Tuesday, 2018-01-09

*** david-lyle has quit IRC00:07
*** dklyle has joined #openstack-tc00:07
*** dklyle has quit IRC00:07
*** kumarmn has quit IRC00:10
*** kumarmn has joined #openstack-tc00:28
*** emagana has joined #openstack-tc00:48
*** kumarmn_ has joined #openstack-tc00:49
*** kumarmn has quit IRC00:52
*** emagana has quit IRC00:52
*** kumarmn_ has quit IRC00:56
*** kumarmn has joined #openstack-tc01:02
*** emagana has joined #openstack-tc01:03
*** emagana has quit IRC01:07
*** emagana has joined #openstack-tc01:08
*** emagana has quit IRC01:13
*** kumarmn has quit IRC01:24
*** liujiong has joined #openstack-tc01:24
*** mriedem1 has joined #openstack-tc01:40
*** mriedem has quit IRC01:42
*** mriedem has joined #openstack-tc01:56
*** mriedem1 has quit IRC01:58
*** harlowja has quit IRC02:32
*** rosmaita has quit IRC04:36
*** mriedem has quit IRC04:47
*** emagana has joined #openstack-tc05:51
*** gcb has joined #openstack-tc06:58
*** emagana has quit IRC07:11
*** notmyname has quit IRC07:49
*** notmyname has joined #openstack-tc07:52
* ttx grabs a coffee08:57
*** cdent has joined #openstack-tc09:01
cdentmorning tc-members and everyone else, it's office hours time09:02
ttxtc-members assemble09:02
* ttx quickly updates the tracker09:02
ttxI wanted to discuss stuck items and find people to help push them to a conclusion09:03
ttxLike the interop / testing discussion09:03
cdentIs there anyone here besides you and me ttx?09:04
ttxcdent: looks like not, or at least not yet09:05
cdentI'm probably a bad choice on help to push the interop thing as I have a fairly strong position09:05
ttxthat is true09:05
ttxThe other stuck one is the Python PTI specific/explicit change09:05
ttx(not counting the driver team stuff)09:06
ttxoh makes me think -- did you make progress on the neutron driver / stadium situation ?09:06
cdentrelated the driver team stuff, miguel did make ...09:06
ttxISTR you were tracking that09:06
cdent... some progress, with some position conversations with "Sam from Cisco" and some rules for handling testing and the like09:07
cdentIt's unclear how much of that landed in any official capacity, but I can get an update from miguel. Last there was email on the topic (to several private participants) everyone seemed pretty happy but in terms of formals results, I suspect some stuff still needs to be written down09:08
cdentmornin' cmurphy09:09
*** jpich has joined #openstack-tc09:09
ttxcdent: morning Colleen09:09
ttxcdent: sounds good. We might be able to abandon those driver team patches after all09:10
ttxcmurphy: would you be interested in raising a thread to get any of the other two stuck proposals to a conclusion ?09:11
cdentI suspect that in order for dhellmann's concerns to be fully (and suitably) addressed we'll need to have something like those patches09:11
ttx(Interop testing and Python CTI one)09:11
openstackgerritThierry Carrez proposed openstack/governance master: Remove the extra-atcs section for trove  https://review.openstack.org/53186009:12
cmurphysure, I can try to start a thread about the interop/testing one09:13
ttxcmurphy: both basically need wider input, should not be a TC-only discussion09:13
openstackgerritMerged openstack/governance master: Add charm-interface-designate project  https://review.openstack.org/51733909:13
ttxespeciall as some of us don't have a strong opinion on it09:13
ttxwhile others in the community do09:13
ttxI'll resurrect the longer cycle thread and bring it to a temporary conclusion09:14
ttx(which is likely to be that we need to take the time to discuss those issues further, which we'll do in Dublin and Vancouver)09:14
flaper87ttx: that'd be great. It feels like the thread died without a clear understanding on what's going to happen next09:15
ttxyes. been a bit busy yesterday but planning to post today09:15
ttxany volunteer to push the "Update Python PTI for tests to be specific and explicit" proposal ?09:16
ttxThis one also needs wider discussion on the ML09:16
* cdent is currently stretched a bit thin09:17
cdentmaybe mtreinish can give it a boost himself?09:17
