Kubernetes Community Meeting 20160609

so as I start every community meeting this is public and recorded and today is Thursday jun 9th and this is the Cooper nidhi’s community meeting that happens weekly this morning we have on the agenda a node problem detector that will be demoed and that’s with lentil oof then we have we have no why do you contribute segment this week we will begin those next week while we’re return to those next week with Huawei we have a 1.4 release schedule proposal from TJ culture Minh and then we have an update on the project or sorry product management working group that was formed and that will be from a who is here as well as David are on chicken Mike sephiroth when they join if anyone has any sig cigs reporting that they wanted to do please put that onto the agenda and we’ll get into that in a little bit and then we’ve got a couple of notices which is we are not going to have the Asia friendly timing once a month anymore we didn’t have enough people joining from time zones where it was more convenient and the next notice of course is now is the time to get your documentation into 1.3 and if you want to help us figure out how to celebrate Cooper Nettie’s first birthday please send a note to me or Bob Burtynsky and the our emails are both in the notes so with that lantau are you game to get started and give us that devil oh yeah actually thank you I’m sure miss me can you see mess right we can yes okay cool thank you Oh everyone i’m nan Chao Maggie’s have IDs randomly oh I’m from the google corn in haste no team and today the topic is about node Club problem detector is an alpha feature introduced in Coober Nick his 1.3 o summa Sarah ok son oh sorry and so shall it is just a daemon running on each node trying to detect the node problem and to the API server and the motivation of doing this is as simple haha during the last several releases Cooper was really suffering from the node problems especially other problems sometimes the colonel encountered dialogue sometimes the talker just hung there forever and when that happened a couplet can do nothing and but for now Corinne he doesn’t have any knowledge about this it will keep on scheduling newport to the bad note and the court will feel and keep on and schedule another one field keep doing this and so this problem firstly we need to make this kind of problem visible to the control plan and if we have this then we can try to divide some remedy system to remedy the band no to save the pulse on the node something like that so the node problem detector is just for the first step to make the node node problems available and now the node problem attacker is demo site planning on each node detect problems and report them to the PI server and we know that people running coordinated some various environment may be different OST shows different hardware’s and different runtimes a lot of arrows we know that no Damon can not handle all these different environment so we make the problem problem the capital totally accessible it just applied problems from different of demons and we called a sub Damon problem demo each problem Damon just a handle a specific kind of problems stuff like that are the problem Damon can be at Damon just a dedicated for the Coronet excuse case we will introduce Colonel monitor later a legion it is just a dedicated folk awareness and wood which really like to work with our community friends to integrate with your existing solution to make it make the to make it more powerful to make grunting who ended his more stable and after this in the future we may introduce the remedy controller to try to remedy the bad note

but that is the future scope enough now and for the user facing API the node problem detector will report the problem as node condition and event the null condition is just a field in the nose stickers we already have some conditions like radial node out of disk or something and in the future the note prompt problem detector will report the permanent problem which makes the note available for other animals the bottom has no condition such as Colonel devil bad disk value CPU dr. Khan something hand for the temporary problem with would also like to collect these kind of information so with a result of nice event event and this is a carbon stickers of the node calm problem detector a eighties it now it is an atom hard running a enabled by default in GC is running on each node each node and we having to introduced the first problem game on Colonel monitor it is just afore the current his urgent requirement and also as a reference of other ones it monitors the colonel log detect some known issues following predefined rules and it will report colonel dialogue condition here instead it is totally extensible you can clear it is here is the config file of the colonel monitor and see that old conditions and the mighty problem matching Titan totally config book and this list can be extended over time one way in filter we have only encounter some other doctor or Colonel but don’t panic maybe we can’t detect this hopefully and then it is a sample time because we don’t have a really don’t really have a bad note in hand so we can only inject some I logs with we have seen before to mimic the idol and now okay and first leg just either okay it is the same with the reverse one actual and you can see that wait wait now we can catch some temporary arrow such as autumn allocating and for example this one they see the permanent i will wait we have debug debug this for quite a long time to unicorn is one point release it’s a kernel bag it is still there or not but we have a work on so in filter if this hadn’t heaven at least we can detect this problem and maybe avoid to schedule calling on to the node in the filter and some other iosys guys dr. Hong ifs more corn something okay and the demo script is just a single strict squid that went for the demo and weight already had I study this class or yesterday we have phone off running and wave or they study either no problem can’t receive any ice when I snowed and let’s just care about a random select one this one let’s just care about this one we can see that now we have a new kernel dialogue condition it is false and it means that the car no votes will and there is no event now firstly lies inject a meaningless love this log is full of Colonel Locke if the knowledge granita is running daughter and if we inject if we inject there’s nothing to heaven so let’s have a try oh wait wait wait ya know then send still falls and then let’s see let’s try to inject her a temporary oh this is an autumn mo a log and if we inject this what Enchantix into the node we should see an event yeah here is an event we captured his autumn memory and the

