masohn/status/1055360677127376896 | 11:49 |
| smcginnis | Maybe we should link to that. | 11:50 |
| smcginnis | After correcting "Red Hat" (ocd twitch) | 11:51 |
| cdent | I've always been a fan of that bit of policy | 11:51 |
| ttx | It's always tricky to put in practice, but i suspect having it spelled out in your "code of business conduct" must help | 11:51 |
| *** e0ne_ has quit IRC | 12:26 | |
| *** cdent has quit IRC | 12:30 | |
| *** jamesmcarthur has joined #openstack-tc | 12:47 | |
| fungi | yeah, that's worth pointing the osf account management team at as something they can suggest to prospective new member companies. might also be good fodder for the contributing organization recommendations list the fc sig has been assembling | 12:56 |
| *** cdent has joined #openstack-tc | 12:58 | |
| TheJulia | It may also be good for them to have and leverage that to help convey context of a good chunk of the community. | 13:09 |
| openstackgerrit | Corey Bryant proposed openstack/governance master: Add optional python3.7 unit test enablement to python3-first https://review.openstack.org/610708 | 13:13 |
| *** mriedem has joined #openstack-tc | 13:19 | |
| *** e0ne has joined #openstack-tc | 13:25 | |
| tbarron | zaneb: Any chance you can spend a bit of time in the weekly manila meeting to talk through the Cloud Vision? It's at 1500 UTC on Thursdays. | 13:37 |
| *** jamesmcarthur has quit IRC | 13:46 | |
| zaneb | tbarron: added it to my calendar, but ping me here if I forget | 14:04 |
| zaneb | it's the same time as TC office hours so I'll definitely be around | 14:05 |
| *** cdent has quit IRC | 14:05 | |
| scas | as a not-tc, there's definitely a perception of "i am $INDIVIDUAL, representative of $COMPANY. i also work on OpenStack." -- i've been polled by knowledge-seekers about my contribution behaviors where the perception was just that very thing. they weren't seeking knowledge about me the individual, or me the OpenStack community member, but me the corporate representative to see how my handlers behave | 14:09 |
| *** jaypipes is now known as leakypipes | 14:10 | |
| tbarron | zaneb: great, I'll put you on next week's agenda. IMO there are no fit problems between manila and the vision - manila virtualizes back end h/w, enables physical data center mgmt, elastic scaling, etc. but | 14:17 |
| tbarron | zaneb: rather than just +1 myself I want to make sure the manila community can talk this through first | 14:18 |
| zaneb | tbarron: excellent, thank you. looking forward to it :) | 14:18 |
| *** jamesmcarthur has joined #openstack-tc | 14:21 | |
| *** tosky has quit IRC | 14:25 | |
| *** tosky has joined #openstack-tc | 14:29 | |
| *** cdent has joined #openstack-tc | 14:29 | |
| ttx | moved https://etherpad.openstack.org/p/tc-health-checklist to https://wiki.openstack.org/wiki/OpenStack_health_tracker#Health_check_list | 14:54 |
| ttx | please push further edits there instead than on the etherpad | 14:54 |
| openstackgerrit | Zane Bitter proposed openstack/governance master: Resolution on keeping up with Python 3 releases https://review.openstack.org/613145 | 14:59 |
| *** diablo_rojo has joined #openstack-tc | 15:00 | |
| fungi | http://lists.zuul-ci.org/pipermail/zuul-discuss/2018-October/000575.html may be of interest for those not following the zuul-discuss ml | 15:00 |
| *** dansmith is now known as SteelyDan | 15:01 | |
| cdent | fungi: that seems like a good idea | 15:19 |
| fungi | basic idea being we could attempt to more fairly distribute ci resource allocation by project or by project queue group | 15:21 |
| fungi | making it harder for any one project/team to monopolize our available quota at saturation | 15:21 |
| cdent | this will super fun for nova | 15:22 |
| openstackgerrit | Lance Bragstad proposed openstack/project-team-guide master: Add section about collecting feedback to PTL guide https://review.openstack.org/613617 | 15:30 |