flaper87ditto! Would love to help here but can't commit right now.09:18
ttxmaybe, otherwise I'll ask again at another office hour09:18
flaper87mtreinish: you da' man!09:18
ttxThe other thing I wanted to cover is the Rocky goals09:18
ttxEmilienM is coordinating the selection, but we don't have many candidates09:19
ttxonly one actually, and one that is not a great 6-month target at this point imho09:19
ttxso we should review the backlog and/or find other ideas09:19
*** dtantsur|afk is now known as dtantsur09:20
cdentIs there any sense/value in goals that are based on trying to achieve alignment with api-wg guidelines? A recent one that came up was inconsistency in pagination links in collection resources.09:21
cdentIn the past we've not done much in the way of trying to make those happen, prefering a "should" instead of "must" approach.09:21
ttxcdent: I think that would be a great use for a goal, yes, if there are good targets there09:21
ttxcdent: I see goals as "should, and if not doing it documenting why"09:22
flaper87Agreed, although, have we considered what would happen if we have a cycle without community wide goals? Would that be considered a bad thing? I'd be surprised if there wasn't something that could be translated into a ocmmunity goal but, just asking for the sake of discussion09:22
*** flwang1 has quit IRC09:22
ttxcdent: so it's a good complement to guidelines that should be widely accepted09:22
cdentflaper87: If I had to guess, loss of momentum09:22
ttxflaper87: I think that would be a bad sign. I'd rather introduce a concept of a multi-cycle goal if we can't find anything that can be done in 6 months09:23
flaper87ttx: oh, mmh, for some reason I thought we had that already09:23
ttxflaper87: we don't, all the goal tracking is scheduled within a cycle09:24
*** eumel8 has joined #openstack-tc09:24
*** flwang has joined #openstack-tc09:24
ttxThe easy way to do it would be to define a S goal today09:24
ttxbasically goving everyone a 6-month heads-up before we actually start tracking it09:25
cmurphythere are a few in https://etherpad.openstack.org/p/community-goals that seem reasonable and achievable, but we need champions for them09:25
cdentI'll add a stub in there about links, for a reminder09:25
ttxThe difficult way would be to sprinkle goal tracking deadlines over two cycles instead of one, which would be confusing if mixed with other goals09:25
openstackgerritMerged openstack/governance master: Upgrade assertion tags only apply to services  https://review.openstack.org/52874509:26
ttxok tracker updated09:26
* ttx looks up backlog09:26
openstackgerritMerged openstack/governance master: grammar corrections  https://review.openstack.org/53139509:26
ttxWe could ask mikal about privsep migration09:28
ttxhe was pushing it in nova and tracknig where everyone else was09:28
ttxZun just landed support for it09:28
* flaper87 reads the list of goals again09:28
ttxWe could ask mordred about the version discovery stuff, if it would make a good goal now09:29
eumel8Good morning, tc!09:30
flaper87ttx: ++09:30
flaper87also, the mutable configuraion and the oslo.privsep ones look good09:30
flaper87IPv6 support everywhere sounds simple enough (last famous words)09:30
flaper87but I believe most services support them by now09:31
ttxI was wondering if we could not get everyone to support basic upgrade. Like “assert:supports-upgrade” level09:31
flaper87eumel8: good morning09:31
ttxeumel8: hi!09:32
flaper87ttx: do you know how many services are missing that tag?09:32
flaper87I could check, just wondering if you have the number at hand09:32
cdentWhat about "default to python3 in tox and gate"? We already _can_ do python3 lots of places, but the appearance is that it is secondary.09:32
ttxflaper87: a lot, but they are probably not that far away... more of procrastinating on proper testing09:32
eumel8If there is time I want to talk about project doc translation and what we need before as requirements from tc09:32
flaper87Remove mox/mox3 <- this one sounds easy too. I'm actually surprised there are projects still using it09:33
