Wednesday, 2018-04-18

*** annabelleB has quit IRC00:02
fungitime to start campaigning and debating!00:02
fungiwelcome mugsie, mnaser, ricolin, gongysh, zaneb and zhipeng! (and welcome back cdent, dims, smcginnis and ttx)00:04
zaneb let the grilling begin :)00:05
funginot a huge ballot, but plenty of great candidates nonetheless00:05
fungitc-members and candidates and anyone else who's around... it's time for another office hour if you have anything you' want to bring up01:00
funginot sure why you' needed an apostrophe, but there's no taking it back now01:00
mnaser(maybe if we used slack we'd be able to edit our messages, hah.)01:01
fungiif you used slack you wouldn't have to see my typos anyway ;)01:02
*** dmsimard is now known as dmsimard|off01:06
TheJuliaThe idea of slack seems... Disturbing.01:09
persiaI like the idea of automatic pastebins, infinite scrollback, wide character support, broad platform support, etc.  Less ideal is lack of federation and housing of all logs in a single jurisdiction.01:13
fungiand... lack of free software01:15
*** harlowja has quit IRC01:16
zanebI once created a slack account (reluctantly) to help with a customer issue. then I realised that if you don't respond to pings fast enough, it emails you to make sure. Nnnnnnnope. Deleted the account the same day.01:16
persiaOn the campaign period: I wonder if it makes sense for those TC members whose term is not expiring to lead the candidates in debate: given that some of the candidate platforms describe specific agendas: this might be an excellent time to start discussion on those issues.01:17
TheJuliazaneb: impressive! :)01:19
TheJuliapersia: that is a good point, although sadly I've not had a chance to read all of the announcements. :(01:19
fungiconveniently, we've got them all aggregated in one location for your reading pleasure01:20
TheJuliaOne thing has been in the back of my head this past week, but I'm fairly sure that unless there is willingness for regular TC turnover, that it wouldn't really be accepted01:20
TheJuliafungi: \o/01:20
persiaWhat's the thing in the back of your head?01:21
TheJuliaterm limits01:22
TheJuliaFrom the standpoint of spreading knowledge and community, it could be a tool to do just that01:23
fungilately we've been turning over ~50% of available seats at each election01:23
TheJuliaThen again, I regularly state that I am crazy.01:23
persiaMy impression from each TC nomination period is that many of those who have previously served are reluctant to nominate themselves.  I don't think term limits would help yet.01:24
TheJuliaI was looking at recent numbers, I seem to be missing a email or two from civs though :\01:24
* TheJulia shrugs01:24
TheJuliaIt is just a crazy idea01:24
fungi50% seems to me like a good balance between continuity and sharing the responsibility with new leaders01:24
persiaOn the other hand, it may be worth an analysis of past elections, to see how many cases where incubants were nominated new folk were not elected (and the new folk were also not elected in the following election)01:25
fungii don't think term limits are crazy01:25
fungibut i do worry that we could paint ourselves into a corner with not enough continuity if we had too many people who could serve and had working knowledge of the tc were forced out by policy01:26
TheJuliaWhich is actually the major source of hesitency in me even raising it01:27
persiaI think the staggered election cycle is enough for continuity, especially if folk previously on the TC remained involved in governance to some degree, and were permitted to stand again after some period (e.g. one can onlly serve on the TC for two years, after which one must not stand for election for at least one year or similar).01:27
fungithough at least the fact that we only put half the seats up for election gives us some guaranteed continuity01:27
fungior, yeah, what persia said01:28
persiaMy concern is whether there are enough folk who have the freedom to be able to fulfill TC duties if we force-expire some of the folk who have found a niche that allows them to serve.01:29
persiaIf we regularly had 15-20 candidates, I'd probably be in favour of term limits.  With half that many, less so.01:29
fungiso i agree, _if_ people with history on the tc had the leeway to assist even when no longer serving, it could work out01:30
persiaHow do you mean "leeway"?  Do you mean "time to engage in discussion" or do you mean "retain certain structural permissions"?01:30
fungimore like granted the time by their employer/life situation01:31
persiaI think it would be valuable to try to cause that to be true independently of whether people had previous engagement with the TC.01:32
persiaI do not believe we can cause new folk to be able to join the TC unless they can find the time to be involved in governance.  This suggests we need to find a way to help funding organisations appreciate the need to spend time involved in governance, independent of activity in an elected position.01:33
fungii consider myself lucky to have a lifestyle which permits me to seek out these sorts of engagements but i respect that i'm almost certainly in a minority01:33