condition is not changed and now let’s inject the hog we suffered a lot of sorry the language is planned register net devised carnival and lesson jacket and the node condition should be the colonel dela condition should be changed let’s check it yet if I see that the condyle of dialogue now it comes true and with the reason and logic or white issue as we can today on your voltage and let’s have another try to inject ER doctor home is if dr. harmful to two minutes the colonel will throughout this caustic and let’s try to inject this and the condition should be changed yeah you can see that still colonel dialogue and within stalker Hong and yeah with the I message yeah in fact that this does it that’s it and I mentioned that the the the configuration is totally extensible in filter we can oh we can detect more and more Colonel box and darker issues and here’s the Future fan so detective not detect the problem in just the first step after that way after that we know that there are some panels we want to we will try to ram did and in the future maybe we’ll introduce the remedy controller it just monitored the noticed a Thursday’s convert convert the problems we ought to attend latent field the 10th the 10th is introduced also in also an alpha feature introduced in 1.3 it is similar risk and schedule but not exactly exactly the same it is more powerful and after that maybe we will try to drain the node and we pull the node start a new one or something that is for Randy plan and more important more important that we were always really liked to detect more problems and to make the node problem detector more powerful and we can all achieve that without the community community health because there are so many different scenarios so many different environments so we’d really like to work with our community friends to integrate with the existing solution to work together an any feedback retirement contribution that will come really welcome and yeah if you are interested to work with us or you have any comments feedback feel free to contact us this is my email give a body and are you can also join our signals group 22 or value show animals okay and we have trapped design dog umbrella issue and now it a no problem detector is separable under the porn it is directly yeah that’s it lantau there are a few questions in the chat log if you have up or I can read them if that helps milk okay the first one is unused eamon set uh well why did you cause something like legs um by Bob Bob asked us to see yeah I was always just a clarification man I think it seems like on the surface that demon said is exactly the mechanism the you with that that was built with handle cases like this where you need to run you know one one thing per node for example um you monitoring activity so my question here is why not use daemon set for this this is tim it is dima said okay I want you did say that to begin with an extra seat it is dima same as a pod but actually there are so much concern for because the even run a team and saddle know which means we have s today’s stage so we have some dependency on the darker and we want this detector all the problem so let’s have the small chicken egg issues actually here yeah but for one point of photographer one point 12 the container will not die yeah i’ll even yeah yeah so so after one point 12 as well get that so there was another question in here of what if you

aren’t using docker is that in future plans well what’s the another date what if you wish I not see this again so yes actually I’m looking are yeah we all look under to provide the alternative or a continental for example for 1.4 have the rattling hiss 1.0 release which is a lot alternative the red container enter yes and we also talked to the oci compatible runtime support and we asked for community help and the from many different partner and so they can they can provide an excellent i’m cracking so we are working on the interface at this moment and make that cocky easier and have the mode as well requirement for each children time we expected well as the cuban ideas resin yes cool hate on it quick question is this derek other particular problems you think would be out of scope for this no problem detector to do you detect you mean what yells problems will be added in the future no I like are they are there particular problems you’ve already decided that fund that would be like not appropriate to detect using the solution just wondering if we should if the queen is looking to help other certain areas that we’ve already thought about that would probably not good fits for this problems affect their and how to actually don’t know the problem the character is just said already well I / incriminate his mold and how we are going to crack the problem and how we are going to propagate with other and the moster problem detected a is dry as a separator a beam inside we’re also talking about that maybe this is a negative is kind of a human egg and we talked about this kinda miss not having to follow last night what’s the macbeth but there’s a lot of problem and so we have only stuff as an Entenmann’s in other one