ttxeumel8: there is time, we are just brainstorming09:33
flaper87eumel8: this is the perfect time for that convo09:33
ttxeumel8: explain09:33
eumel8most of the documentation is now moved in project repos. The idea of the I18n is to start translating this things there09:34
* cmurphy thinks of ttx as https://i.imgur.com/lUIlU4l.gif09:34
eumel8this requires some no jobs and a prio list of projects09:34
ttxThis person looks exactly like me09:35
eumel8for example openstack-ansible has the most progress in documentation09:35
ttxeumel8: as far as priority goes, i would start with the "openstack" bucket from http://www.openstack.org/openstack-map09:39
ttxstart with the bold stuff, then from bottom of that box to top09:40
ttxhmm actually09:40
ttxstarting with openstack-user bucket might make more sense, more international users there09:40
ttxeumel8: is there anything to translate on the openstackSDK/openstackclient / Shade side ?09:41
eumel8ttx: not yet, so far. I would focus on documentation. That's more international stuff.09:42
eumel8ttx: but the map is nice, thx for the link09:43
ttxeumel8: it's the "official" map that will be used on the website / software navigator soon09:44
ttxopinionated obviously, but all maps are09:44
ttxreminds me I need to write a blogpost about this one09:44
eumel8ttx: really good09:45
ttxeumel8: so what do you need from the TC ? A list of priorities ? some blessing to get review work on the jobs prioritized up ?09:46
eumel8so, if there are no special other requirements from the tc for this plan, we could start with the first project as a PoC to see how it will work?09:47
ttxflaper87: on the "basic cold upgrade capabilities" goal, it requires someone with time and grenade/QA experience to drive it, but otherwise would make a great goal09:47
ttxeumel8: I don't have requirements... If you can't cover everything I would prioritize like mentioned above, using the map as a guideline. But to be fair, whoever translates should decide based on their interest / personal / org needs09:48
ttxAll else being equal I would start with the most deployed stuff09:49
flaper87sorry, got distracted.09:49
flaper87mmh, mtreinish again? :D09:49
flaper87wish sdague could help with this goal :/09:50
eumel8ttx: We've made a small survey and there are some few projects mentioned. But to cover the prios from the map sounds as a good plan.09:50
ttxflaper87: the QA team has attracted a number of new people, we should stop thinking about it as mtreinish + sdague09:50
ttxand provide opportunities to step up, like chandankumar did for the tempest goal09:51
ttxeumel8: so yes, just doing it sounds like the best thing to do :)09:51
flaper87ttx: well, I was not suggesting that mtreinish and sdague are/were the only members of the team *shrugs*09:51
eumel8ttx: ok, thx :)09:52
flaper87I'm aware of the new folks and I also know chandankumar :D09:52
ttxflaper87: I don't blame you, I fall guilty to it too... mostly because I didn't look into those teams recently to see who is active there09:52
ttxOK, so next steps on goals... I'll reach out to mikal on the privsep stuff, in case he would feel like driving a goal there (or at least check if he thinks it's time for one)09:55
ttxI can also ask mordred about the version discovery stuff09:55
ttxMaybe cdent can look into API-WG guidelines to see if there is anything that would make a good target09:55
cdentttx I've already added something to the etherpad, and will talk to the rest of the group09:56
ttxAny other champion-hunting activities ?09:56
cdentsince mordred is the most recent sufferer he might want to hold that one too09:56
ttxLet's sync with EmilienM at the other office hours this week09:57
*** liujiong has quit IRC10:16
*** rosmaita has joined #openstack-tc13:13
* chandankumar scrolls up!13:15
fungilooks like i missed the first really active office hour of 201813:26
fungibut i did get some good sleep, so i don't feel _too_ guilty13:27
openstackgerritDai Dang Van proposed openstack/governance master: Update policy goal for trove  https://review.openstack.org/52944713:39