persiaMaybe this is some badge or something we can grant for active reviewers in the governance repos?  Maybe something similar for specs repos for project teams?01:33
persiaAnd maybe have some copy that helps funding organsiations appreciate that folk who have such indicators can help them achieve their goals in openstack in ways that others may not be able to do as effectively.01:34
zanebit's rare to see a current TC member run again and not get re-elected... but of late people seem to be doing a good job of rotating themselves out, so I don't think we need to regulate01:36
fungiwe've been asked in the past to highlight people who are in need of recognition, so yes people getting (or staying) involved in governance even when they're not elected/electable would be a great choice01:38
TheJuliazaneb: so what got me on the whole train of thought was a discussion of recognition and a complete stranger telling me all sorts of things because I looked like his former next door neighbor.01:41
*** spsurya has joined #openstack-tc01:52
*** ianychoi_ is now known as ianychoi01:57
fungione of our more productive office hours, especially for this timeslot. thanks everyone!02:00
*** dklyle has joined #openstack-tc02:31
*** ianychoi has quit IRC02:44
*** ianychoi has joined #openstack-tc03:20
*** harlowja has joined #openstack-tc03:43
*** harlowja has quit IRC04:12
*** dklyle has quit IRC06:32
*** dklyle has joined #openstack-tc06:32
*** ildikov has joined #openstack-tc06:43
*** mordred has quit IRC06:45
*** dklyle has quit IRC06:47
*** mordred has joined #openstack-tc06:49
*** jpich has joined #openstack-tc07:25
*** mugsie has quit IRC08:09
*** mugsie has joined #openstack-tc08:21
*** mugsie has quit IRC08:21
*** mugsie has joined #openstack-tc08:21
ttxOn the number of candidates... On one hand 10 candidates for 7 seats is not that much, on the other they are all strong candidates. I'm also kind of happy that we have 3 candidates from China/Taiwan. Given that they had trouble getting elected in the past with the popularity-contest style of our election, I'm actually happy that we don't have 15 candidates -- increases our chances to have representation08:55
ttxfrmo that region on the future TC :)08:55
cmurphy++ that will be really good for the community08:56
ttxRe: term limits -- they are useful to create turnover, but if you do have turnover already, mandatory turnover can be more harmful than helpful.08:56
ttxIt feels like we have healthy turnover over the past election cycles, so I'm not sure we really need to add term limits right now...08:57
ttxI guess we could have a rule that states that if we don't have at least 2 open seats (people not running for reelection), we force the oldest person out.08:59
ttxThat way it only triggers if you don't already have turnover.08:59
ttxso no term limits, but minimal guaranteed turnover.09:00
ttxbut again, not sure there is anything really broken there today that we need to fix.09:03
*** cdent has joined #openstack-tc09:35
mugsieI think in our current situation, term limits may not be needed anymore. They may have helped in the past, when we had a lot more candidates, and less TC turn over, but now we seem to be doing a good job of self management10:02
cdentI think the conversation earlier nailed it: people need to be more able to find and justify time for governance, even when not (yet) elected10:13
cdentin addition to that is: people need to feel like governance is a worthwhile use of time10:13
cdentmost of the people I spoke with in the run up to this election cycle did not feel that was the case10:13
mugsieyeah, it is a hard sell. and in the past we did not do a good job of welcoming people with things like governance patches. I think that is improving, and we should hopefully see a pay off in the medium term, where people a - feel empowered to get involved, and b - feel that there is actual result from engaging with the process10:19
cdentb is critical10:22
*** zhipeng has joined #openstack-tc10:27
mugsiewe do need to work on b10:27
smcginnisI agree with the idea that if we were not to have any open seats, then we may want to do something like make the longest serving member take a 6-12 month break.10:28
smcginnisBut I also agree it could cause more harm than good with the current state of things otherwise.10:29
cdentbalancing continuity with freshness is complex10:30
mugsieIn previous organisations I have worked in or helped re-org, we had term limits, but balanced it with full time staff in the org, who could act as an insitutional memory. That worked, but it required a large amount of full time staff interaction with elected leaders10:32
mugsie(something similar to a civil service in governmental style orgs)10:33
cdentin a sense ttx and fungi are kind of like that10:39
* cdent posts a board meeting topic10:39
cdentI hope it turns into a fire10:39
* smcginnis likes a good bonfire10:42
mugsieCfgMgmtCamp wouldn't let me get one of these for the conference - I wonder would the foundation be OK ?
