Monday, 2019-02-25

*** bsilverman has quit IRC00:42
*** jamesmcarthur has joined #openstack-tc02:17
*** jamesmcarthur has quit IRC02:59
*** jamesmcarthur has joined #openstack-tc03:00
*** jamesmcarthur has quit IRC03:05
*** whoami-rajat has joined #openstack-tc03:12
*** jamesmcarthur has joined #openstack-tc03:30
*** jamesmcarthur has quit IRC03:36
*** jamesmcarthur has joined #openstack-tc03:45
*** jamesmcarthur has quit IRC03:51
*** jamesmcarthur has joined #openstack-tc03:53
*** jamesmcarthur has quit IRC04:09
*** jamesmcarthur has joined #openstack-tc04:12
*** spsurya has joined #openstack-tc04:24
*** jamesmcarthur has quit IRC04:30
*** jamesmcarthur has joined #openstack-tc04:31
*** jamesmcarthur has quit IRC05:05
openstackgerritAdrian Turjak proposed openstack/governance master: Add Project Deletion goal for Train  https://review.openstack.org/63901005:08
openstackgerritAdrian Turjak proposed openstack/governance master: Add Project Deletion goal for Train  https://review.openstack.org/63901005:11
*** jamesmcarthur has joined #openstack-tc05:35
*** jamesmcarthur has quit IRC05:40
*** jamesmcarthur has joined #openstack-tc06:11
*** jamesmcarthur has quit IRC06:16
*** Luzi has joined #openstack-tc06:51
*** jamesmcarthur has joined #openstack-tc06:53
*** jamesmcarthur has quit IRC06:58
*** jamesmcarthur has joined #openstack-tc07:33
*** jamesmcarthur has quit IRC07:37
*** dtantsur|afk is now known as dtantsur08:21
*** jamesmcarthur has joined #openstack-tc08:33
*** zbr|ssbarnea has joined #openstack-tc08:35
*** jamesmcarthur has quit IRC08:38
*** e0ne has joined #openstack-tc08:51
*** tosky has joined #openstack-tc08:55
*** jpich has joined #openstack-tc09:03
*** jamesmcarthur has joined #openstack-tc09:10
*** cmorpheus is now known as cmurphy09:12
*** zbr|ssbarnea is now known as zbr|out09:13
*** jamesmcarthur has quit IRC09:14
*** cdent has joined #openstack-tc09:47
cdentgood questions evrardjp09:48
*** jamesmcarthur has joined #openstack-tc09:49
*** jamesmcarthur has quit IRC09:54
*** cdent has quit IRC09:56
*** jamesmcarthur has joined #openstack-tc09:59
*** jaosorior has joined #openstack-tc10:37
*** cdent has joined #openstack-tc10:43
*** jamesmcarthur has quit IRC10:50
openstackgerritFrode Nordahl proposed openstack/governance master: Add ``ceph-rbd-mirror`` charm and dependencies  https://review.openstack.org/63906811:21
*** jamesmcarthur has joined #openstack-tc12:02
*** jamesmcarthur has quit IRC12:08
*** jamesmcarthur has joined #openstack-tc12:13
*** cdent has quit IRC12:17
*** jamesmcarthur has quit IRC12:19
*** jamesmcarthur has joined #openstack-tc12:22
*** cdent has joined #openstack-tc12:34
mugsieevrardjp / cdent : I feel like i am writing essays in univeristy again :P12:36
cdentmugsie: sadism is alive and well in openstack12:36
mugsiethis has to be one of the more active campaigning seasons we have had in a while12:40
* cdent blames himself12:42
cdentsorry, let me fix12:42
* cdent congratulates himself12:42
*** jamesmcarthur has quit IRC12:43
cmurphyimo it looks a little brutal, you all were so desperate for any candidates at all and now they're all being given the third degree for their generosity12:47
*** e0ne has quit IRC12:47
cmurphyif i was running it would have taken me days to come up with thoughtful responses to all of these emails12:47
cdenthmmm, interesting point cmurphy12:49
cdentbut I really do think that we need a vastly enhanced culture of _writing_ than we've previously had in openstack12:49
cdentand hopefully if you've thought hard enough about wanting to be on the TC, many of the questions are things you've already thought about?12:50
cdentIf not, why not?12:50
cdentAs in: if you're not able to relatively easily construct some answers to these questions, in writing, maybe you're not the right person for the job?12:50
cdenti'd personally prefer fewer members of the TC who are facile with writing (because that's the only reliable communication tool we have)12:51
*** jamesmcarthur has joined #openstack-tc12:53
cdent(that is, a small TC overall)12:54
dtantsurfair point, even though it'll created a clear bias towards candidates who are native English speakers12:56
dtantsur(I considered applying this time, this is is one of the reasons I did not)12:56
cdentYeah, I realized that after I just said that, and thought that it didn't align well with my response to asettle, so it's tricky12:57
cmurphyeven as a native english speaker the amount of writing involved in this campaign period would intimidate me12:58
cdentBut to be a bit more clear: being facile with writing isn't necessarily the same as being correct12:58
dtantsurnot necessary indeed, but it's much easier to *show* that you're correct if you can do fancy writing without too much effort12:58