* smcginnis finishes reading scrollback13:41
smcginnisSeveral good goal ideas discussed. I think all were better than the one currently proposed.13:41
smcginnisAlso wondering, kind of related to the longer cycle discussion, if we should make most goals span two cycles.13:41
smcginnisThings like wsgi kind of felt rushed to get in just to complete the goal.13:42
* smcginnis goes back to making breakfast13:42
*** mriedem has joined #openstack-tc13:54
EmilienMgood morning13:57
EmilienMttx: I'll read irc backlog in a few, meetings first13:58
cdentfungi: is there any way back from the *-dev *-ops split, or is volume a killer?14:00
cdent(in your opinion, caveat, disclaim, etc)14:01
ttxMLs are defined by their subscription, and -ops attracted a lot of ops who didn't feel like sorting the occasional ops topic in the middle of -dev discussions and other calls for help in the general ML14:02
fungiwe first split openstack-dev from the general openstack ml, right?14:03
ttxthen -ops was split from the general openstack ML as well14:04
fungii feel like the mistake may have been splitting on specific demographic lines that could lead to a schism, though i don't know how possible it is to predict that in advance14:05
ttxfungi: it was back when they were being created, havign their own ML was seen as something you need to have to differentiate your community14:06
fungiand yeah, really no idea how to unroll that split at this point (aside from the volume problem, the groups developed separate identities for themselves in the ensuing years)14:06
cdentI've wondered if creating new lists, more topical, might work, but they would need pioneers (people and topics) to populate them14:07
ttxthe only way would be to redesign our ML portfolio entirely and ask everyone to resubscribe so that everybody is affected14:07
cdentsigs list is kind of like that, but not taken off14:08
fungiit's possible there was already some "operator" vs "developer" group identities forming before the ml splits, but segregating them to separate lines of communication certainly deepens the divide14:08
cdentttx, jbryce is it too late to "provide input to OpenStack 2017 annual report". The email say jan 5, but it looks like perhaps it is still in progress and I have some input for the api-sig14:11
cdentdiablo_rojo: I see you're looking at it, perhaps you know?14:12
ttxcdent: I don't think it's too late14:14
ttxcdent: I'd say they no longer can guarantee it will make it (although I can say in confidence that that it most likely will)14:15
cdentroger that14:15
*** jroll has quit IRC14:19
*** jroll has joined #openstack-tc14:19
*** kumarmn has joined #openstack-tc15:27
dmsimardfungi: from my perspective, I'm not sure it's such a bad idea to split openstack-dev and openstack-operators -- the topics discussed (and signal to noise) is very different for the two15:28
fungiit does however seem to unnecessarily reinforce that people are either developers or operators15:29
dmsimardHowever, openstack and openstack-operators are perhaps closer in terms of audience and subscribers15:29
smcginnisHas it been an issue to have separate -dev and -ops, or is the concern that we are enforcing a division that we would rather not have?15:32
smcginnisMy feel working with the ops team is the ones that have the interest and bandwidth for it subscribe to -dev, but those that do not are happy with just -ops.15:35
cdentI think for some the concern is that devs are "allowed" to keep themselves separate15:37
dhellmannAre there specific cases of missed communication being raised or is it a general concern?15:39
smcginnisIt seems to me we've been fairly good about cross-posting where appropriate.15:39
cdentdhellmann: the topic came back up today because of general comments fungi made about the creation of zuul lists15:39
ttxthe issue is taht lists should be based on topics rather than current membership15:40
dhellmannI noticed jeblair's well-reasoned comments about not having separate lists15:40
ttxi.e. they should both have devs and ops15:40
dhellmannideally, yes15:40