cdentfrequently it's never the things I think matter that become a fire10:57
*** zhipeng has quit IRC11:08
*** zhipeng has joined #openstack-tc11:27
*** zhipeng has quit IRC11:34
*** cdent has quit IRC11:56
jrollcdent: +1000 on your email <312:29
jrollinteresting plot twist on the kolla-kubernetes topic:
mugsieyeah - it is12:32
ttxjroll: yeah, they have been in touch with me and apparently all parties agree on a sensible way forward12:35
ttxThe only person from "the kolla-k8s team" that was still active-enough to chime in (Rich) is actually the one who proposed it, so I'm more confident now12:37
jrollI find it odd that this is the response to concerns that changing governance might kill the project, though I'm not quickly finding if that was a tc-people concern or a kolla-people concern12:39
jrollttx: yeah, that is a good thing, and he +1'd jeffrey's proposal there12:40
dhellmannjroll: I take it as a sign that we discovered the problem too late.12:40
jrolldhellmann: discovered that kolla-k8s is effectively dead too late?12:41
dhellmannI wonder if we had known about the team structural issue earlier if we would have been able to take action that would have allowed kolla-k8s to thrive12:42
* dhellmann has to step away12:42
jrollI wonder what that action would be, without having people to do real work on it12:43
jrollfwiw, it seems to me (by rich's analysis) that openstack-helm covers the use case of kolla images on k8s well, so it seems like a good example of healthy competition, ending with dropping effort where it doesn't make sense12:47
jrollsaid analysis:
*** zhipeng has joined #openstack-tc12:48
jrollhe does have concerns about -helm switching away from kolla to loci or something else. I also think that would be fine. if there's an image provider out there that's better than kolla, I don't see why we'd want openstack-helm to stay on kolla images for the sake of kolla. it should be about what's best for openstack as a whole, not a single project.12:49
dhellmannjroll : if kolla-k8s hadn't been under the kolla umbrella, would helm have even been created?12:53
dhellmannbut yes, I agree that competition is fine and deciding to end a project is fine, too12:54
dhellmannsome of the history with this particular team leads me to think if we had taken different action at the start we might have different outcome now; not necessarily better or worse in terms of the tech but perhaps in terms of the social aspects12:54
dhellmannperhaps *better12:54
jrolldhellmann: I would hope kolla-k8s-but-not-kolla would have used helm instead of reinventing the wheel, but that's a good question :)12:55
dhellmannsorry, I meant openstack-helm, not helm itself :-)12:55
dhellmannIIRC, that project was started in part because of social issues dealing with the kolla team. Those issues obviously have 2 sides and I don't know all the details, but I wonder if they would have had the same influence if there were multiple PTLs or teams involved.12:56
dhellmannmaybe I'm mixing up openstack-helm and loci, though12:56
jrollI would hope that kolla-k8s, if outside of kolla governance, would have eventually turned into openstack-helm, though12:56
dhellmannquite possibly12:57
fungiwell, there's also work to get loci images working with kolla-ansible, so i don't see it as just a scenario for helm... it could be possible for one orchestration solution to support both i suppose, if they want to12:57
dhellmannevolution of a project in place feels less disruptive than disbanding one in favor of another12:58
dimsjroll : the design of kolla-kubernetes was heavily influenced by how kolla works ... so doubt it12:58
dimsthere would still be 2 sides12:58
jrolldhellmann: but yeah, I'd have to look far back in history to really have a good answer to the social question, and I don't have time to do that today12:58
