Red Hat OpenShift OKD Working Group, 12 Nov 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: OKD Working Group 2019 11 12 Full Meeting

Description

OKD Working Group 2019 11 12 Full Meeting
co-chairs: Diane Mueller, Daniel Comea, Christian Glombeck

A

So for those of you who don't recognize this page in front of you right now, this is we're using the community projects page to drive the agenda.

A

If you have other agenda topics that you want to talk about or would like addressed, then let's add them in here and we will try and either find a speaker or give you the podium at an upcoming meeting or even today is if we have time so with that I'm going to just kick it off with the first proposed agenda item, which is basically my recap of the upcoming events.

A

The first thing that we have that I've just announced we found a room during koukin and we're going to hold okd office hours from 11 a.m. to 1 p.m. Pacific on the November I should have said 19th year, just edit that so, if you are coming to coop con, you will be Christian and myself and anyone else who's interested will be hanging out.

A

There for two hours, drinking copious amounts of coffee and.

A

Then the other thing that we have is the I have a number of speaking slots for upcoming, open ship Commons gatherings, and if anybody from the okd working group would like to do, the lightening talk, we'll have a small slide deck from the gathering. That's next Monday in San Diego at koukin and I just like to each time. We do an open ship gathering over the coming months to keep promoting okd, getting more people to test it and give feedback on it and contributing to it.

A

So if you are able to come, January 29th to London, to San Francisco on April 26 or to the bukhan at Amsterdam on March, 30th, I'm hosting gatherings there and going to put that in if none of you show I will do it and then the other thing is or I wanted to see. If we could schedule sort of a face-to-face, okay, Dee Fedora, core OS working group meeting at Def, Con cz on January 24th to 26 and I've, asked for a room for that and Christian and I and I.

A

Think we put Danny's for me his name on it as well. I've submitted a paper to see if we can get some get room and then brother sized about okay, Dee and the road map and everything there as well. So that's those are the other upcoming events, if you guys know of other opportunities to promote, recruit and get people to work with us. Let me know and I'll try and find speakers or get there myself or figure out how to get Christian wings so that he can fly around and Danny any questions about that.

A

Many volunteers, I'm.

B

Pretty sure Christian can have wings anywhere. He wants I.

A

Think he comes with wings, so that would be I, think so um yeah. So that's just my spiel for the beginning to kick us off I'm trying to get the word out there about APD, for we have an open task to write an article for the fedora magazine. Maybe once we get through today's update, we can transcribe that and some of the instructions on how to build and turn that into the fedora magazine post that we proposed writing at the last meeting.

A

So without further ado, if there's no questions about those talks, just email me or put your name associate your name with one of those updates in the agenda and I will keep asking at each meeting so Christian you want to take over take over the screen now and maybe do your okd update and that's person and.

C

Yes sure I can do that. Okay, let's do that.

D

Okay, so I'll, actually just which the machine for this one sec, yeah.

A

I'm waiting with a lot of patience for this because I heard a fourteen roadblocks yesterday, so I'm hopeful that they got cleaned up.

D

Hey you should be able to see my screen now. Yes,.

A

We do you might want to make it a tiny bit bigger if you can make the font a tiny bit bigger.

C

Does it work for you that works nicely yeah yeah.

B

I see a dashboard that doesn't say openshift container platform. Oh.

D

Yeah, you can see the okay dear ending and we are actually if we, this is a spa. A cluster buddy has created for us. So thanks buddy, this great work by the way um he's been doing lots of fixes and moving us forward. So without him we wouldn't be here right now, which is yeah.

D

Thank you about him for that I wanted to say that first, let's have a look at the notes that are in the cluster right now you see it's three masters three workers and let's look at this master, though the OS image we were booting from is a Fedora core OS preview image, which also means that we're that we're running spec, 3 ignitions back three yeah, and this is really needing the first really working faster.

D

We've got so a.

E

Great progress X will have.

A

D

Will have the proper alpha release, I hope, bye, bye next Monday's openshift gathering offensive comments gathering so, though, that everybody will be able to download it and test it out. Yeah.

D

There's not much more to show maybe I'll.

D

Let's have a look at the operators. Real, quick, oh.

B

Maybe good is operator hub automatically configured with ok because it used to not be, and there wasn't really any instructions on how to set that up. So.

D

I'm not really sure what configured users right now, what we wanna ship in the in the release is open. The the operator have with the community catalog I think it's called all the community operators, which I think this.