fungiyeah, the zuul project is proposing to have zuul-announce@lists.zuul-ci.org and zuul-discuss@lists.zuul-ci.org, the former for low bandwidth announcements redirected to the latter for general discussions15:40
ttxthat is why I think crosspostnig is evil15:41
dhellmannI feel like we say "should" a little more often than it's true15:41
smcginnisMy feel is we need both - the general -dev and -ops type lists, along with topic specific ones.15:42
dhellmannttx: I added a proposed goal to review our dependency list to eliminate unmaintained tools.15:42
fungii mentioned in my reply that i feel like the separate discussion channels we've formed fairly early in the life of the openstack community allowed us to evolve a somewhat unhealthy division between "developers" and "operators"15:42
ttxIf you cross-post, you use the lists in a membership-centric manner rather than a topic-centric manner15:42
smcginnisThere are general issues that affect one group or the other, then there are specific things like SIGs and various efforts that could benefit from targeted communication.15:42
dhellmannsmcginnis : maybe we need an openstack-new-features list15:43
ttxyou basically discourage devs from joining -ops list and the other way around15:43
fungiperhaps the difference for me is that i don't think there should have been "two groups" but rather just a healthy mix of people writing and using software (some of whom do one or the other but many of whom do both)15:43
dhellmannI suspect the mail volume on the zuul lists is going to make it significantly easier to have a small number of lists there15:43
*** david-lyle has joined #openstack-tc15:44
fungifor what it's worth, i feel like the debian community set a fairly bad example with debian-devel and debian-user mailing lists, basically telling users to ask questions on their own mailing list and not bother the devs... end result being almost a wasteland of misinformation because the people who have teh answers are telling the people who need answers to go ask other people who need answers15:45
ttxthere is no good solution I think15:45
smcginnisI'm definitely for more targeted lists, but I would actually hate to see -dev and -ops go away.15:46
fungithe real answer is to get people to ask smart questions, file bugs and get involved in productive discussions, but it's far easier to just tell them to go away15:46
ttxfungi: the worse is when you have a -private list where devs can quietly speak to themselves :)15:46
*** rosmaita_ has joined #openstack-tc15:46
fungittx: yeah, don't even get me started on that15:46
ttxIt exists more often than not15:47
*** rosmaita has quit IRC15:49
fungiahh, you mean in general. i first thought you were just referring to the debian-private ml16:11
fungibut yes, an anathema16:11
*** hongbin has joined #openstack-tc16:17
*** jpich has quit IRC17:07
*** alex_xu_ has quit IRC17:07
*** hongbin has quit IRC17:08
*** hongbin has joined #openstack-tc17:08
*** alex_xu has joined #openstack-tc17:08
*** david-lyle has quit IRC17:23
*** david-lyle has joined #openstack-tc17:26
*** eumel8 has quit IRC17:58
*** david-lyle has quit IRC18:06
EmilienMflaper87: hey, I'm catch-up scroll-back from the last TC office hour; I saw your proposal about IPv6 goal, I think it's "just" about converting some existing CI jobs (based on devstack I guess) to deploy on ipv6. The only question I have is for fungi / clarkb : would it be possible to have a nodepool group of resources that we're sure are running on ipv6 network?18:13
*** dtantsur is now known as dtantsur|afk18:17
EmilienMttx: I'm preparing a summary of Goals discussion and post it on ML18:26
jrollironic gets super tricky on ipv6, fwiw :(18:26
jrollin short: pxe booting only works on v418:26
cdentjroll: I'm still not quite used to you being around. Every time I see you around I have a bit of a start18:39
* persia thought PXE on 6 worked, but required special firmware (so doesn't work with nearly all hardware shipped with default configurations)18:40
fungiflaper87: v6-only nodes won't really solve the testing problem18:49
