Wednesday, 2018-07-04

*** harlowja has joined #openstack-tc00:26
*** harlowja has quit IRC00:27
zanebfungi: bwahahahaha00:43
zanebfungi: reminds me of the AWS recruiting emails. "come work for us so you can work on open source technologies. like, um, iOS."00:44
fungiyup ;)00:44
fungiheadline could just as well have been "why google believes in santa claus"00:46
fungi#startmeeting tc01:00
openstackMeeting started Wed Jul  4 01:00:01 2018 UTC and is due to finish in 60 minutes.  The chair is fungi. Information about MeetBot at
openstackUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.01:00
*** openstack changes topic to " (Meeting topic: tc)"01:00
openstackThe meeting name has been set to 'tc'01:00
fungi#topic Office Hour01:00
*** openstack changes topic to "Office Hour (Meeting topic: tc)"01:00
fungitc-members and community at large: now is the when we say things01:00
fungior more often we don't01:00
fungiyour choice, i don't judge01:00
zanebcoincidentally I am here01:00
TheJuliaThe time window doesn't seem ideal for regular discussion01:01
zanebafter an unscheduled trip to the doctor01:01
TheJuliazaneb: hope everything is okay01:01
zanebpro tip: if you drop a chef's knife, don't try to catch it01:01
fungicommitting that one to memory01:01
zaneb(not me, my wife)01:01
fungihope all's well01:02
zanebyeah, all good. just a few stitches01:03
zaneblooks like cdent is getting his wish for that discussion thread to blow up01:09
zanebI've spent all day working on an epic reply and now there are like 4 new messages that I'm sure will all warrant similar discussion :)01:10
fungii made and then deleted two starts at a straw man for "abolishing the project walls"01:12
zanebI'd be interested to hear from Nova/Cinder/Neutron folks what they think about the nodelet idea01:19
zanebI'm definitely stealing that name from ttx01:19
zanebobviously they're gonna hate it, but I want to know what are the things that'd make it difficult specifically01:20
*** edmondsw has joined #openstack-tc01:42
*** edmondsw has quit IRC01:46
*** ricolin has joined #openstack-tc01:48
notmynamefungi: abolishing project walls. like just one big ol' code repo for everything? we all just check it out and go to town wherever we need improvements made?02:01
funginotmyname: sounds great, right? ;)02:02
zanebnotmyname: that's why I've been trying to steer it more in the direction of 'abolish walls between projects *on the compute node*'02:03
fungikept heading toward reductio ad absurdum02:03
zanebI think the idea is more to stop people from installing OpenStack piecemeal (and thus denying critical mass to a bunch of important bits)02:04
notmynamewhat are the walls that need abolishing? code repos? gerrit reviewers? irc channels?02:04
notmynameah. so back to the co-installability debate02:04
fungi"abolishing project walls" was a specific phrase in kfox1111's recommendations02:04
fungiit initially came without much context02:04
zanebI don't actually think this is the way forward in general - there are actually some benefits to separate projects as far as the control plane is concerned - but I do get kfox1111's point02:05
fungiyeah, i had assumed originally that he meant socially (get rid of separate teams, no more ptls, one flat megagroup of core reviewers across all the code, et cetera)02:06
fungiso that's our hour done (and some change). thanks all!02:09
*** openstack changes topic to "OpenStack Technical Committee office hours: Tuesdays at 09:00 UTC, Wednesdays at 01:00 UTC, and Thursdays at 15:00 UTC | | channel logs"02:09
openstackMeeting ended Wed Jul  4 02:09:28 2018 UTC.  Information about MeetBot at . (v 0.1.4)02:09
openstackMinutes (text):
fungiand for those in the states, don't forget to put cheetos and pop out for captain america before bed so he'll leave illegal fireworks under your flag while you sleep!02:09