| TheJulia | scas: I think many people struggle... or have not found reference or knowledge of community over project team over company... at the same time that can be a very foreign concept and easy to completely disregard. | 15:52 |
| fungi | cdent: fair queuing solutions there will likely impact tripleo the most, followed by nova and then neutron | 15:57 |
| fungi | cdent: though in probability, it will be more like fair queuing in the gate specifically will impact the tripleo shared change queue the most, followed by the integrated projects shared change queue | 15:58 |
| clarkb | http://grafana.openstack.org/d/rZtIH5Imz/nodepool?panelId=13&fullscreen&orgId=1 gives you a rough idea of the breakdown. centos there is basically just tripleo | 15:58 |
| fungi | cdent: meaning changes in the integrated gate queue will now get an even share of resources even if there are twice as many node requests for tripleo changes | 15:59 |
| clarkb | we also deployed an update to zuul yesterday that will track per job and project resource usage (in our logs) that if I have time today I'd like to start processign to see if we can produce any interesting info | 15:59 |
| cdent | that will be interesting | 16:00 |
| TheJulia | wow, that is an interesting raph | 16:00 |
| TheJulia | graph | 16:00 |
| fungi | so basically the yellow line is node requests tripleo changes and then the pink line is node requests for all the projects running jobs on ubuntu-xenial? | 16:01 |
| clarkb | fungi: that is in use accounting so not just requests, but cpu time | 16:01 |
| fungi | ahh, right. because proportional consumption would be number of nodes multiplied by the amount of time they're in use | 16:02 |
| clarkb | the nice thing about that graph is it doesn't have to fudge the multinode math, the downside is it doesn't directly tie usage to specific jobs or projects (which we should now be able to do via other methods) | 16:02 |
| fungi | so the fact that lots of tripleo changes run multiple jobs on multi-node centos-7 deployments and use them for 3.6 hours is what's really pumping up that number | 16:03 |
| clarkb | fungi: yes | 16:03 |
| fungi | er, 3.5 hours | 16:03 |
| clarkb | also lots of gate resets | 16:03 |
| clarkb | its like the perfect storm of all the bad things | 16:03 |
| fungi | got it | 16:03 |
| * fungi is glad they're working hard to remedy the situation with faster jobs, fewer nodes in those jobs and fixing the bugs causing their tests to fail | 16:04 | |
| clarkb | the graph will also be useful to tell how well we've transitioned forward to bionic | 16:04 |
| clarkb | right now its about a 1:10 ratio bionic:xenial and I expect that to flip | 16:04 |
| *** dtantsur is now known as dtantsur|afk | 16:07 | |
| *** dtruong has joined #openstack-tc | 16:07 | |
| *** e0ne has quit IRC | 16:10 | |
| *** jpich has quit IRC | 16:31 | |
| *** mriedem is now known as mriedem_away | 16:34 | |
| zaneb | what is still running on trusty? stable branches? | 16:36 |
| smcginnis | I would expect that to be the only thing. | 16:38 |
| fungi | infra still runs some tests on trusty too | 16:46 |
| fungi | as we have some services (a dwindling few thankfully) running on trusty still | 16:46 |
| fungi | we're eager to get the last of those dealt with before trusty reaches end of support from ubuntu | 16:47 |
| *** bauwser is now known as bauzas | 17:01 | |
| *** jamesmcarthur has quit IRC | 17:16 | |
| *** lbragstad is now known as elbragstad | 17:37 | |
| *** jamesmcarthur has joined #openstack-tc | 18:01 | |
| *** cdent has quit IRC | 18:03 | |
| *** mriedem_away is now known as mriedem | 18:28 | |
| *** e0ne has joined #openstack-tc | 18:35 | |
| *** jamesmcarthur has quit IRC | 18:48 | |
| *** e0ne has quit IRC | 19:14 | |
| *** e0ne has joined #openstack-tc | 19:19 | |