D

Okay, so there might be more operators in here right now, then we'll eventually get will probably just get those to the community thing in okay be for the beginning. That's.

F

Because this contains an actual pull secret, so we get all the things as if it was a standard OCP, but eventually we would be able to deploy it without a pool secret and some operators, which is be gone.

D

Exactly so yeah and I hope by next week, we'll have that all set up so that you can just download it without a pull secret and install, and you know, tinker with it.

B

That all of the installation methods that are currently in OCP should work with ok d.

D

So we'll definitely exclude Ezzor for the beginning, because we sort of had to rip that out of the Installer for now to get these factory support, but yeah in the coming months will will aim for for parity there and moving forward. We want to support new platforms in okb first, so.

B

We will have at least OpenStack and Overton things like that all supported.

D

Yeah, if that's the part in the Installer right now, we should be able to support it in okay,.

B

A

Testing comes into play here now getting other members of the working groups that try and test on other platforms. The very.

B

I mean yeah, so those two are the main ones that I'm interested in trying this out with, because that's what I have so.

E

Hey Christian, you standing here the the boy secret when he's gonna come how you guys sort out the the what's the plan over there Whitaker's to the the Installer which still validates on the procedure to not be a to be an empty array. So how do you guys what's the plan over there? Can you share a bit.

D

I'm not actually super sure do we need a pull secret if the images are just available without any login. If.

E

You're using the Installer, which I know and Vadim can can jump in and correct me fear is the Installer which I know it does validate on the poi secret to not be an empty array.

E

Okay, then we'll probably change that in the F.

D

Cross branch off of the Installer until we find them yeah until we sort of aligned that and merge everything from the F cross branch back in will we're not quite sure yet when that can happen, maybe in the.

C

D

Time frame, maybe we'll have to wait until the 4.5 time frame really depends on management and the architects to decide when OCP is going to move to spec 3. Those are ongoing discussions and yeah and.

E

The images and the orchid images which spit out from the CI they end up in Quay, is it or you guys, found a different home, no I think it'll be quite all right. Why cheers.

A

Couple questions in.

C

A

Phillip is just asking: when do we think we can have a working release of okay, d4, and you know Diane, has happy years and thinks that we could have something out. At least that's the next topic or one of the next topics is, you know.

F

A

On how to build your own, but the secret thing sounds like it still needs to be resolved. Is that going to take more than next Monday to resolve the secrets requirement or.

D

No I think we'll have to fix that before to be able to actually have it openly out in the open. Yes, I'll put that on the on the okd project board on github right now, yeah.

A

Okay, so do that, but then it shouldn't be.

D

A huge task we probably just have to remove the validation, then, if there is a.

A

B

Or we could have a dummy value.

D

Right then, that might be an easy work groans well, yeah.

B

Because, like I expect that so my thinking is that if people want to transition from okd to OCP, then being able to put in a real full secret and switch things around from the Installer or whatever would be useful and also having the dummy value means that the validation part doesn't have to be special case or anything like that.

E

I'm not sure whether that is gonna fly straight off the bat Anil, especially when the or the the solo part of the post-acute is one thing. But then you also have the on the pile of a site where you know the operator does check to see whether someone you know whether is a very secret to kind of pull all this stuff.

E

So the mechanics over there as I said I, mean I'm, not sure but I'm sure the idea, the volume and the encryption with my wife shared some rights in the coming day so or week or so well, what you said I think is not for the year they won so to speak. I wouldn't I, wouldn't put my my by my unit. Well.

B

I mean we have to do something to have it working by Monday.

E

Yeah maybe work around a hack, Co, quick hack, but what you said to the transition from okay, dito, CP and just have just sort the year. The secret I think that's a medium term, so to speak.

B

I'm just saying that, with the way, if we have a dummy secret value that we don't have to change too much logic about how it's being validated, because the value can still be used to pull- and it's just okay, it's fine and then it just keeps coming down anyway. So it's an open secret or something like that.

B

Whatever like it's, it basically allows you to not have to do special casing internally to manage the differences between okay, Dee and Ossie P, because well, my experience is that the more special casing we have to do between the two, the less likely okay, these going to stay working.

A

That's a good point: yeah.

D

And until we we really merge the f cost branches of MCO and installer back in will sort of have to do things a little hacky and that's I yeah. So in the alpha version we'll get next week, they'll definitely be some workarounds. We don't want to do in the long term, like maybe with a poetry, secret right now. I'll have a look into this this week to figure out what what the easiest path is to get something working til next week, but that might yeah the the solution might change in the future. I, don't yeah!