who customers know the problem detector that demon said we make you could plug into different a different problem female Robin detector be met and if example for how to our issues actually no we talked to the entire day they really won’t have on the five things I think you guide the memory protection to liquid once they have that demon Randy’s we can pack into our loaded penalty exit scene inside and then we put those problem for some our baby’s kisses we also have not there are internal TKE Mexican the MIP and then we have some signal give to us so we can say okay dced knows that there’s something which one of node and there are some issues this one to make it a thousand and propagate and you also talk about chillin enter people’s community constantly have some few sauce in idiot the problem though if you seek knowledge when human it also is pad awful what possible the resources in India the problem detector yeah I’m gonna have to cut us off here the best place to go discuss this is signaled correct and if you end up to break off to another group then you guys get to do that so there was there was a comment or two from Justin Santa Barbara about other status names things like that it sounds like those should have social side into the signal group so thank you very much lenta for the demonstration and we will jump on to the next topic oh I’ll stop this out there we go thanks linty you wanted to talk about 1.4 timing yes exactly I’m going to try to share my screen and hope this works I’m usually I’m about the current mouth phone is everyone seeing that yeah I think I usually near the top of the colonel done one point I’m just as a quick reminder we’re hoping to cut beta tomorrow and then launch in two more so about 11 or 12 business days tell 1.3 if all goes well but from one sorry always I it’s a kind of loud I think someone might need to be muted but yeah did you get a volunteer to help with the fried burned down um no I didn’t yet I’m actually i should be able to okay yeah that’d be great i think basically a fridays burn down meeting is going to be thumbs up or thumbs down on beta whether or not we can cut beta is going to be the key thank you sir yes thanks right for the derail no no that’s great but then pivoting go 1.4

1.5 we want to start to set out our schedule and goals for when we can launch these things so just for I wanted to put it this is actually on our community wiki so you can see github to come Cooper Nettie’s community repro and then wiki is here so anyone can edit this I guess don’t don’t go deleting it or giving anything too crazy but just for history so we launched in July of 2015 1.1 was a just about 16 weeks later and then 1.2 was 18 weeks after that but that included thanksgiving in the US and winter holidays in a lot of vacation time so probably closer to 16 weeks again 1.3 is on track I would say for about 14 weeks so it does appear to be moving a little bit quicker which is nice so that’s sort of what things are looking like so I sort of took that as an input and the other input I took as axiomatic was that we want to have two more minor version releases this year so in 2016 we would want to have both a 1.4 and a 1.5 and so given that um it looks like we’ll have to move a little bit quicker compress the milestones and so this is kind of how 1.4 and 1.5 would lay out given those items for 1.5 I picked December 9th as a Friday that gives us about a one-week buffer to the 16th if if things slip I’ve been a week after the 16th is the 23rd and that’s probably not a really time that we’d want to release there’ll be a lot of people on vacation and things like that so basically all I did was work back from december 9th looking at weeks on the calendar and where things are and that would look like to 11 week milestones and so if we sort of do our typical between coding and the sort of bug fix and release parts that would be seven weeks of coding and nine weeks of bug fix the difference between 1.3 is laid out here so 1.3 was nine weeks of coding in five weeks of stabilization should be shaving one week off stabilization and two off of coding the two off of coding I think we stand a decent chance for because we have been better about pushing planning earlier realistically we probably did six weeks of coding in 1.3 the first several weeks were still a lot of planning and some early design and things like that I didn’t see a lot of code moving early and the five weeks of stabilization you know that’s that’s a toss-up if if we can move to four I think that we’re putting a lot of effort into flakes and submit cube fixes now and so if the submit Q stays healthy that will certainly help getting everything in the last few weeks with coding if we keep the flakes under control obviously that helps during stabilization so it’s a it’s a little aggressive on the timeline but I don’t think it’s unreasonable um that said like looking for feedback and everything the only other thing I built in was 11 week post release of 1.3 before we start 1.4 in earnest for sort of a flake week people had been bouncing around the idea of having a fix it having sort of everybody rally around fixing flakes and fixing test issues and fixing test infrastructure issues and whatever all the things we need to do to really set us up for 1.4 time in that enabling the community teaching in the community to help with that going forward yeah totally so that’s basically it the only other note is uh 41.5 I did do five weeks of stabilization and release because the Thanksgiving us thanksgiving holiday is in there but but the winter holidays would not be in we probably will start 1.6 January second something like that but looking for feedback and everything I think that we’re relatively constrained though if we actually want to get to minor releases out this year you’d been talking more broadly about moving to a time-based released as opposed to a feature-based release is that really by saying we’re going to do two more this year is that really the proposal here is that we will we will do the two more that’s the the presupposition as you said and and we will fit the features that fit into them as they come and then they will slip any feature that doesn’t make the release cut off will slip to