* dims has been talking to both openstack-helm and kolla-k8s from the beginning12:58
dhellmannnor I, so I'm just wistfully pondering12:58
mnaser(and then you have loci)12:59
mnaserwhich seemed to aim to be the more generic image that would ideally be consumed by everyone afaik12:59
dimsmnaser : that was an attempt at reconciliation . so yes12:59
* jroll <3 loci12:59
dimsdhellmann : in the end folks don't have time to work on kolla-kubernetes or have moved on to other things ...13:00
mnaseri would have loved to see a 'generic' form of images which then is consumed by different downstream deployment worlds.. in an ideal world13:01
mugsieloci is pretty generic isnt it?13:02
mnaseri think it aimed to be, but there seemed to be push back from the kolla teams to consume them13:02
dhellmannit's not up to the kolla team what sorts of images other teams consume13:02
dhellmannthis is specifically an area of "competition" and differentiation we encourage13:03
mugsiewell, it doesn't have the kolla JSON API stuff alright, so kolla probably wont use them, but other should13:03
dhellmannit has always rubbed me the wrong way that the kolla team tried to be the only container team13:03
dimsmnaser : see thread from about a year ago -
jrollmnaser: that's a bummer, I had thought it was fairly generic as well (and lightweight!)13:03
mnaserjroll: they are!  but i guess they don't suit kolla's needs13:04
mugsieloci seems to produce more traditional container images, with mounting config etc13:04
mnaserdims: 100% agree with your questions which make it that kolla is "speciality" images13:04
fungimugsie: there was a thread on the ml a week or so back about extracting the "json api" and associated start scripts so that kolla-ansible could use them with loci images13:05
jrolldhellmann: +100013:05
mugsieyeah - but that JSON api isn't even doc'd yet ...13:05
mnaserimho that "json" api should be in the deployment tool, not the images13:05
mugsiemnaser: ++13:05
dimsmnaser : there was at least a couple of folks asking "can i use loci images with swarm or just docker to stand up some standalone services"13:08
dimsi take that as progress13:09
dimsbut then there are very few folks working on loci too, which is a bummer13:09
dimsfor example we can evangelize say cinder with noauth running in a docker container as a thing folks could use for some scenarios13:10
mugsiewhen I was playing with it recently, it just seemed to work, with no major issues - it is just stable?13:10
dimsmugsie : possibly13:11
mnaserdims: i think the problem is that it solves one half of the problem and people want the full solution13:12
dimsit's more about getting folks not working on openstack-helm to kick the tires i fues13:12
mnaserno one wants to write their own deployment tool just to consume loci (i guess)13:12
dims+1 mnaser13:12
mnaserit's easier to just use kolla-ansible or $your_favorite_deployment_project13:12
dhellmannperhaps we should be encouraging people to use the community deployment tools anyway. That gives us a common place to address issues like upgrade complexity.13:13
mnaserunfortunately i think people don't get so excited about deployment tools and they end up being maintained by one or two organizations (read: evrardjp email about OSA)13:14
dims+1 dhellmann13:14
mnasertripleo seems to be mostly redhat, i'm not sure about kolla-ansible or openstack-helm13:15
dhellmannmnaser : yes, I think that's a pretty big issue for us now13:15
dimsdhellmann : i'd like for example something like openstack-ansible AIO to stitch together a single node (devstack-style) environment13:16
mnaserit can already do this afaik13:16
dimsmnaser all our services running as containers in a single docker daemon?13:17
dims(loci containers)13:17
mnaserosa uses lxc containers :X13:17
mnaserso they're more os-like containers13:18