D

I! Don't really have have an answer to this right now, but we'll we'll see as we go ahead. Okay,.

A

So I'm just gonna if any, because someone mentioned coop con. If any of you are coming next week for a poop con, we could. We could walk through some of this as well in the office hours on Tuesday, the 19th and so check the back to the agenda and there's a link to the room. And if you send me, your email I'll invite you officially to that.

A

So you get the directions in your inbox and if any of you are doing, talks or anything like that at KU con and can give us a pitch to get more people's on this. That would be wonderful, too, and, of course, we're gonna have a short talk about the gathering.

A

By Christian on the 18th, and if you'd like to come to that, just let me know and I'll get you in so I think probably the the question that I have next is before we go and get bedding to talk a little bit about instructions on how to build your own Christian. Can you talk a little bit to how you want to gather feedback on this? So if someone internal and or someone external to red hat want to give you logon issue get in a forum? Where do you want to get the information.

D

um So right now we really have multiple options for or receiving feedback, which would be the Google Group, the mailing list and the slack Channel I'm, not sure we have a proper place to really file an issue on github. Yet we might like yeah I'm, not entirely sure at what the best way would be or if we maybe even create an bugzilla component for that for people to be able to file bugs on Bugzilla.

D

B

Debug category in in Red, Hat, Bugzilla, I, don't know if anybody knows it's there or using it, but it's already there. Okay.

D

That's great yeah, I didn't know it was already there, so that might definitely be a place. Somebody.

B

Should go see if it's organized well yeah.

D

Yeah I think once we have the first release out, I'll definitely try to write some instructions on how to we're on how to give feedback and where to file stuff. I haven't really given this too much thought until now, but yeah, that's definitely thing I'll figure out yeah and then I'll outright to the email to the mail, miss what what exactly we came up with I.

F

Would prefer all the issues for for this to stay in open ship community repository? Because we don't know we our end-to-end test, not yet passing and half of our teams are not even aware that okay D on the door on caress is a thing. So if you would file a bug for console, for instance, your bug may be rejected immediately because it doesn't, it has a difference, release image, so we don't want to scare different people. I would rather prefer filing issues in open ship, community and github.

F

We would triage them and send them to the respective teams. I.

B

Would really rather know those were really rather not, but.

E

If you, if you look in the in the past, Anil on the ok, DV or 3x, the kind of you know you had this component and not many people were internally world were chasing this this stuff. So what what is missing is just a compromise for now did until we get all the deductions. Oh no I am.

B

Aware, I, just I, don't want code issues in the community repo there's a perfectly good origin, repo that has an issue tracker as well. If we want to use github or.

E

Regional origin, origin, repo, okay, go on so I I stopped. You I got you yeah.

B

So the origin, repo already has an issue tracker, and that makes more sense for where code issues are gonna go because at some point they all will be OCP issues anyway, so putting them all in the in the right place. If we're talking about github is where the code is, the community repo is more about process and procedure at dealing with topics and meetings and whatnot. So I would rather see us if we're using github issues, let's put them with the code repo, which is openshift origin, yeah,.

E

But in for you have, apart from the have tons of other gazillions repose, which which has a very low form for EXO, maybe what you're saying is fine, don't don't use the Bugzilla abuse may in don't use community, because it's what it is, maybe just maybe create a one. Other good rapport like chorus has right, but.

G

They are finding issues. That's I think what Neil is trying to say is well, let's keep our community like discussions separate from our bugs yeah. Let's, let's keep the oke community. Are the open ship community repo, at least for discussions, because I feel like bugs, are probably going to drown out the discussion at some point and so having either a they'll.

B

Do that almost immediately right, yeah.

G

So like having either a new repo that we create that specifically for that or reusing origin is probably best if we create a new repo. There's really, no reason why we can't move issues around since github has that ability. So, for example, if there's a issue that gets opened in the wrong place or there's a more appropriate location for it.

G

You know if you are a significant owner of the two repos that you want to move, you know from one to another, then those issues can be moved so that is that a good answer yeah, but.

E

But don't forget the in terms of being the the oh, no such like those that that can be done. If you you part of the the open shift org. In other words, if you had employee for the community side, I know Clayton done some some magic over there to kind of allow, for example, non-cotton Plato to be able to to have more privilege, so I would say rather than them. You know migrating issues from left right center. Maybe just come up with just one single point and that's it and that's children. I was saying if.