the next one yeah I think that’s what we we would have to full-on make the pivot to say that’s how the project works and how the project runs in 1.3 we did hold ourselves to a pretty high standard on this feature complete date saying there are several things were very close and we push them out and that’s painful at times but stabilization has been better this time than in past time so I think we’ve seen the result of that up really cutting down mature and on time so yeah I think it would it would just mean that if you miss this August nineteenth or this November fourth date listed feature complete that’s that’s it feature just slides to the next one are there thoughts or comments on that outside of outside of this specific timeline so what exactly is the definition of feature complete the definition is that we’ve said that you’re in the queue with all code required for new features plus in to end tests in 1.3 we said but not counting Docs and of course not accepting that the code could be party there could be flakes like there’s a lot of things that you discover after the feature goes in and really gets beat up by the PR builder so when you say and I am tests are you talking about and then tests running it as part of the PR builder or something that’s non gating non-blocking off to the side running as part of that PR builder like the test that you submit on feature complete date should be running continuously you know the day after victim I just feelin all upset but in the overall ed suite in general so I’m asking from the perspective of the Eric Tunes fantastic work on the so whole feature proposal workflow I was sort of curious if we’re actually like committing to doing that and using that to sort of document what feature is and what feature complete is and what the steps are to move through the process that sort of thing yes definitely um I I do want to go use that I think that will help a lot towards the end towards like you say both ensuring what features feature complete applies to of course and then also the end of milestone as we go build docs and go you know build all the collateral around announcing it will have that sort of ready in one place okay I look forward to seeing that this will also be touched upon in the product management update that we’re getting next because there’s a broader picture of this yeah yeah absolutely awesome any further discussion on this or are we good to go with these are our dates and we hold each other and ourselves accountable for them going forward for 1.4 and 1.5 if you’re on mute and family arguing then we’ve missed it and I’d say we makes make these go forward this is awesome TJ thanks hey Sarah um one thing I would just suggest having gone through big release cycles like this before it’s just a do a retro after the next release cycle and just sure that we’re not making assumptions or things that you know our future selves are not being held accountable for things that we can’t see right now for our past selves mistakes wait exactly posted you yeah so that’s that’s my my two cents yes so we generally do a retrospective after the release in this community meeting and we were not as good last time okay so 1.2 we kind of did it at hawk one or sorry 1.1 we did it at hawk 1.2 we did it with about four hours notice so people didn’t feel that there was a lot of time to give feedback and this time I will put it on the schedule right now for the week after the release I I would strongly advocate that that is needs to be a lengthier meeting with a single agenda and that we shouldn’t try to squeeze it into the you know 20 minutes I’m a community meeting it needs to okay right and I’m i happily volunteer to facilitate if that’s helpful I’m done a lot of those so but if that’s something to be useful I’m more than happy to do that so awesome thank you Jason alright so we will schedule an out-of-band or out of this meeting meeting about a post-mortem we will publish it publish information about it here so that we

know when we’re having it and get the right people invited and then we will go forward with that awesome alright any further commentary on this or are we on to the product management agenda item sweet sorry yellow don’t go ahead nope I think you know more and Bob Arum here with yes okay got it yes we can hear you okay sorry TJ and i are in the same room not projected so that’s the failure alright so as we talked about last week we want to formalize the feature process I which things get approved and into Cooper Nettie’s as a whole for those that have been paying attention on the coup benetti’s dev mailing list or at last week’s community meeting we shared many times I also tried to reach out to as many folks as I possibly could track down to ensure that everyone was aware of the formation of this group and that we are currently in the process of putting together some plans since then we’ve already engaged and started to create some high-level documents right now we do currently have a feature repository that you can go and create features in there’s a walkthrough for how to go about that and things like that um but most importantly we do have a feature proposal of which I believe all right I don’t see egoriy they’re like hey there Mikey I’m in here yeah yes so we do currently have a feature proposal and this is again how we would like to run the organization as a whole or the project hole where the community decides on the features that ultimately make it similar to some other best practices that we’ve seen across the you know open source world so I guess without further ado um would you like to present a