*** edmondsw has joined #openstack-tc03:30
*** edmondsw has quit IRC03:35
*** evrardjp has quit IRC03:38
*** evrardjp has joined #openstack-tc03:39
*** evrardjp has quit IRC03:50
*** evrardjp has joined #openstack-tc03:52
*** gcb has quit IRC03:53
*** tdasilva has quit IRC04:39
*** alex_xu has quit IRC04:49
*** alex_xu has joined #openstack-tc04:57
*** tdasilva has joined #openstack-tc04:58
*** alex_xu has quit IRC05:13
*** alex_xu has joined #openstack-tc05:15
*** edmondsw has joined #openstack-tc05:19
*** edmondsw has quit IRC05:24
*** chkumar|ruck has quit IRC05:30
*** e0ne has joined #openstack-tc05:42
*** e0ne has quit IRC05:53
*** edmondsw has joined #openstack-tc07:07
*** edmondsw has quit IRC07:12
*** e0ne has joined #openstack-tc07:20
*** ianychoi has quit IRC07:20
*** jpich has joined #openstack-tc07:35
mnaserEarly morning hello07:58
* cmurphy hands mnaser some coffee07:59
mnasercmurphy: thanks, much needed :(07:59
*** diablo_rojo has joined #openstack-tc08:00
*** diablo_rojo has quit IRC08:33
ttxnotmyname: I don't think we should abolish project walls. I'd like a two dimensional approach (similar to the one Kubernetes has been using in the past, although there are talks about moving away from it now): organize work based on objectives (horizontal: user experience, things-that-run-on-a-compute-node...) and based on where the code actually lands (vertical: which git repository)08:33
ttxCurrently we mostly do the latter, which makes working on things like cinder multi-attach more difficult than it should be08:34
ttxand things like consolidation at the compute node level almost impossible08:34
ttxSo for example, having a self-healing SIG that tries to make the set of projects in that space work better together is a great idea.08:37
ttxand I wish we had a backup SIG that organized the work between Freezer and karbor to reduce overlap there08:38
*** edmondsw has joined #openstack-tc08:56
mnaserttx: you might be most familiar with this but where/how is the project navigator managed08:57
mnaserAs part of my health check, a project that recently became official isn’t sure how to get added there08:58
ttxmnaser: it's part of the website managed by the Foundation web team. It's updated after release, so if that project was not part of te last release it's normal that it does not show08:59
mnaserttx: I figured as much about the site being managed by the foundation. Let me see if it was included09:00
ttxmnaser: which project is it ?09:00
mnaserttx: blazar09:00
*** edmondsw has quit IRC09:01
ttxYes, should be included...09:01
ttxwas in Queens09:01
ttxI can relay the ask09:03
ttxbut we are working on a new version that is more directly tied to the map anyway09:05
mnaserttx: if the work involved isn’t too much I’m sure the team would love to be seen there09:12
ttxit's totally something that was missed during the Queens update and shall be fixed.09:17
*** lbragstad has joined #openstack-tc09:47
*** lbragstad_503 has quit IRC09:47
*** lbragstad_ has joined #openstack-tc09:59
*** lbragstad has quit IRC09:59
*** cdent has joined #openstack-tc10:31
cdent[t 26rm]10:40
purplerbot<zaneb> obviously they're gonna hate it, but I want to know what are the things that'd make it difficult specifically [2018-07-04 01:20:51.795447] [n 26rm]10:40
cdentwhy obvious? I suspect that is the shared brain potentially hating it, not all individuals10:41
*** edmondsw has joined #openstack-tc10:45
*** edmondsw has quit IRC10:50
*** dtantsur|afk is now known as dtantsur11:10
* fungi imagines the shared brain swimming in a jar of pink goo and communicating through a gravelly speech synthesizer11:50
fungithe shared brain hates all things with appendages11:50
cdentgroupthink is a powerful force, especially in nova11:52
persiaInterestingly, groupthink is influenced nearly as much by what people outside the group think of the group as what the group actually thinks.11:54
* cdent thinks persia has a set of modern-communication-buzzwords set up as irc pings11:56
persiaRather, my IRC client doesn't do channels in the conventional way: I see all traffic over about 150 channels as a single infostream.  "highlights" happen through keyword recognition by my brain, rather than algorithm.11:58
ttxew sounds like a full time job11:58