B

We knew that it's not that much more different than just reusing the okt Bugzilla project, because that already there we just need to set for we just need to add a four X component. Call it good.

A

So what is there an just? This is me being trying to keep it as simple as possible. Is it possible that we just create a label? Okay, D for problem.

B

Is that indexing and searching through them is going to be pretty perilous, as it is I think is within the open. Did the origin things like having labels for bugs, but the problem is the code? Doesn't all live in the origin, repo anymore, which is I, think what vadym was saying or some Danny.

E

B

G

Yeah, and the other thing too, is if they all live in the same repo and you just use labels. It's you know, I'm.

B

Like nobody who, just.

G

Cares about direction and and taking part in you know, discussions and I. Don't necessarily want every boat report to show up in my inbox and it's hard to do that if they're all in this thing, repo yeah.

B

And github doesn't really provide you a good way to subscribe to subsets of issues so.

C

B

That's that's more of a feature that Bugzilla has then github does so that again, that makes things a lot more complicated, so I mean what we could just start with the fact that there is a or there's an okay de product in Bugzilla just add a Forex version, and we start from there. If we decide to do something different, it's easy enough to just transition later.

A

I, just okay: how open is that boobzilla? I mean, I know that I have to log in to get to bugzilla and all that people.

B

Can view it without logging in and you can search it without logging in you just have to log in to file a bug report, which is.

D

Yes, so that's something from.

B

Github and you.

D

Can use your Fedora ID, which right everybody can create for themselves to log in and.

B

You don't need a fast ID. You could just create a red head, Bugzilla account if you don't want to attach to any SSO. So it's probably the most open of the systems. We have right.

G

The only the only thing I have against Bugzilla and you know Red Hat employee.

G

So it's not like I, don't have a bugzilla account, but I have to admit there have been times where I have not opened bugs, because the bugzilla interface is just not quite as good as you know, going to where the code or the discussion is already happening and opening a bug there so like I, obviously will use Bugzilla but I, wonder how many people would use not to file a bug just because I just feel like github is probably you will probably get more.

G

They are more participation and Bugzilla is a little bit hard to navigate. Sometimes I.

D

Agree dusty here because.

G

D

User experience- it's not really great github is much better I can't.

B

Refute that either it's just the only the only thing that I that makes github a little bit more troublesome is like map.

B

Well, I guess it doesn't really matter, but like one of the nicer things about using bugzilla is that if there's cross-functional things that need to be fixed like there's a cost issue along with a with an upstream Fedora issue and a no kitty issue that goes into a downstream those CPSU, it's super easy to map those relationships within get within Bugzilla github makes this extraordinarily painful to the point that it's not really a good.

B

It doesn't really work out well, but if that's not really a concern that we care about, then then that's fine I, don't have a problem with either option. It's just I, don't want I, don't want it to be so I don't want things like open show community be cluttered with with reports. I got.

A

You can I, ask Christian, um can you can you share your screen where the engineering tasks live? Were you tracking the engineering tasks yeah.

D

Sure so the engineering tasks live in the ok default project in the github organization on github. So it's like a Kanban board or Co efficient way to share my screen.

A

D

Then you see my screen: I can.

A

D

This is the the github.com slash orcs, slash, OpenShift, slash projects, lash.

D

D

D

So this is what we use right now to track our work, and it's really across repos sort of a boards. We can pull in any any bugs from any repos in the organization yeah and the long run I mean some some bugs may make sense on on the repo themselves, but I think it also makes sense to be able to try ash them love all the community. Cogs are going to be accepted by the respective teams mmm yeah.

D

This is what we do right now, but we'll need a solution that is a little bit more yeah, more open in the future, for people to just report bugs I think it's either gonna be bugzilla or a new github repo. At this point, yeah oh yeah I'm, not sure who will decide that in the end, but we'll we'll try to set this up before the alpha release. Next week, next Monday yeah.

B

um In terms of the project planning stuff, like this Kanban thing, the Fedora taiga instance could be an option for like having a more open setup for this I. Don't know how well that would work out. If for what you want, but at least it's it's an option and that one I think also lets you log in with not only fast but maybe you're getting up a count. So.

D

I'm not sure if you can create an account directly on the Tiger instance, but it would be yet another system and for people to yeah. So.