cdenttrue12:59
* cdent considers12:59
dtantsurit usually takes me several complete rewrites, a lot of time and some googletranslating to convey a non-trivial thought12:59
evrardjpthat was definitely not my goal to intimidate13:01
evrardjpI found it by far more intimidating when the campaining happened IRL during PTG.13:01
*** jamesmcarthur has quit IRC13:01
dtantsurevrardjp: for the record, I'm not blaming you or anyone. I more or less agree it's necessary.13:01
dtantsuroh, f2f compaining is even worse (incl. for the same reasons)13:02
evrardjpI would expect people to not answer anyway. But I like at least people to think about those, and it would be great to see an answer13:02
evrardjpI appreciate those who take the effort too13:03
* dtantsur ponders having a repository with questions as part of openstack/election13:03
evrardjpmmm13:03
evrardjpmaybe I did wrong,maybe I should have asked those to answer personally...13:03
evrardjpI mean, not on the ML.13:03
dtantsurwell, as a voter I like seeing the questions and the answers, so thanks you!13:04
evrardjpbut I have the impression that could be useful for everyone13:04
evrardjpdtantsur: :D13:04
cdentI don't think having a repo is a good idea. we want the questions to reflect the time they are asked in13:04
evrardjpbut cmurphy is right -- it is ... a lot.13:04
dtantsurjust thinking that if we collected the questions and then the answers in openstack/election, it would be a tiny bit easier13:04
cmurphyi like that idea13:05
dtantsurcdent: new questions every election13:05
cmurphyit means when you decide to run you know what you're getting into13:05
dtantsurcollect them 2 weeks before the application period, for example13:05
evrardjpdtantsur: so those questions would become a program?13:05
cdentThe other thing to keep in mind, of course, is that people who have the time/energy to _ask_ the questions have a sort of first moves advantage13:05
cdentI guess a repo would help that13:05
dtantsurevrardjp: a part of a program, yes13:05
cdenton the other hand I fear that the more we formalize the process the less often people will want to participte13:05
*** jamesmcarthur has joined #openstack-tc13:05
cdentand until we started the onslaught of questions and answers I was thinking the TC was mostly dead13:07
dtantsurthis is hard to predict13:07
evrardjpI would probalby answer myself differently if it was stored in git as part of a 'program for running'13:07
gmannML discussion looks more global and easy to read by large number of audience/voters13:09
*** jamesmcarthur has quit IRC13:10
cdentit's a bit like a televised debate: the questions are asked the sake of being seen13:11
*** jamesmcarthur has joined #openstack-tc13:18
*** ijolliffe has joined #openstack-tc13:26
*** jamesmcarthur has quit IRC13:29
*** e0ne has joined #openstack-tc13:35
*** jamesmcarthur has joined #openstack-tc13:50
jrollas a candidate, this edition of the campaigning period is brutal13:50
jroll:)13:51
cdentjroll: yeah, but I made you care, which == success13:51
cdentI can retire into my dotage satisfied13:52
jroll:P13:54
openstackgerritFrode Nordahl proposed openstack/governance master: Add ``ceph-rbd-mirror`` charm and dependencies  https://review.openstack.org/63906813:55
*** elbragstad has joined #openstack-tc14:01
*** jamesmcarthur has quit IRC14:08
*** jamesmcarthur has joined #openstack-tc14:25
dhellmanntc-members: could we get a few more reviews on this patch from evrardjp to open up the stein goal selection process? https://review.openstack.org/#/c/638676/14:44
* cdent tries to speed up his network14:44
smcginnisTurn the crank a few more times.14:46
cdentif only14:46
*** mriedem has joined #openstack-tc14:47
dhellmannfeed the pigeons?14:48
cdentI probalby need to feed pigeons less14:49
openstackgerritMerged openstack/governance master: Remove team fragility stats script  https://review.openstack.org/63672114:49
cdentas they may be to blame for the solar panels not performing14:49
*** cdent has quit IRC14:53
fungifor those who haven't seen it, here's the current state the osf project confirmation guidelines draft which the board will review next week: https://wiki.openstack.org/wiki/Governance/Foundation/OSFProjectConfirmationGuidelines14:54
fungicmurphy: i'm not even up for reelection yet and it's taking me days just to find sufficient time to read through those ml threads! i doubt i'll be done before elections start at the end of tomorrow14:55
cmurphyfungi: same!14:57
fungiat least maybe i can get through them before the end of the election week14:59