dimsmnaser : it would be easy to write a devstack equivalent using loci, but i'd prefer that we don't go down that path and just start with ansible (or another common deployment tool)13:19
mnaserdims: that would be really cool and would totally speed things up i'd imagine13:20
mnasermy concern would be openstack is really strange in that it requires a lot of bootstrapping for things like adding endpoints etc13:20
mnasermaybe a keystone service catalog plugin that uses consul or some service discovery tool (hint lbragstad :p)13:21
mnaserthat'd eliminate a whole lot of bootstrapping13:21
lbragstadyeah - that's been proposed before, but never implemented13:22
*** annabelleB has joined #openstack-tc13:40
ttxlbragstad: maybe we could give it a second life with the edge people pushing for more hands-off operations13:43
lbragstadi wouldn't mind trying to off load stuff to consul13:45
zhipengmnaser +113:45
lbragstadit seems like a way to reduce wheels13:46
lbragstadbut i haven't really had the bandwidth to prototype anything yet =/ (despite the fact that i imagine it would be really useful)13:46
dhellmannknowing that it's of interest may encourage someone who has more time to do it13:53
lbragstadsounds like a wishlist bug?13:54
lbragstadi thought bknudson brought this up in the past, but i couldn't find the context13:56
*** kumarmn has joined #openstack-tc13:58
dimslbragstad : dhellmann : found this -
*** dklyle has joined #openstack-tc14:15
*** kumarmn_ has joined #openstack-tc14:30
*** kumarmn has quit IRC14:33
*** hongbin_ has joined #openstack-tc14:40
zanebmnaser: openstack-helm is effectively a single-vendor project:
zhipengnot looking good for kolla-kubernetes as well:
mnaserzaneb: most deployment projects are, unfortunately. and (50% of the independent is a rackspace employee afaik)14:59
mnaserlooks like kolla has some interesting success --
zanebkolla-kubernetes was actually very diverse as well
smcginnis'/win 2015:06
lbragstaddims: dhellmann apparently we're terrible about persisting discussions we have in IRC15:06
lbragstadthere are all the occurances i've been able to find for mentions of consul in keystone15:07
pabelangerdims: lbragstad: dhellmann: consul is pretty neat, i wasn't aware anybody in openstack was using it.  But, I haven't actively been looking either15:15
lbragstadpabelanger: yeah - we haven't actually prototyped any sort of integration with consul, yet.. but we've talked about it a lot15:16
clarkbout of curiousity, since etcd is already a base service wouldn't it be more appropriate to etcd?15:16
pabelangerclarkb: yah, likely. I haven't really looked myself at consul / etcd in the last 2 years15:18
* lbragstad is in the same boat15:19
*** zhipeng has quit IRC15:19
dimsclarkb : not saying consul per se ... just that we need something15:30
clarkbsure, just thinking we shouldn't add a second distributed key value store with support for service discovery and dns :P15:31
*** kumarmn_ has quit IRC15:37
lbragstadclarkb: i think mordred would agree with that point
*** dklyle has quit IRC15:38
lbragstadi don't think any consul discussions have been revisited since the ectd15:40
*** kumarmn has joined #openstack-tc15:41
*** kumarmn has quit IRC15:49
*** annabelleB has quit IRC15:55
*** annabelleB has joined #openstack-tc16:02
*** kumarmn has joined #openstack-tc16:02
*** harlowja has joined #openstack-tc16:06
*** jpich has quit IRC16:07
*** kumarmn has quit IRC16:16
*** kumarmn has joined #openstack-tc16:16
*** cdent has joined #openstack-tc16:24
dhellmannsmcginnis : can I get you to +2a so the rest of the dependency management section patches will merge, please?16:30
smcginnisdhellmann: Sure, I'll take a look.16:30
openstackgerritMerged openstack/project-team-guide master: add info about lower-constraints.txt
openstackgerritMerged openstack/project-team-guide master: add some process instructions for updating dependencies