G

Yeah, probably a good rule of thumb for now is to you know, keep it simple and then once we need advanced features, yeah.

A

G

Other things yeah.

A

So, but and I think that was dusty speaking keep it simple. Is that used Bugzilla as is or keep it simple? Is that.

G

Just saying this project board that Christian has you know, will use that until we bump up to the limits of it, and then you know it's no longer serving our needs well enough and we will explore something else. Right, I, don't see any reason.

B

That that's fine.

G

A

B

Do that out, as a suggestion.

A

How would um so, if we're using this project board here, how would someone log an issue here? Can you just.

D

You, you can't just add an issue here: I need to pull in an issue from any other repo or I need to just add a card on one of one of those I'm, not sure who has the the ability to do that the permissions.

D

So this is probably a triaging thing, so what may be I think the most feasible. It may probably be just create a new repo to track engineering tasks that we then triage by adding them on on this project board. We could do the same from Bugzilla to this project board, but that, right now, that's still a manual process, adding it here. I, don't think people from outside the organization can add anything on that board. Right now, yeah.

A

Can I just I'm gonna take the screen back hang on a second and make a suggestion here. Just for now I think you can see my screen. Can we create another project under community that is okay, d for bugs and issues or something like that, and that would be.

D

So I'm not really sure these projects and because you can't really create issues in a project you can create cards or you can link existing issues from other repos yeah.

A

So they can lock. This is I, know yeah I hear you're. You can learn about mixing community stuff, but if people added an issue, an issue here with a label bug or whatever you guys want to propose, and then we had a separate project board for okd, you know alpha release bugs and we could triage the bugs here on the community page, so I would I would.

D

A

D

Our whiteboard for this, because you can pull in issues from any repo within the arc. If you do, if you do it in a project within a repo like a community thing, you're.

B

D

B

C

B

Gonna point out that I was just gonna point out that if we're gonna use projects, we should always do them org-wide, rather than a repo specific, because repo specific ones have exactly the problems that christian pointed out. Okay,.

A

So I'm there thanks for explaining so yeah, so lets christian and I and and make will use danny as a guinea pig, because your external will work on this this week and get that set up and Danny or whomever else is external on the working group make you all test it make sure you have access to it because I'd really like and then we could again report a video on next week when Christian and I are face to face kook on during the office hours.

A

If nobody comes but I think people will come and document it and put it out there in a blog post or transcribe it into a readme file or something so now that we've beaten that dead horse, which is now a live horse, oh they'll, be soon vetti. Are you still with us.

A

That, even you did is he.

F

Yep, let me see all the screen screen.

A

And I asked if he would give some instructions walk through how to build your own bouquet d4, and this will get updated a few times between now and then probably, but this would be a good way to get started. So take it away that night, okay,.

F

Hope, you're gonna read the text. My editor, oh.

B

A

F

A

F

The difference between OCP, four and OCP three and the same, throw kitty is that everything install is now and compassed in one release image. It's also called release. Payload release anything. So it's an actual container image in a special format and you can decode it using command OC atom release info uh most likely. You would have to use a pull secret.

F

You know to pull it from quake, for instance, so the parameter the Shea you can put the pass to the post secret and it would be displayed so shows all the information about it, its name, the nightly it has been wrenched from digest, and the most interesting part is a list of images it contains. So it's literally just a container which has the references to different images, all of them most of them are operators and we pull one by one set them up and the whole cluster gets running.

F

The digests are not really useful if you're planning to work on advocates just change to open, you can show using parameter committee, or else you can show the URL for each particular comment. This image has been built from so the way forward. If you have followed the blog instructions on disconnected install, you would mirror all of those release images using commands release a mirror from its initial part to whatever the station. You prefer and create a new release image in that destination, and these are the official instructions because later on, you would announce it before.

F

You would lose your word and see if you change the release image, but you know gd4 world. We can do that, though. You can use the command to release new the creates a new, updated image and we're replacing one single part of that.

F

For instance, we were pointing the cluster network operator to a different image, with whatever fix you have prepared. The way to build each operator strongly depends on the way on the way the separator is being built. So that's very specific: you would have to ask maintain us of that package and after that command you would get an info how to update your installed config so that you would be able to use your mirror image. The once you're done. You can export an environment, variable release, image, overwrite and point it to your new release image.

F

So that's the usual workflow. You would change something in the relates payload, creating new one and run the cluster with it overriding the default for this image. So a few more advanced called topics.