| *** e0ne has quit IRC | 19:23 | |
| clarkb | from earlier discussion http://paste.openstack.org/show/733154/ produced by https://review.openstack.org/613674 | 19:38 |
| *** jamesmcarthur has joined #openstack-tc | 19:46 | |
| clarkb | the breakdown is tripleo: ~50% of all cpu time, neutron: ~14% and nova ~5% | 19:47 |
| notmyname | wow | 19:48 |
| clarkb | unfortunately small sample due to only recently having logged this data, but I think it is likely representative | 19:50 |
| smcginnis | Great that you're able to get that data now. | 19:58 |
| clarkb | kolla is ~3% and openstack-ansible ~4% to put tripleo on a scale with some other deployment projects | 19:59 |
| notmyname | clarkb: FWIW, https://review.openstack.org/#/q/p:openstack/tripleo-heat-templates shows a *lot* of failures, and https://review.openstack.org/#/c/585882/ has a non-voting timeout after 4 hours. maybe these are some low-hanging fruit to look at | 20:01 |
| clarkb | looking at the new foundation level projects: airship is .05%, zuul is .1%, starlingx is .05% | 20:02 |
| clarkb | notmyname: oh there definitely is. One of the common pushbacks is "if you can't give us real numbers then how do you know we are actually doing bad" | 20:03 |
| clarkb | notmyname: we had real numbers before zuulv3 and lost them in the transition and now we have them again | 20:03 |
| dtroyer | I'm trying to get those numbers up! That low means we're not testing enough :) | 20:03 |
| clarkb | the other thing that comes up a lot is concern that all these new projects are going to use all our resources | 20:03 |
| notmyname | clarkb: normalizing the cpu time to number of patches may be a good way to compare. maybe the high cpu usage is just because a lot of activity | 20:03 |
| clarkb | which looking at it today is just false | 20:03 |
| clarkb | notmyname: ya I think there is a correlation there, but with tripleo in particular the other factors that come in are 3.5 hour jobs that fail a lot causing gate resets that run on multiple nodes | 20:04 |
| notmyname | clarkb: lol "no worries! the existing projects take 4 orders of magnitude more resources than the new ones!" | 20:04 |
| clarkb | anyways we now have fairly definitive data (that should be looked at longer term, that is the one caution its a 13 hour window) that shows airship and zuul aren't nova's problem with merging code | 20:05 |
| notmyname | that's really cool. but I'm a little sad you didn't use asciigraph ;-) | 20:05 |
| clarkb | also there are 376.8 test days captured in that 13 hour window | 20:06 |
| clarkb | if you want to completely misinterpret the data we do more than a year of testing every half day | 20:07 |
| clarkb | :) | 20:07 |
| fungi | "zuul tests more than a year's worth of patches before you even get up in the morning" | 20:13 |
| notmyname | "if current trends continue, openstack ci will consume all of the earth's available energy output sometime next tuesday" | 20:15 |
| *** jamesmcarthur has quit IRC | 20:29 | |
| *** openstackstatus has quit IRC | 20:42 | |
| *** openstack has joined #openstack-tc | 20:44 | |
| *** ChanServ sets mode: +o openstack | 20:44 | |
| *** jamesmcarthur has joined #openstack-tc | 21:13 | |
| fungi | that's rather optimistic. my projections are all saying late sunday | 21:15 |
| *** jamesmcarthur has quit IRC | 21:24 | |
| TheJulia | That is rather disturbing... Could we convert it to a hybrid of wind and solar power? I wonder if maybe we could build some sort of CI data center into the San ?Junicto? mountain ranges, and use the wind near palm springs to power it. I'll even happily drive to that data-center and push buttons. | 21:26 |
| fungi | i miss pushing buttons | 21:45 |
| *** mriedem has quit IRC | 21:52 | |
| *** diablo_rojo has quit IRC | 23:40 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!