persiaThat my brain happens to see "modern-communication-buzzwords" as "interesting", such that I inform my client that I want to see the context of the statement (and maybe reply) is perhaps telling about me.11:59
persiattx: It's just a comms channel choice.  Other folk have full time jobs reading all the mail on multiple mailing lists :)12:01
ttxyou see right through me12:02
* fungi searches the cosmic background radiation for grocery lists from parallel universes12:03
fungieveryone needs a hobby12:03
cdentpersia: I know you don't do much email, but you might be intrigued by some of the discussion on the thread around . Both forward and backward in time.12:03
cdentfungi: are you sure those aren't grocery lists sent to you from the future instead of other universes? Your descendants are trying to adjust your diet.12:04
fungithat explains all the algae12:06
* persia uninstalled his email client in disgust a couple weeks ago due to inappropriate licensing changes, and hunts down a new one12:06
*** edmondsw has joined #openstack-tc12:33
*** edmondsw has quit IRC12:38
mnaserhow difficult is it to enforce PTL +1s for changes to certain repos13:05
mnaserlike requirements or releases13:05
mnasercould we have a job that gets the gerrit change, looks up the ptl and checks for a +1 and then we can recheck on that? (though that sounds like a lot of wasted resources)13:06
cmurphywhy does that need to be automated?13:07
mugsieI think we are better using humans - locking it down to one person can get messy13:23
mnasermugsie: because some teams are overloaded and i feel bad enforcing more stuff on them13:31
mnaseri.e. "please make sure you get a ptl to +1 for requirement changes affecting said project"13:32
mugsiemnaser: oh, I know - but we can paint our selves into a corner by forcing a "ptl +1" for these teams if the ptl is awol / different TZ / is blocking for $reason13:32
cdentis this specifically about requirement changes? if so, why?13:34
cdentas in: what makes them special?13:34
mnasercdent: in this case..
mnasera change to stable requirements didn't involve the team, it broke stuff for a while till they found it13:39
mnaserand the revert took ~2 days to merge13:39
mnaserso 2 days to merge revert, bad stuff merged since june 2513:40
cdentoh you mean changes to the requirements repo. how would you identify the relevant PTL?13:41
*** zhipeng has joined #openstack-tc13:42
cmurphyseems like a prerequisite to automating that check would be making a policy for the requirements team to get a ptl +113:45
cmurphybut yes who the relevant ptl is is probably not always straightforward for requirements13:45
mnaseryeah in requirements it's a pretty hard thing13:49
mnaseri told the team in question that moving forwards if they're blocked maybe they can reach out to the tc and we can ping the people necessary to get it moving13:50
*** zhipeng has quit IRC14:07
*** edmondsw has joined #openstack-tc14:22
*** edmondsw has quit IRC14:27
*** lbragstad_ has quit IRC14:34
fungiinfra team has a similar sort of policy for job configuration changes, where we get the relevant ptl or any of their designated infra liaisons to +1 changes, but it's similarly not always straightforward to figure out so really only a best effort14:36
fungii don't think i would try to automate those sorts of determinations14:36
persiaIf there is interest in such automation, asking teams to contribute tests that test if changes to infra, requirements, etc. affect them is probably the least painful way to do it, rather than trying to test whether a useful discussion happened between humans.14:40
mnaserthat's pretty valid too14:44
mnasermaybe my approach wasn't the best, but it was an idea.. i certainly dont want to bother requirements team even more about figuring out impacts14:44
*** hongbin has joined #openstack-tc14:51
persiaI don't mean to shoot down your idea.  Ideas are good.  I just think tracking human interaction is a hard problem with human intelligence, and a harder problem for automation.14:52