proposal and walk us through your thinking yeah sure absolutely sure my screen yeah and I’m going to share the link to our shirt for everybody can wizard it like to link to our draft document and so no take a please send it to them so there are a number of kind of absurd on Doc there are a number of proposals that’s been around right so we started with proposal Eric and mother I believe put together back in April right this is a proposal that describes the idea of feature repository right and so we basically we created he currently created a second document we should probably go to merge into this one if everyone agrees right so this document I think there has been a number of comments and kind of it’s more about how to create the feature what should be in the feature what happens with the feature next right then so and there is a faq on why it’s important and so what we did is extended further kind of chick specifies on the features let me quickly walk beep over it right and so the document link is shared for the cone is open for the comments so there is basically it you built a kind of erotica project management to on top of github which as all of us know please get help is non-trivial and and the point of B or having it kind of well I think of a lot of people the experience was it’s wild a bunch of PRS it’s pretty hard to track here are quickly what features actually go in into the release what you said state of the features people use sticks community meetings constantly asking questions so what happened to his pad sets what happened to Z is what happens is that right so there it is to kind of as much as possible to automate this process and be able to maintain continuous dashboard on top of github features and undo I NPR’s about the status of each issue right so we start these features kind of all the features need to be created based on previous document and then when people start creating pr’s right the suggestion is instead of carefully embedding links to all the PRS into the future which kind of may or may not happen to insist that

the otrs should be labeled specifically non bug pr’s right things like feature proposal documentation and to end test they all should be contain a label with speech in its ID or something of this type so that we can build some simple code to scan all the Reapers and basically automatically link the RS into the feature set into the fee into the features that way we can maintain there are here what’s related to the future it also allows us to scan for the orphan pr’s and then signaled Lee elites should be responsible for appropriately labeling this PRS any questions concerns on that aspect of it why specifically call out John dog TRS I mean kind of we can define kind of right this is a proposal would I am open to kind of suggestions right but he is the kindest feature or discussion proposal or documentation right the ID it has to belong to some feature box maybe the box from the previous life sure I interest in the context of like the current release cycle right like there’s still a lot of bugs that are open that we’re going to squash before we get the release out so I’m trying to understand if bugs or something that are the closure of bugs is not necessary to reach feature complete or if the closure of bugs is necessary to reach feature complete so yeah so it’s kind of the below was right so if the bug Israel clearly related to the future here all right then yeah it should be labeled to the feature right but sometimes their box which come out out of previous releases and say it’s hard to Utley be contribute to current set of features being released so I don’t think those necessarily have to be labeled right that’s basically the logic behind it all the new box related to the future see as it should be labeled and it also helps with dragon kind of feature quality so to speak it’s a bugs don’t necessarily block a feature but issues that are labeled with the feature labeled do bloco feature yes okay so I mean the it’s scalable only up to a point right so we have to rely a lot on sig leads or if they have a lieutenant to whoever they kind of heaven zc great who is responsible to foreign force in this behavior right the alternative is just automatically not merge the any would any PRS that don’t have related Fisher label so we if this is in fourth kind of its regĂ­strate forward to create dashboard which features a part of the milestone and what’s the progress in each of the features such as proposal kind of being accepted proposal in the discussion codon is in progress and so on and so forth right which will give visual indicators to what’s going on with release and we can maintain its own way online so that people stop asking a continuous questions was going to be in the release what’s the status of this particular feature so the other things kind of and that shouldn’t relief with the dates at TJ proposed some of the proposals observations from one to three and prior to that right all features must have an owner and committed implementers bye week for this is arbitrary date we can do weeks three week why if I know that stopped discussion is TJ and community basically its future proposal freeze what’s happened in one two three is there a bunch of features which came out of cigs they will mark t0 and then half meets through the cycle it turned out that nobody’s implementing them because resources were moved so to avoid this situation and then all PRS with actual implementation proposals need to be lgtm bye week seven and the idea behind it there’s limited number of maintainer pretty busy people so if you can front load all the approver approves of proposals it will help tremendously this can throwing away features at the end which royal GTM proposal was lgtm like one week before the code freeze before the fish fries any objections to kind of enforcing that type of discipline quadrature where are we non core project members being able to add labels given it is a requirement of proposal yeah I mean so the sig leads they typically can add labels right or we need so we are also working on this owners proposal