fungias for committing questions and answers to a git repository, that sounds like a fair amount of additional process for little actual benefit. i'd rather we avoided heaping on new processes except when absolutely necessary, but maybe that's a future campaign platform if i decide to run for another term ;)15:01
smcginnisI could see maybe having a small set of standard questions that we would like included in any nominations. But beyond that, I agree it would be too much overhead.15:02
fungiwe did try that in some elections a few years ago, if memory serves15:03
*** e0ne has quit IRC15:03
fungithough i think it was back when a lot of this was still managed in wiki articles15:04
fungieffectively the election officials at the time thought it was their job to also come up with questions for the candidates and manage/curate the debate answers somehow15:04
*** e0ne has joined #openstack-tc15:05
fungibut that turned out to be rather a lot of additional work15:05
fungimainly the cutting and pasting to interleave everyone's answers to each question15:06
fungiand trying to effectively randomize the orders of those answers so as not to introduce any unintended bias15:06
smcginnisThat sounds challenging.15:06
fungiafter that we said just take an extra week to ask the candidates questions on the ml instead15:07
gmanndhellmann: fungi sent the legacy job migration plan. I am deadline it on April 1st, hope 1. that is enough time for projects to test and 2. not so late to achieve our goal to finish that before stein - http://lists.openstack.org/pipermail/openstack-discuss/2019-February/003129.html15:07
fungiand let the q&a process manage itself15:07
fungigmann: and 3. that people don't think it a joke because of the choice of date ;)15:08
gmannfungi: heh :) i did not realize that. should choose other date15:08
funginah, i'm sure it's quite fine15:08
fungithough you should definitely anticipate a lot of pushback. there are quite a few projects whose representatives i've talked to from time to time about their legacy devstack-gate jobs and how they feel stuck because they don't have enough familiarity with whatever random pile of hook scripts and plugins they wound up strapping together years ago to begin to be able to unravel it for a rewrite, even if15:11
fungithey did understand enough of the job definition language to be able to accomplish that in a timely fashion15:11
fungibasically their integration tests are mountains of technical debt they feel incapable of tackling with their since dwindling maintainer base and loss of historical context for much of it15:12
*** zaneb has joined #openstack-tc15:12
dhellmannhrm15:12
fungithough in my opinion, that's a good reason to just scrap those jobs. if you don't have anyone who knows what those jobs actually do and why, then they're a liability anyway15:13
dhellmannright, what would you even do if they failed?15:13
*** Luzi has quit IRC15:13
fungithe answer to that usually turns out to be asking the infra/opendev and zuul maintainers to help you unravel what the job's doing, which isn't really a scalable solution either15:14
fungii mean, i've personally shelved a bunch of knowledge about how that stuff used to work15:15
fungiso the longer this goes on the less able i am to be able to remember how to troubleshoot it either15:16
fungithis sort of came up late last week when tbarron was looking to get a patch merged to devstack-gate for some problem manila was running into, and at this point we've basically frozen the devstack-gate codebase along with the shared "legacy" job playbooks which use it15:17
fungiwe also from time to time end up fielding questions about changing or extending the zuul-cloner shim, particularly from the tripleo team, who haven't been able to switch to using ansible variables in some jobs (i think mainly in third-party ci?)15:19
dhellmannoh, that came up in the release meeting last week with a question of whether we need to do a step in the process that involves changing devstack-gate to add info about new branches.15:20
dhellmannif there's a new way to do that, I'm not sure anyone else on the release team knows what it is15:20
tbarronit's like maintaining old monolithic spaghetti code15:20
fungialso to consider, it seems like a majority of the third-party ci operators out there are still running zuul 2.x deployments with jenkins and running devstack-gate15:20
tbarronyou can tweak it to fix specific issues15:20
* dhellmann prefers to write new monolithic spaghetti code15:21
tbarronbut migrating off is hard15:21
tbarrondhellmann: :)15:21
tbarronthe jobs were written by contractors when netapp started up manila and they are gone and15:22
tbarronbecause manila integrates with devstack via a plugin15:22
tbarronthere's a lot of hackery to undo15:22
tbarronwe've converted some jobs but there's still a lot to do15:23
