Red Hat OpenShift OKD Working Group, 8 Jun 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: OKD Working Group 2021 06 08 Full Meeting Recording

Description

OKD Working Group 2021 06 08 Full Meeting Recording

A

All right, let's see how many people we got seven people here. Let's wait like two more minutes uh before we get started christian is here, looks like there's some network uh issues um yeah. Let's wait like two more minutes before we get started and if you notice, I have a hackmd uh link in the discussion and so that's um an attempt to create sort of a linear document of an agenda and that I think, will um be helpful for folks to have that on hand.

B

Could could we add in um attendees to that as well? If you want absolutely all right, so that would be good. That's really! All I have been tracking in in actual notes for each of each event.

C

D

Sorry I haven't been, I I couldn't make it the last a couple of uh times but um and now I finally got my okay t-shirt. So.

E

And I haven't gotten one, yet it still hasn't.

D

F

E

It got to germany and it hasn't come to canada. Damn I'm.

A

Gonna arrive today actually.

C

Yeah mine got here earlier this week. I also got mine fits perfectly.

D

Yeah, it's awesome.

B

And we will get more of these, so if someone's coming on and they didn't get one- um don't worry, we'll we'll put them in the uh you know, work a linkedin store that you can um a linkable store where you can self-service and get them.

B

So there you go someday we'll have a meeting where everybody has an okd t-shirt on.

A

All right well looks like we've got enough folks here to at least uh uh get uh rolling here so again, there's a link in the uh channel uh posted again for anyone knew that joined. um Please put your um attendance uh at the very top of that uh and note uh who uh or what you represent, um and you don't have to represent anyone. You might represent yourself.

A

So, uh let's get started with some quick introductions, uh I'm jamie mcgarrett and a member of the working group for a little over a year now, a year and a half and um using okd in university of michigan various environments in the university of michigan, um we'll be spinning up probably later today, a cluster that's going to be used for a couple dozen developers doing data heavy applications in the political and social sciences and uh anyone else want to introduce themselves diane. Maybe we'll go next to diane.

B

I'll go next: um diane, the director of community development over at red hat in the cloud platform bu, and um one of the instigators and chairs of this working group, along with everybody. So.

D

I haven't been here in a bit, um I'm christian glennbeck and I am an engineer um in the openshift or in within red hat, and I've also been one of the co-chairs of the openly.

A

You want to introduce yourself.

G

Hey so I'm timothy and I work at um at red hat on the federal court as well chorus team which, among other things, work a lot of does a lot of the work for fedorak rs.

A

Bruce you want to introduce yourself.

H

Sure you know I was just trying to sign or put my name in the document unsuccessfully, even though I logged in, but I'm bruce link and I'm just a okd user. I teach at bcit and managed to put several students through the okd experience successfully doing projects and things like that. So that's been a lot of fun.

C

Joseph yeah, I'm joseph meyer from germany, I'm working for rhodium schwarz as a cloud architect and I'm an okd user and supporter since uh yeah more than three years now.

A

And the man who needs no introduction vadim.

I

I'm letting blagowski and my webcam is apparently broken. um I work redhead focusing on upgrades and release activities for okay.

A

Excellent neil, you want to introduce yourself.

J

Yeah, hey, uh my name is neil gampa. I I work as uh at datto as a senior devops engineer, focusing on software delivery systems, including source code package management, container stuff, and things like that, and you know I'm here in the okiti working group, working as both a person in fedora and to a lesser extent, a person from datto to do awesome things with uh okd4.

A

Excellent uh anyone else want to chime in that we didn't get yet.

A

Tree, you want to say hi.

K

Sure, I I say, hi yup hi everyone, my name's sri ramanujan. I work with neil at datto as a senior software engineer, uh working on various products and services that are, you know, built on top of our in-house uh kubernetes and openshift clusters. So, and I'm also here as an interested hobbyist, especially in like the home lab space, I'm looking to expand okd's presence there and sort of bring up its cache in that community.

K

A

And tree is a master of disaster recovery. I'm led to believe.

A

Did you get it back up yet or no? No.

K

I the cert thing that uh that that other person mentioned- I don't know his name right now. It didn't work for me, so I just blew it away earlier this morning I re-spin it probably afterwards.

J

Did you actually? Oh, you haven't responded yet I was. I was wondering if you'd actually brought it back online after everything fell apart. No, not yet after work, probably.