from Brian right so hopefully they’ll be my own or squirrel to add features to the particular report yes my queen is that it’s going into the features repository which we will be able to make much more permissive so that people will be able to add their own labels to it so we will start adding adding more people who have permissions on the features repository as in not a free-for-all but close yep OPRS with code still near look could you comment on large features that might get implemented over time in kind of a phase delay some people might call that an epic you might have a big feature that is open over multiple releases or there’s the attempt pop here that should get broken optimistic I think the intent is kind of right we can have a feature and call it epic as you pointed but then have sub features which fit into single release rate because if they start spawning over the release it’s against hard to track what went in what went out ah so I I would I would concur and agree that that would be the way to do it it’d be nice to see that spell out here in the proposal okay well okay um I’d like to come in on that label issue since we moved to restricted branches we should be able to more freely give out the right access to the crew benetti’s rizzo because the people who can actually merge code is not the people with push access so we probably can work to getting more people the ability to add labels I agree Eric but anybody that can add labels can add lgtm which will cause things to get merged I would like the owners implementation finished before we start doing that okay I mean I share the concern but I think the label is kind of requirements are relatively minimal and I think kind of cig lace can be able to be able to address labels in the few weeks was a release cycle sir alex karen is very scale problem as the number of labels decreasing get up here just wondering that label drop-down selector a huge yeah switching the github doesn’t have any other mechanism right to track anything the question in the community does does anyone use sin hub or an alternative like overlay to github I’ve heard generally good things but I you know I wouldn’t want to explore unless we had kind of consensus we tied up internally and then number of limitations which don’t scale for larger projects okay I mean I’m frozen I don’t know we tried it a couple months ago maybe some hub keep suppressing them they also require full right access because they store all the metadata is a github it sells we also think that we kind of have made a decision all along that we need to enable people to use vanilla github like just random people who run up to the project and so the concern was a lot of these tools is that they that while they provide utility they kind of make in ellicott hub impossible to use I know this proposal extend for the other three pros that are underneath she would sell free and independent alone so like getting a feature in the sea advisors follow this path each stir maybe even another problem detected it was just entered Britain evidence other than any cheaper to do them I’m sorry I couldn’t hear your question very well if somebody else could repeat it I think we discover the problem detector for example or other sub repose of the Cure netezza like we are we’re just we’re scope right now just to work thinking about the teacher repo we’re obviously open if it works here or any other stood or if any other remote or would like to adopt similar tools techniques and so on that’d be great but let’s let’s write during this talk about this yeah I’m just kind of had like a feature like like signo much XO again adding device not for accounting supporter right and majority that work is in the advisory so I’m just wondering like do we have to plan everything around all these can’t mentally throws with this in mind order it like speak like the labeling would have been kind of not well served on that particular future so I’m just trying to think through that yeah I mean this is not this is neither process for processes say ignored do we have to have like a hundred percent like every you know one hundred percent solution on day zero our goal is that it should be the

exception not the rule orcses me the rule not the exception that everything follows this guy and you know if if it makes have an exception or something like that if the community decides that hey it’s okay then fine you know we we’re not worried about this but right now we have no process no rule whatsoever we’re trying to kind of take our first MVP at this yeah and once again that is kind of to simplify trican Watson’s a particular release and what’s the status of the particular features is right if some things if it goes like nineteen ninety-five percent of the PRS and features I think it would be great you can deal with the rest of it out of them so in terms of the PRS with proposals line that I’m looking at right now does that mean that I wanted something in release one for that I need my Ozel merged into the mainline repo bye week 4 or week 7 in order for it to still be counted for release 14 I mean in this one case it’s week seven so it’s a so I guess I would sort of anticipate that week seven sort of looks like it lines up with feature complete coding should have soft and we should some bug fixes so we probably want the proposals in earlier than that yeah i know no Thursday as it says in the comment the dosa kind of arbitrary weeks picked up on 16 week cycle so shorter week’s show the release cycles to the judge those weeks corresponding describing is something of a little more formal gate baked into the release process by which we say if those aren’t actually reviewed and vetted and have reached some sort of phase i don’t know if that’s alpha