fungidhellmann: on the branch matrix, check with the qa team. i think devstack may still use that one file from the devstack-gate repo but it needs to get moved somewhere else if it hasn't been already15:23
dhellmanngmann , fungi : thanks for working on that migration plan15:23
dhellmannfungi : ack. smcginnis, ttx, diablo_rojo_phon , evrardjp ^^15:24
fungipretty sure it was mostly gmann we have to thank for that plan15:24
dhellmannoh, the way he phrased it I thought you helped, maybe he was just pinging both of us :-)15:25
gmannnp! :).  I am working on base patches and will take projects side testing as much as i can.15:28
fungiyes, thank you gmann!15:28
dhellmannexcellent, that's a lot of work so I really appreciate you digging in on it15:28
gmannfungi: dhellmann also grenade use devstack-gate file for defining the source/target repo15:29
fungiyes, and last i looked the effort to switch grenade to a non-legacy job had stalled several times15:29
gmannyeah, that is in progress. tosky is working on that.15:30
openstackgerritMerged openstack/governance master: Setup Release Goal Folder for Train  https://review.openstack.org/63867615:32
dhellmannfungi , gmann : it would help if you reviewed the release team process docs for RC1 week and work with smcginnis if any changes are needed: https://releases.openstack.org/reference/process.html#rc1-week15:32
dhellmannwe try to keep that up to date so we know what to do at the end of a cycle, and it sounds like things have changed again15:32
tbarronjust read gmann's email and fwiw manila has migrated even the legacy jobs to bionic except grenade jobs15:33
*** zaneb has quit IRC15:33
smcginnisdhellmann: Maybe we just need to change that to state something like "work with the QA team to make sure any end of cycle devstack changes are ready"15:33
dhellmannperhaps. although having the actual steps written down somewhere would be good. maybe that part just needs to move to a qa repo?15:34
smcginnisYeah, I do like having the actual steps written at least somewhere.15:34
smcginnisJust wonder if the release process might not be the best place for it.15:34
dhellmannthe idea is to know what to do before we get to the point of having to do it, so we don't end up with a delay because we couldn't reach the 1 person who actually knows :-)15:34
gmanntbarron: cool. I will push grenade job to bionic and then you can test it. you are using grenade base job or derived one15:35
dhellmannyeah, like I said, maybe the qa team could own that part of the doc15:35
dhellmannin $pick-a-repo15:35
*** elbragstad is now known as lbragstad15:35
tbarrongmann: manila grenade to bionic and py3 migration review depends on a grenade review that hasn't gone anywhere: https://review.openstack.org/#/c/630961/15:36
gmanndhellmann: smcginnis agree. and in last release i noted some steps for QA team and that i can push in wiki. But i was waiting to make some central QA doc place15:36
* dhellmann nods15:36
gmannqa-specs is one candidate or just keep using wiki15:36
dhellmannwhatever works is fine, as long as we keep it up to date15:36
gmannk15:37
dhellmannif it's not going to be in the repo with the code, maybe the releases repo stays a candidate for this set of instructions15:37
dhellmannand we started that document out as an unpublished text file, so you could consider that approach, too15:37
tbarrongmann: and here's the manila review that depends on it: https://review.openstack.org/#/c/630727/15:37
gmanndhellmann: ohk, that can be good choice.15:38
gmanndhellmann: smcginnis for final-library-release-week-before-milestone-3, I am on step4 (as diablo_rojo_phon mentioned to me) and will push patch this week-  https://releases.openstack.org/reference/process.html#final-library-release-week-before-milestone-315:39
smcginnisgmann: Great, thanks!15:40
gmanntbarron: thanks for link. please add the same in etherpad. meanwhile i will review your grenade patch15:41
tbarrongmann: ack and thanks15:42
*** diablo_rojo has joined #openstack-tc15:58
*** jamesmcarthur_ has joined #openstack-tc16:06
*** zaneb has joined #openstack-tc16:07
*** jamesmcarthur has quit IRC16:08
toskyit's OT here, but if anyone feels like it, you can start commenting the WIP native zuul v3 job for grenade and its dependencies (fixes fore devstack and grenade)16:11
*** cdent has joined #openstack-tc16:18
*** e0ne has quit IRC16:26
*** e0ne has joined #openstack-tc16:31
*** ijolliffe has quit IRC16:38
*** dklyle_ has quit IRC16:40
smcginnisInteresting - https://www.cncf.io/announcement/2019/02/25/cncf-launches-cloud-native-network-functions-cnf-testbed/16:50
cdentfighting words! ;) "to showcase the performance improvements from avoiding virtualization overhead."16:51