*** kumarmn has quit IRC16:40
*** kumarmn has joined #openstack-tc16:40
*** kumarmn has quit IRC16:42
*** kumarmn has joined #openstack-tc16:42
*** kumarmn has quit IRC16:45
*** kumarmn has joined #openstack-tc16:46
*** kumarmn has quit IRC16:49
*** kumarmn has joined #openstack-tc16:49
*** kumarmn has quit IRC16:56
*** kumarmn has joined #openstack-tc16:56
*** kumarmn_ has joined #openstack-tc16:58
*** kumarmn has quit IRC17:01
*** kumarmn_ has quit IRC17:09
*** kumarmn has joined #openstack-tc17:10
*** kumarmn_ has joined #openstack-tc17:11
*** kumarmn_ has quit IRC17:12
*** kumarmn_ has joined #openstack-tc17:12
*** kumarmn has quit IRC17:14
dhellmanncdent : the article you posted on symmathesy is interesting, thanks for sharing that17:21
cdentflaper87 has suggested that we gather in vancouver to chat about it17:22
dhellmannI would like for us to explore ways the TC can work together more as a team17:23
dhellmannI don't have a good way to express exactly what I mean by that, yet17:23
cdentyes, I have similar feelings, and also don't know exactly what I mean17:23
cdentin my response on the "what we should talk about with the board" thread, the second topic is driven in part by figuring out how to be more symmathesitic (if that's the right form of the word)17:24
* dhellmann reads that email again17:26
dhellmannyeah, it feels like we have some "process debt" to deal with still17:27
cdentthe idea being that upstream unicorns need to increase their generativity17:28
*** harlowja has quit IRC17:44
*** harlowja has joined #openstack-tc18:18
*** harlowja_ has joined #openstack-tc18:23
*** harlowja has quit IRC18:24
*** annabelleB has quit IRC19:00
*** kumarmn_ has quit IRC19:00
*** kumarmn_ has joined #openstack-tc19:05
*** kumarmn_ has quit IRC19:10
*** kumarmn has joined #openstack-tc19:11
*** kumarmn has quit IRC19:11
*** kumarmn has joined #openstack-tc19:11
*** kumarmn has quit IRC19:17
*** kumarmn has joined #openstack-tc19:17
*** kumarmn_ has joined #openstack-tc19:20
*** kumarmn has quit IRC19:22
*** kumarmn_ has quit IRC19:37
*** kumarmn has joined #openstack-tc19:39
*** kumarmn has quit IRC19:40
*** kumarmn has joined #openstack-tc19:40
*** kumarmn has quit IRC19:42
*** kumarmn has joined #openstack-tc19:44
*** kumarmn_ has joined #openstack-tc19:45
*** kumarmn has quit IRC19:49
*** mriedem has joined #openstack-tc20:34
*** eandersson has quit IRC21:10
*** eandersson has joined #openstack-tc21:13
* fungi is lost in buzzwords21:18
smcginnisfungi: Are you leveraging your synergies to think outside the box?21:19
fungiif i say yes will the bobs let me return to my cubicle?21:19
cdentfungi: if you don't have a reference for it already, and those are the buzzwords that are losing you, this is the article doug and were talking about: (and led to many buzzwords)21:20
fungii do enjoy going to see the opera21:20
* dhellmann pictures fungi in a Hawaiian shirt with tails21:21
fungithe lei and monocle go well together21:21
cdenta good look21:22
* cdent waves goodnight21:22
*** cdent has quit IRC21:22
*** kumarmn_ has quit IRC21:28
*** kumarmn has joined #openstack-tc21:30
*** kumarmn has quit IRC21:34
*** kumarmn has joined #openstack-tc21:34
*** kumarmn has quit IRC21:43
*** kumarmn has joined #openstack-tc21:43
*** kumarmn_ has joined #openstack-tc22:04
*** kumarmn has quit IRC22:07
*** kumarmn_ has quit IRC22:12
*** kumarmn has joined #openstack-tc22:12
*** kumarmn_ has joined #openstack-tc22:15
*** kumarmn_ has quit IRC22:15
*** kumarmn_ has joined #openstack-tc22:15
*** kumarmn has quit IRC22:17
*** kumarmn_ has quit IRC22:37
*** kumarmn has joined #openstack-tc22:37
*** kumarmn has quit IRC22:42
*** kumarmn has joined #openstack-tc22:57
*** kumarmn has quit IRC23:01
*** hongbin_ has quit IRC23:18
*** andreaf has quit IRC23:18
*** andreaf has joined #openstack-tc23:18
*** mriedem has quit IRC23:32

Generated by 2.15.3 by Marius Gedminas - find it at!