A

All right all right! Well, let's um jump into the agenda items I do like to have the intros, because there's always going to be people listening or watching that aren't familiar with who we are and what we're about and again um okd is the uh communities um kubernetes community supported kubernetes technology. That is supported by a very interesting and lively group of individuals from a diverse with diverse backgrounds and it's a real fun project to be involved with.

A

So, let's get into release updates with vadim.

I

um Sure so we didn't release for a release candidate this week, because the code freeze is planned to be happening this week or next week once it happens, we'll release a new release candidate that it shouldn't have, can changes just a bunch of bug fixes. So if you could kill the nokia 48 a try, that would be very appreciated.

I

Meanwhile, on 47 we released a new stable release it. The main highlights of it is that machine os moves to fedora 34, and we might have an upgrade issue number 668 um affecting for seven to um to the latest table, upgrade I'm still looking into. That seems like we missed, including one package and uh the most concerning party today was not caught by ci.

I

um We'll do um our best to fix it quickly, see if the workaround is feasible. If it's not we'll move to a new way of building machine learning, content.

I

For a release patchway soon, the issue is that also we're seeing quite a lot of table clusters upgraded to the latest version, something about 16 telemetry.

I

So we'll need to figure out more information about this issue. So if you could give an upgrade a try on your test environment and reply to this bug, that would be very appreciated.

I

um I think that's all we have from the technical standpoint.

A

Okay, great- and um we always like to bring our our um friends um from fedora core os on to give an update on that. So timothy timothy, you want to give us an update.

G

Sure so, right now we have two items on the agenda. The first one is about systemd, which is a new part, a new demon, I think from systemd that has been introduced in fedoras 34 if I'm correct, if I'm not mistaken, and that is not yet enabled in federal courts by default and we're looking at whether it makes sense for container workloads to have that part of the system.

G

uh So, to give a short of view, I think uh the assistant d, umd essentially is is a demon that is using the new c groups, v2 and psi interfaces from the command to figure out which process it should kill. In the case where the kernel of the system lacks memory is under pressure, memory pressure and so essentially on okd well, this might have at least consequences.

G

Maybe we don't well, we don't actually know if what what would be the consequences of using their that in okd uh so yeah? The idea is to bring that out to your attention and trying to figure out if we should enable that we should remember that by default. Of course, it could be enabled in federal queries, but disabled. You know kd2 so, but well, if, if it doesn't generally make sense for container workloads- and we just probably won't enable it at all and uh well, this is like the right time to gather feedback.

G

If you have any on this issue,.

I

um That feature seems to be very useful, but for now it won't affect us because it requires the group to we don't have, and it's not clear if it would be defaulting for nine. Even um another problem is that it's probably undesirable in kubernetes world, because the hublet cannot read system d events to find out. If your pod has point that reads kernel events, if the system o and d would kill a process, it would look as if some random restart that would just confuse all things um again.

I

We would need to work upstream in in kubernetes to to reconcile these two approaches until we can enable this by default.

I

um So what we can do now is uh to to stop, depending on on fedora cross decision on this really and just disable it in our custom payload. uh But that's probably the input from the okd on this all right.

G

Well, that's that's great I'll pass that forward to the record streams.

B

G

B

Right in the chat, james is asking um teach kas to talk to omd, or vice versa.

D

So there is a when you, google, this uh cubelet oomd, uh there's a talk coming up by a redhead employee that is mostly about sea groups, e2 and kubernetes.

D

um So maybe this is giuseppe livano and um maybe we should just be getting his opinion on this because he probably knows a lot more than we do about this whole.

A

Complex did someone want to reach out and um and gather some information from that person, or at least maybe invite them to um come here and talk with us.

D

um I'll reach out and ask whether he wants to join next next time, otherwise I'll I'll. Just ask him to give me a few words about sure awesome. Thank.

A

You all right timothy, what's next.

G

Yes, the second item is a little bit more like forward thinking thing, uh because it's about february 35., so we've just rebased fedora 34.. So it's not coming right away, but it's coming in approximately six months and it's the right time. I would say to take a look at what's going to change five and see how it might impact okay into the right. Third.

G

And so far we haven't identified cursing, I would say: oh, that would be strong.

B

G

One second.

B

I'm just gonna mute other people and unmute yourself. um You need to.

A

All right, so click typewriter, okay, go ahead. Timothy.

G

Yeah, all right so fedora certified census. Mostly nothing stands out so far, but we're tracking that and the federal change process window is not closed. Yet, if I'm correct so so yeah. This is uh this is ongoing and that's mostly it for me for federal course. There are any questions.

G

C

Did I understand that uh right that fedora, of course, 34, is coming in half a year but is used now in okd.