smcginnisI would think telcos would still want to run their k8s clusters on top of an OpenStack cloud. I can't imagine any major telecom operator being OK routing their central business through AWS or GCE.16:53
cdentbaremetal cloud, perhaps16:53
* dtantsur raises his head16:54
* cdent sets bait, reels in16:54
*** dklyle has joined #openstack-tc16:55
*** spotz has joined #openstack-tc17:10
*** e0ne has quit IRC17:18
*** jpich has quit IRC17:20
*** dtantsur is now known as dtantsur|afk17:23
*** bsilverman has joined #openstack-tc17:46
jrollsmcginnis: what about something like AWS outposts?17:59
jroll(on prem aws)17:59
smcginnisjroll: I wonder if that would be attractive for those kinds of folks or not. Not sure.18:00
jrollsmcginnis: I would imagine so, but dunno18:00
mugsieif they a) actually ever appear, and b) allow people to decide what random fork of the kernel they want, set CPU pinning options like isolcpus and map PCIE devices into VMs, then maybe.18:06
* mugsie is embittered by his time in telco cloud :) 18:06
mugsiebut k8s on an ironic cloud seems like a good bet18:07
cdentmugsie: s/ by.*$//18:09
mugsie:)18:09
*** e0ne has joined #openstack-tc18:44
openstackgerritMerged openstack/governance master: clarify the hand-off of chair responsibilities at election  https://review.openstack.org/63884519:11
openstackgerritMerged openstack/governance master: mention using CIVS for chair election  https://review.openstack.org/63884619:14
fungialso that article once again seems to use "openstack" as a euphemism for "virtual machines"19:14
fungirunning kubernetes on ironic _is_ running on openstack19:14
zanebfungi: y-yes, but, there's a wide range of possibilities there, from full OpenStack cloud with Ironic for bare-metal-to-tenant up to standalone Ironic with no auth used purely as an abstraction layer for BMCs19:51
zaneball of those scenarios are good for Ironic, but not all are a win for OpenStack as we envision it19:52
fungisure, but to say that running something on ironic isn't running on openstack (which seems to be happening more and more these days) is certainly misleading19:54
fungia number of popular container deployment models out there now seem to want to hide the fact that the underlying server management is in fact being performed by an openstack component19:56
fungiand instead say they're running "without openstack" when what they means is "not on virtual machines"19:56
fungithough for this particular article and the associated tooling they seem to want to prove that running kubernetes on terraform-managed resources is more performant than running kubernetes on openstack-nova-managed virtual machines19:58
zanebI agree the whole OpenStack==VMs thing is irritating, but it was the OpenStack community itself that created that perception19:58
jrollfungi: which is an even stranger thing to prove, given that terraform supports nova :)20:02
clarkbfungi: fwiw its not kubernetes on virtual machines. It is network functions on VMs vs network functiosn on containers. It just happens that the containers option relies on a cloud existing to provision the instances that kubernetes runs on and they happen to use a baremetal cloud20:04
clarkbaka something that ironic could provide to you if you didn't move your telco backend into packet20:04
fungier, right, running vnfs in containers on terraform vs running vnfs in virtual machines on openstack, i guess20:06
fungibut it could just as easily have been running vnfs in containers on openstack bare metal instances vs running vnfs on openstack in virtual machines instances20:08
fungii agree20:08
fungithough if it had been, they wouldn't be as easily able to persist with the whole kubernets vs openstack cage match paradigm20:08
smcginnisEven running vnfs in containers in Nova VMs would be a more interesting proposition than the scenario they are doing.20:11
clarkbhttps://github.com/cncf/cnf-testbed/issues/186 is the scenario they are testing fwiw20:11
clarkb(I found that to be qutie informative)20:11
openstackgerritJeremy Stanley proposed openstack/project-team-guide master: Go into more detail on procedural -2  https://review.openstack.org/63919121:15
*** e0ne has quit IRC21:21
openstackgerritZane Bitter proposed openstack/governance master: Get rid of popularity discussion in PTI  https://review.openstack.org/63804521:24
*** whoami-rajat has quit IRC21:47
*** e0ne has joined #openstack-tc21:54
*** e0ne has quit IRC21:55
*** e0ne has joined #openstack-tc22:05
*** e0ne has quit IRC22:08
*** jamesmcarthur_ has quit IRC23:17
*** jamesmcarthur has joined #openstack-tc23:19
*** jamesmcarthur has quit IRC23:30
*** jamesmcarthur has joined #openstack-tc23:36
*** jamesmcarthur has quit IRC23:47
*** diablo_rojo has quit IRC23:53

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