fungiyeah, i think down the road where the yet-to-be-named pilot project managing community infrastructure will want to not be expected to know the configuration policies of every tenant it's hosting, so will want to delegate a lot of those decisions to teams in those other projects to decide (and they can automate some of those determinations or not as they desire)14:52
scasenforcing the ptl to do something depends largely on the ptl in question. due to the relatively low volume of chef-related changes, i try to weigh in on most every related patch i see, warranted or not15:23
scasthat said, some changes to escape my purview if i'm not tagged on them15:26
ttxsame for releases, we ask for a ptl or liaison +1 to approve release15:33
ttxit's part of the reviewer job to check for that15:33
*** e0ne has quit IRC15:34
scasthat reminds me. at some point, i need to bloviate about releases15:34
fungiis that short for webloviate?15:34
scassounds like some sort of javascript mixin15:35
scasbut, sort of15:35
cdentI'm _still_ undecided about adjutant.15:42
scasthere was an email thread some time ago about chef's release process. rather, its lack of a modernized workflow using the releases repo. being that i'm hamstrung with natural progression and people growing as people do, anything that looks like it puts more work on me gets a stern glare from my cheshire cat. i do, however, want evidence of chef's aliveness to be where i can get it. even our friends15:44
scasfrom the gfw have shown minor interest, if only minor due to visibility. but, i want to deliver my artifacts in a manner as kolla, delivering to my own destination, just not with individual tarballs posted. acts like kolla, looks like openstack-ansible, seems to be the logical option. however, i have to account for 15 individual code repositories15:44
scasbeing that i don't know what i don't know, what i'd need is guidance on if that's a thing15:45
scasi apologize in hindsight for my extra wordiness. blame my past.15:48
scasthe reason for the wall-o-text is that i'm looking to release another major revision sooner than later, and don't want to be the one still manually cutting releases16:01
*** edmondsw has joined #openstack-tc16:10
*** edmondsw has quit IRC16:15
*** e0ne has joined #openstack-tc16:21
*** e0ne has quit IRC16:25
*** shervy has joined #openstack-tc16:43
*** shervy has left #openstack-tc16:46
*** jpich has quit IRC16:47
*** dtantsur is now known as dtantsur|afk16:56
*** e0ne has joined #openstack-tc17:08
*** edmondsw has joined #openstack-tc17:59
*** edmondsw has quit IRC18:03
*** e0ne has quit IRC18:36
*** edmondsw has joined #openstack-tc19:48
*** edmondsw has quit IRC19:52
cmurphyscas: my understanding is that everything under openstack should be released by the release team using the releases repo, as per "Releases of OpenStack deliverables are handled by the OpenStack Release Management team through the openstack/releases repository. Official projects are expected to relinquish direct tagging (and19:55
cmurphybranch creation) rights in their Gerrit ACLs once their release jobs are functional.19:55
fungisounds like maybe he's also interested in creating chef-ecosystem-specific release artifacts which get published to a chef-ecosystem-specific place?21:07
*** cdent has quit IRC21:13
scasthe repos are the artifacts, actually21:13
scasit's just a matter of how they're presented to the userspace tooling21:14
scashaving a link to a tarball isn't super valuable in context. it can be made work, but is almost worse than doing nothing at all21:14
scasbut, yes. the idea would be to publish them to the chef-specific place called, unsurprisingly, supermarket21:15
scasthat part is the trivial one, from what i've been able to ascertain. the releases repo, however, is the big unknown since i am responsible for a bulk of the chef-specific changes21:16
*** ricolin_ has joined #openstack-tc21:18
*** ricolin has quit IRC21:21
*** edmondsw has joined #openstack-tc21:36
*** edmondsw has quit IRC21:41
*** hongbin has quit IRC22:19
*** ricolin_ has quit IRC22:22
*** edmondsw has joined #openstack-tc23:25
*** edmondsw has quit IRC23:30

Generated by 2.15.3 by Marius Gedminas - find it at!