G

uh No, no we've yeah, so, as vadim said, uh fedora korres based on fedoras 34 is coming right now in okd, just right now, federal 34 has been released a couple of weeks: uh okay, okay, thank you and, um and so yeah we've. We finally finished all the rebates for federal race and uh right now we're starting to look at fedora 35, which is coming in approximately six months.

C

G

Thank you to make sure that we catch up things before they happen before we that we can provide feedbacks or everything is in fixed.

I

Excellent right from the looks of the changes, it doesn't look, anything would break us, but I think there was a good practice which we applied for for age release. We our machine, os content, has been using fedora's next devel stream to fetch the latest changes which would only land and stable in a couple of weeks, so we can catch them and the installer or 48 is also currently using.

I

Testing requires rather recommends you to use test thing ice. The initial fedora cross images so that we would change check that both of those are perfectly valid once they land in stable. I think we'll we'll do the same for rokiti for 9 or 410 until it becomes stable so that we could verify the changes ahead.

I

That has been pretty valuable, as we call quite a few regressions we're able to report them before it lands in four seven or in for the occurrence stable.

A

A

All right, let's move on now to uh issues. Let's uh take a look at our uh issues and one of those is.

A

That I want to highlight actually before we get into sort of the general overview of issues. Is the operator wish list. So this came up in conversation um in an email thread?

A

um Do you want to summarize the the email that that was sent to the group and sort of um explain your response to that, and, and maybe also christian's response to that.

I

uh About the operator uh wish list or yeah the email.

A

That, um basically, asking for for clarity and for a documents, change um in regards to what was actually available and what was supported and what not. And there was actually a okd document change because of it.

I

Right, that's because, for instance, we had to rip out some parts of the documentation, because those logging operator is not directly available in okd and while we are waiting for folks to create a new kitty um uh catalog for the operators to start pushing the upstream versions that we would have to remove some parts of the documentation.

I

um I don't think we have any estimate on this yet, but we'll be it's because effectively out of our hands. What we!

I

What we need is the collaboration between various teams to commit to supporting those in no giddy, as in supporting by community novel as not redhead support, but nevertheless it's still some level of commitment required and once it happens, we would be able to bring those parts of the documentation back.

I

But until then we effectively have to remove them and tell folks that they unfortunately have to build those operators manually or look for alternatives and community catalog. We have and that's basically uh that's the current.

D

D

Like a few probably months by now back, um we talked about having an okd specific catalog. uh Just for okd. Has there been my question is: has there been any uh development on on that side.

I

I have not heard any it's mostly uh the talks between the teams. They want to ensure that we don't plan additional work on them, but still this catalog gives them benefit of well community support.

I

It appears that technical part of this should be trivial. It's just a matter of uh ensuring that the team is actually committed to supporting those uh versions of the operators before they land in official red, head operators or to put, or uh they want to develop an entirely different community version. It's just a matter of how they would structure their work. uh Creating a catalog itself is fairly trivial.

I

It's just a matter of uh making teams uh commit to that.

D

I I wonder whether we should be setting up or just helping the people that would do that set up this catalog, so we have something the respective teams could then release to without worrying about uh you know this kind of like it doesn't even exist yet um and also, I think well well. This is probably more internal. The management has now given them more clear. I think mandate to the teams to release to the community catalog or release a community version of their uh operator.

D

um So maybe, if we kind of make it as easy as possible for them um to actually release the operator that will help with that process.

A

One thing that I was thinking about is when we talked about this at the time, and I put a link in the chat to the particular issue that christian created we've got the wish list, but we don't actually have an explanation of what needs to happen.

A

um We've got them categorized, currently unavailable available in community etc. Is it possible to maybe put a little explanatory text in that issue um as a comment, or we create a separate page? That's like hey! This is actually what needs to happen for these various components to come together. So people can understand what needs to happen and then also we can look at timelines and also for the people who, in this group or who are interested in okd development for them to get started on this.

A

Maybe some tips so to provide them some some guidance as to hey. You know, if someone's like. I would really like to see this operator in okd. Well, it needs to be rebuilt.

A

Here's some general place to look about how to do that, or something like like each each repository for each represent for each operator is going to have specifics, but it might be nice if we inspired people who are interested in okd to to take this on by giving them a little lift.

I

Yeah, well probably, we should regroup them into two major groups as an additional order of attacks like maintained by red hat and uh something like others. The problem is that why we can submit logging operator, for instance, into the community operator, because is that ocp users would see them duplicated, they would see official redheads logging operator and one coming from community and they would be very confused by which one is actually being supported.