beta GA right but like some of it of that yeah I mean it’s people in general kind of an agreement said there should be this gate i’ll work with DJ to edit to his release 104 and 105 days yeah i don’t disagree i mean i think that the two things that are that are here the only two things that are captured here is we really don’t want to have kind of the issue that we ran into this time which is there were a set of poesy rows where owners were not identified early on and and they just kind of languished out there number one and number two there’s the reality of when our code freezes and and and those are the only two realities represented here if the community decides that we want a proposal gate i am all for discussing it and and seeing if it makes sense because it’s think you avoid games are released chicken by providing clarity and clear exit criteria as you move from one phase to the next yep so we absolutely do have that it’s not represented here you please go to the crew benetti’s feature repo and click on issue template in that you have a series of gates where needed to alpha beta stable I’ll posted in the what’d he call it and each one has a set of check marks next to it that are required so it’s not it’s not date driven but it is kind of stability or I guess completeness driven and and we’ve heard from the community generally that dates or even t-minus dates are aren’t fantastic you know more give clarity around the thing that said if something isn’t you know clearly passing alpha by call it you know that the code freeze day it’s not going to make the release clearly I agree we’re trying to be couple the lifecycle of features from the life cycle of the crew brunette ease project but we need to be very clear about what the touch points are when we sink those things back up that’s all that I’m asking for yeah fair enough and then I mean if they’re features which have to be kind of merged past this deadline right we need to all kind of go with it so there has to be elders community p.m. whatever the group has to do the discussion and approve it at least informally and there should be fully understanding that it means there is a scope creep and that Mary jeopardize release dates so and as I said I did old box race to the feature should contain official able to track it back and then the other observation wasn’t planning right out of lust 123 planning there were a lot of cigs which came out only with p0 type features I don’t think it’s fully realistic I think they should be kind of right there should be some ratio where there is some amount of p0 some amount of p1 sp2 sp3

is coming out of this planning right I am kind of open to discussion what the ratio should be one-third p0 is 50% p0 is just my generation we cannot have six coming out and saying you’re going to tempe zeroes that’s not gonna end well and then so the reality in 1230 a bunch of features with deprioritized in the middle of the release cycle 2 p-10 features and so why not don’t we accept the reality from get go to slightly better prioritization up front and split it issues this way and then kind of reiterates a little bit right so owners of the features need to align the resources by week four we don’t want the situation where their features is no nurse no resources aligned for that I mean to be to be clear you know if you have the people and you have the proposal in and the community feels like this is a good thing and you can accomplish it within the milestone it’s kind of set to go it’s not like you know you’re gonna put in something that we feel is that that you know person a says well I I don’t care about this feature at all a person be sis I care about it deeply and I’m ready to allocate you know three devs to work on it and you know based on the proposal we have here it’s going to be accomplished then and the proposal is approved then it’s in it doesn’t you know it doesn’t matter that the person they said that you know this is a Pete you know this is a p100 ya know when agreement it just kind of it’s nice to have this discussion advance it’s that instead of marking somethin as whatever is a priority see and then turns out it’s again it’s all just too it’s the clarity and then turns out that nobody align actually any people to work on it so I kind of that may go back to the features repository to have at least some proposed list of github IDs who is responsible for this feature so that if people ask questions in all from to reach out and then I kind of another proposal is to introduce and I think this’ll a kind of exists that owners and maintainer should abide by it so ang features they manage right and if they manage way too many features that maybe this feature should be dropped in advance right it just doesn’t scale and it just leads to a lot of the archer nan’s the end and it’s a way is basically right people should be able to answer within 12 working days so and whatever mean comments done mate so basically just a call to manage a little bit more aggressively by everyone what goes into the release any objections questions also the link is posted in the dark I would appreciate all the comments and all the work and with the whole room there is ngj to make sure it’s all kind of captured thanks so much Mike this is awesome and i actually just tweeted how exciting it is to have the revised workflow coming from the from the outside community saying let’s refine this better and let’s hold each other accountable so that’s super awesome alright we’ve got a couple of minutes left were there any other cigs that wanted to do a very fast report out we can catch up with cigs next week then that sounds like a plan for running rapidly out of time so let me cover notices again for the people who weren’t here at the very beginning we tried the Asia friendly timing for two months on one one meeting a month and we had more people for whom it was super inconvenient than people appearing that where it was more convenience at that time so until we get a larger call for something that is friendly for asia and parts other parts of Asia specifically India was more inconvenience for example by the other time we will leave this at 10am every week just rise and shine and join us for ten a.m. pacific time there’s a point about documentation TJ mentioned that documentation was not part of feature complete this time 1.3 that means now is the time to get your documentation in we had a lot of trouble around 1.2 getting documentation in after the release so we want to get as much of it in before the release so please prioritize please ask for help if you need help please please get us documentation in for the 1.3 release and