F

We are using the very same release image containing the very same four three operators, as if they are the CP release, but we are replacing two parts: the open, shipped installer, we're building it from branch called f.cuz and machine config demon needs the support for ignition versus three. We also have the Rockaways branch. So if you're planning to create new pull requests there or file images, please add something: we should notice that it comes to the apricots branch instead of the master.

F

Now topic is we're using a standard, Fedora chorus images which are released by the records team, but all of these needs additional.

F

Binaries, in our case that's hypercube and cryo, so we have created a simple repo which shows how to how to add them to your payload. Previously, if you have noticed there is a special operator, we just called machine us content. That's not in fact an actual operator. It doesn't smoke a set of people at ease, yawns, it's a no streak on it, packed in shape of the container.

F

So here we have a docker file, which does the very same a very similar thing to what we do in our CI when we're building a machine us content for the request. So here we use using the karez assembler as a base. We would need it later on to add additional binaries to our payload. We are using a fork of machine configurator because we also need a binary from it and in the entry point.

F

We first of all fetch the latest dream from Fedora chorus. What we need from it in this mess of. We need a path to the OS tree, combat tar later on. We fetch it and extract it to the repo. Then we fetch a few rpms using yum downloader.

F

Here are the list of packages we need so the coral damage, cryo I, open shield, hypercube and clients and a few packages for cluster of us, so that our end-to-end test would pass- and here is the list of proposed we're using the ones the yum downloader has patched all the RPMs. We extract them and move a few folders to accommodate to the RPMs three layout of the files. For instance, let's see is stored in user as being as a symlink to user.

F

As being so, we have to copy all the binaries there and remove it, and once we are done, we would call command caress, assembler def overlay and build a new choice to commit tag it with the same ref we use previously, and that would be our updated machine, OS content. We need all of those sorts.

G

Of these things that you know are kind of hacky stuff for now, it's not always gonna be like this right. These.

F

Are high key stuff because we need unprivileged, we need unprivileged builds on our CI in a proper way. You would build your own work or further across using, of course assembler. For instance,.

G

Yeah, this is just like an intermediate stuff to get us something now right. Yes,.

F

We have to depend on on whatever cleric OS is releasing. We cannot afford ourselves a new stream and deal with other things. That's too much too much work right.

G

F

G

How to do future is to get a line better alignment there, so that the differences aren't as great right.

F

Yes, on the other hand, it's just a matter of getting privileged builder in our CI, so we'll see which way will go. Okay, thank you.

F

That's pretty much all I've got I will drop a link to the notes and all the repos in the document.

A

Would it be appropriate to put that note in some form in the community Rico somewhere or some more public, or is that just too too hacky at the moment? No.

F

The that would work we just need the real full specs for the images once we release them and push them to way.

A

The more we can keep in the repo, the better. In my humble opinion, um can you go back to a minute ago and I'm gonna yeah, just asking about the two, the two alternatives: there's a Fedora core, OS, instabuilder installer and then the other thing. Was you explain that a little bit more, the CI piece yeah so.

F

On CI, we fetch the exact tourist recon it the door across is using, and then we update it using grow assembler dev scripts, because that one doesn't require privileges will literally run it. You know and then opens you've built.

F

Eventually we would move to a more privileged build and we would update rpm to be using chorus. Assemblers other features to actually creating new commits that would require virtualization, because that image has to be built in, but on the other hand, I would update rpm TB, which we've come to now. So these are the options we have, but the first option is our start point we're using it now, but we're definitely going to kick it away at some point.

A

In the open, shipped ansible script, playbook, rather it's it's a transition, so I'm three to four, so I'm still catching up there. Thank you any other questions. There seems to be a couple in chat.

D

Maybe let me just add to that, and what we're doing right now is the binary is most of the binaries. We overlay that we extract from the RPMs in in that script are built in a Santos environment on CI, so we sort of ditch in sent us build binaries into a Fedora image which isn't doesn't have to be a problem. Necessarily I would really like to try in the longer run, to build all the boundaries in in a fedora environment as well.

D

So we can be sure, there's no, no problems due to the drift between build system and and the system that the thing is actually running on. There are some blockers for that, though, like we have a few patches on the goal and compiler in in Santos and rel that we don't chip in fedora and yeah. We're actually going to discuss this this week here in Boston internally to see what the way forward with this will be, but right now the binaries we put on the F course machine. Os content are build in a Santos environment.