fungiwe had "v6-only" nodes in osic and jobs ran on them just fine18:50
fungimost jobs are relying on local networking over the loopback interface or through layer-2 tunnels between nodes in a multinode job anyway18:51
fungiworth noting, "v6-only" implementations usually still have rfc 1918 ipv4 addresses with v4 egress via overload nat/pat18:52
openstackgerritChandan Kumar proposed openstack/governance master: Mark the completion of tempest plugin split goal for Blazar  https://review.openstack.org/53228318:52
EmilienMfungi: thanks for answering, it was actually my question18:52
fungiso that they can reach v4-only resources on the internet at large (for example, github.com is v4-only)18:52
EmilienMI guess what we want to test is the control plane working well on ipv618:53
fungiyeah, that's entirely up to job configuration. we can't really prove v6 readiness by "taking away" v418:53
EmilienMI don't care much about github, etc, I'm more interested by if nova-conductor can schedule a VM if the mysql db is on ipv6, etc18:53
EmilienMwell, nova conductor doesn't schedule VMs, but you get my point18:53
EmilienMI meant to say, let's test a real scenario where the control plane communicate via Ipv618:54
EmilienMlike we do in Puppet OpenStack CI and TripleO CI18:54
fungisounds great to me, but it's not really an infra question. we've had the ability to do full ipv6 testing even in environments which only have global ipv4 networking, just needs people interested in writing jobs to do it18:55
EmilienMso that is the goal I guess18:55
EmilienMwrite the jobs and run them18:55
fungiwe already recommend jobs generalize above the node-level network constructs anyway because they're inconsistent from environment to environment18:55
fungiwe can't guarantee a node will have a specific number of actual network interfaces, what sorts of addresses each of those will have, et cetera18:56
fungiso using loopback interfaces and the provided tunnels between nodes is the only real consistency you can count on, and you can make those v6-only traffic if you want18:57
*** david-lyle has joined #openstack-tc19:10
*** harlowja has joined #openstack-tc19:33
*** harlowja_ has joined #openstack-tc19:41
*** harlowja has quit IRC19:43
*** cdent has quit IRC19:44
*** harlowja_ has quit IRC19:55
*** harlowja has joined #openstack-tc19:55
dhellmannjroll : hi! are you back?!19:56
jrolldhellmann: hi! yes, not 100% upstream, but I'm around :)19:57
jrollI'll even be in dublin19:57
dhellmannnice :-)19:57
fungiwoah, someone make sure i buy jroll a whiskey in dublin19:58
jrollwow, I get drinks just for joining irc channels, I love it19:58
dhellmannheh, that could get dangerous fast19:58
fungiwell, you get drinks for coming to ireland. pretty sure that's just a feature of their fair country. mugsie can hopefully confirm?20:04
jrollha, fair enough20:19
TheJuliaI have been told several times over the years that it should always be expected when visiting Ireland21:10
TheJuliaWell, specifically Guinness, unless in Cork which then it should be Murphy's21:11
persiaFor the avoidance of doubt, the water in Dublin has been considered safe for human consumption for over 50 years21:16
TheJuliapersia: Good to know!21:17
*** david-lyle has quit IRC21:40
mtreinishEmilienM: for cross project goals next cycle, what about the follow on to https://review.openstack.org/519751 and making everything follow that?21:50
* mtreinish notes he needs to rev that to remove the quotes around posargs21:50
EmilienMmtreinish: it's a good option I guess21:53
mugsiepersia: the water may be safe, but why take the risk. Whiskey and Guinness are completely safe22:09
persiaYes, that is the traditional view :)22:10
*** hongbin has quit IRC22:57
*** kumarmn has quit IRC23:12
*** kumarmn has joined #openstack-tc23:30
*** kumarmn has quit IRC23:35
*** ianw has quit IRC23:50
*** ianw has joined #openstack-tc23:51
openstackgerritSean McGinnis proposed openstack/governance master: Add Rocky goal to remove mox  https://review.openstack.org/53236123:59

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