then my last notice is the Cooper Nettie’s first birthday is july twenty first and we want to do some fun celebrations around the world and i need help with that as does bob produce key to make that happen in more locations than just where we are so please send us emails if you are interested in helping plan celebrations maybe it’s a special meet up we can help with you know I don’t know cake and cookies and pie or something anyway we will reach out please if you have time and are interested in helping make a goober Nettie’s first birthday party in your region and that’s it for notices does anybody else have any other notices to be added beta is coming the beta release is coming birthday dad one question I was chatting in the background some folks is it’s not always clear what saved aligns with necessarily a particular feature someone in the community might do okay um like in my head I was just thinking through things that we have worked in the past and how to think of the cigs I participate in and I think not always a clear overlap so do we intend to associate past features with existing cigs I think that’s an open question for past features i think it seems like for example like someone you know jobs when i’m a big guy right someone goes into scheduled jobs right what sig is that associated with and like in my head i could say oh no machinery but I participate on that suddenly there so I guess very much little puzzle is tied for cigs and I feel like a lot of the features we work on this in this way associated with a clear cigarette right now and maybe I’m wrong in that assumption now I think that’s true I think that’s a reasonable question and to your point it’s about associating existing features with cigs as opposed to the new ones going forward which the feature hosel template should handle um I think it would be a great exercise and certainly is something that we can work on as we roll forward and touch and update those features we can try to align them with a special interest group is that the right choice yeah I don’t know if it actually is the right choice which makes me question the original proposal somebody so sig biased versus maybe more future owner bias or the existing owners in the code base bias so that’s just the thing I was trying to rest like through my head you think there’s enough dude do you or the rest of us think there’s enough meat here to have a longer conversation next about that I would are you game to lead one yeah I mean I think it’d be a little bit like the set of features that we’ve worked on us Adam just to ask where they align but if you’re asking next week in the community meeting i guess i don’t know if i will have the time to prepare for that but it’s just an open question and they we can talk about it on communities down okay you do that and see if it bubbles up to be something that should be a conversation proposal kind of thing here i guess i would add one quick thing on this which is basically i find that these taxonomic organizations just never fit perfectly and but that the organization itself is worth it so I wouldn’t worry too much i guess about finding the perfect fit for any particular feature the point is to find oh yeah and to find some owner and and move forward with that and so like what David said earlier about how like this isn’t supposed to be totally perfect it really applies I don’t know maybe cigs is added fit that we wanted to try something else but I don’t want to spend too much time and I was analyzing this and tried to define the perfect economy dignified yeah that’s what I was trying to suggest Brendan so like and that’s why I was trying to correlate it back to like owners that are like the owners file that’s already in the current repo is probably like the ad hoc signet will form around something if you’re touching it like like whoever’s like if you I’ll just pick an example like if you want to do a feature in jobs like whoever is the current listing of owners and like a job package that seems like the ad hoc cig you should work together on this and maybe we should update the workflow proposal to kind of reflect what we anticipate would be more ad hoc that’s all I don’t know I feel like if it’s too ad hoc it’s gonna be hard to get it to run and so starting it in the cig and then having them defer to people who are

in those owners files may be the right way to just to have a touch point for a place to start from right yeah like the CLI I don’t think is there a clear safe for that maybe I’m just not attending it or others you ax right isn’t there a cig doing you know fish like I think that’s what covers the CLI is the and that’s kind of broad I guess you x is kind of broad but so we are at eleven o’clock it seems like there’s enough that this discussion should continue maybe on the email list or in the slack channel and brian has also been sending notes in here about the owner where the owners are documented and asad that he will lead either has created or will create github teams for future owners so there is more discussion it sounds like so apologies it does eleven o’clock i am losing my room and i will see you all next week thanks at all happy thursday happy thursday you can read