D

Just to add that context to this.

A

You might have just blown my mind, but that's okay,.

E

Hey buddy might have a quick one for you, the the CBO, based on your quick test, you done with it. Does that work nicely with you. You know um injecting you're a new version or based on faith course and such like. Oh that's, not sorted yet yeah.

F

Cbo doesn't really care which exact operators are you inserting there it would justify them. It's a.

C

F

You know, as we had to purge machine containment, to properly apply order to commit this ignition in version 3.

E

A

Though, um let's see does that answer everybody's questions from the chat, Philip L D, why it's not on the playlist, it's it's up there in the YouTube he's asking about the recording from the last meeting. Add it to the playlist. I must have forgotten to.

A

B

When do you think what was just presented will will be posted out, so we can start tinkering with it.

A

Well, I'm hoping Monday, okay. Next, probably that means Christian are you? Are you in Boston the whole week on East Coast time, Christian yeah.

D

Exactly I'll fly over to San Diego on so today, yeah, though.

A

Yeah, that's my goal is to have it. You know whether it's Vette Eames raw notes up there on there and a demo done at the gathering on the 18th of November by Christian and be able to do you send an email to the okd working group and to the Commons and another work mailing list to ask people to take a look at it. Give us their feedback. Okay,.

A

So that's if you look in the notes, if you have a platform like James, was talking about bare metal or whatever that you would like to take on testing this on mom, please, the notes is in the notes, are the notes link is here again. Let me see if I can grab it, it's in the chat at the very top and yeah. Basically, now what we're on the road looking for is people? Oh.

A

That notes, maybe you're, not meeting James the notes from.

E

He was the notes from Fatima Fatima already said, he's gonna after they either called he's. Gonna provide some some bits and pieces. Whether that's story in in Guiteau in this Google Doc as a link or whatever we'll figure out.

B

The mailing list- yeah that'd, be great yeah I've got it I've got a small stack of Intel, looks for them for the bare metal. Does it require ipmi, or would it work on on something like the the little Intel looks.

F

Work, it should work on Intel Noakes using the UPI flow with IP I yeah, you need, you need a PMI or something else. The metal cube wouldn't stand in. My way definitely will work. Okay,.

B

I could use virtual BMC.

A

B

It doesn't matter because the IP I for bare metal doesn't work anyway, so.

A

All right going back to the working group meetings here, I know: I I took you off that dusty off about the packet stuff, still not ready for that and we're almost at the end of the hour, so I think that's all we have for today I'm going to move this back over here. Unless there's another topic somebody wants to address and please anyone who is coming to coop con.

A

Let me know my email address is D Mueller at Red, Hat, calm and I'll get you that you do not need to be registered for ku Khan to come to the okay D working group session, because it's in a room outside of the Convention Center at the Marriott The Gaslight Marriott, on the eighth floor on November 19th. So it's we didn't want to exclude people if they didn't have passes I'm the coop um I.

A

Don't think we may boot up a blue gene sessions. If people are interested in that and have it just hosted virtually as well, that's something people are interested in doing next week, certainly send the no doubt about that. That.

B

Might be interesting, it just depends on what time that is. It.

A

Is going to be Pacific time, 11:00 a.m. to 1:00 p.m. next on the 19th.

B

A

Set up a virtual invite and put it on the calendar and send it to the the working group and we'll just see how it rolls. Okay,.

B

Sounds good to me yeah, absolutely, because that I have actually got a free block of time doing there. That was an I'm, pretty sure yep.

A

The 19th Tuesday I, don't know what that is at Eastern. I can't add any more, but so.

B

If it's 11:00 a.m. then that means it's 1:00 p.m. Eastern yeah or no is three hours right, so 2 p.m. damn it I can't math can.

A

I not this morning.

B

A

Let's do that and we can just use it as to make sure that our documentation is right for the update process.

C

A

Well and just use those two hours wisely because we have Christian in for a few more hours and then he has to run to the airport because he's not staying for pucon. But if you know anybody from any of your respective companies or organizations that are coming to coop con, let them know and I'll send out a note again to the okay D working group. So that's what we have for today and.

A

Going going gone, everybody gets what whole whacking seven minutes back of your hour. Thank you very much of a team for all of your efforts and Christian amazing and we've gotten this far and I really can't wait to show it off next week. So well done everybody thanks.

D

Thank you all for joining in.

B

Thank you thanks.

A

And I will edit and get the playlist updated too so grab all.