I

I think that that was the main reason.

D

Engine separator, I have the okd catalog right.

I

Yeah exactly yeah, but there is an engine's operator which is totally fine to submit to a stream operators and community can start working on this right now and it's already listed in the wish list. So what we need to do is to review the list.

I

Find a list of things folks can start submitting right now like engines, and I think nvidia operator, or maybe everyone uh and after careful review, we can start. uh We can find the documentation on how to submit who wants to take care of it. How to hear the responsibility should we include these kind of new wishlist items into these meetings or who would maintain them and so on and so forth, um the other internal red head operators.

I

They would need a bit more work because there is a whole team behind each of those and we need uh their support and they would be maintaining it without. Okay use work, group kind of uh requirements and such.

A

Okay, well, let's put this on the agenda for the next meeting and in between now and then, if folks could think about how we could organize this, um and should it be a sub group, you know like a subcommittee or something like that. Could it be the main group and we'll talk about in the main meeting, but how we can move forward on this, because it's been lingering, like christian, I think published this in in january, like so, we have a list that we can start working on.

A

We can, we could be adding stuff to it as well. um If we start moving on it, I think it makes um okd more attractive to have, because some of the ones in there are pretty significant right and they're. It's things that people are really looking for so.

D

A

I think one thing that.

D

Go ahead: christian, oh yeah, sorry! uh So one thing I think that the community can definitely already get started on is um looking at the currently unavailable operators and all of those which aren't in the openshift organization or even the openshift k native organization.

D

So all non-red hat operators, um those won't be released by by us, anyways, so and and those would be going into the upstream catalog, I guess- um or if they're, if they're really made just for openshift, then they would go into the community catalog, but any of those all those operators that aren't owned by red hat that are currently are unavailable. um Those would definitely be good candidates for the community to look at them and add them to the respective catalog on operator hub.

D

There's also that there have been some changes um recently to the operator hub, so people can um more yeah release new versions uh by themselves without uh having an admin from the operator hub repository um have to acknowledge those changes so that there's gonna be, I think, owner files for each operator. So when you submit an operator, you'll become the owner of it, and then you can also update it without anybody else um kind of having to acknowledge that from a third party.

D

So maybe, if there is interest within the community to get one of those operators released right now, um all the non-reddit operators will definitely be good candidates for you to look at for us internally, all the operators that haven't been released. Yet we obviously want to get them released to okd or okay by the respective teams um yeah, and we can also pick some folks.

A

Excellent, all right so we'll put that on the agenda for for um our meeting in two weeks uh and talk about this more and have fleshed it out a little bit more and probably put in a separate page, as opposed to this issue, actually maybe put an okd page either on the website, or something like that. That's that we can actually edit in line um as opposed to just adding comments to etc, that that we can improve upon and update the lists and and whatnot all right moving on.

A

Are there um there's a um a new issue that was submitted um vadim? Do you want to talk about this? A little bit deploy your installer provision clusters on bare metal looks like it was submitted an hour ago.

B

Is this the um documentation one yeah yeah, so I I was going to talk to that just quickly.

B

um We just spent uh the prior hour to this meeting with bruce and mustapha mustafa from elderberry from red hat and bruce from bcit, and this is uh I'm just bringing it up here to give sort of an example of um of how we're working through um the docs.okd.io to find um defects in the documentation and report them back, and so we've we've done our first pair pro pair, not programming, repair review and we're trying to document the process, and there is a recording that I'll, we'll I'll put up as soon as I figure out what the youtube issues.

B

But this example, and I'm just going to share my screen- I'm just going to grab one of them here and I'm just going to share the troubleshooting. One is the process. If you find a defect or an issue in the documentation that we're using right now and we'll we, I will document it um in a a brief blog post on the okd website um and as well. So basically just finding the section and we're reviewing him with michael burke who's.

B

I don't think on this call, but in the documentation team for openshift, um and so this isn't about actually fixing the issue.

B

So in some examples there are code examples that need to be reviewed and retested, but this is a good way of make sure you put the bracket okd in the subject line, put the link to the section that the issue you're finding in and then potential um resolutions, and if you don't have privileges like most of us, don't to assign this to somebody um that slash a sign at the bottom of my screen there for michael burke and myself um will allow us to track it, and at the moment I don't have privileges to add a label to it.

B

So I'm going to work on that, I'm going to double check with michael. He may have privileges, but there are two labels that we're using one is okd dash only and the other one is um that it's documentation, so there's another label in there for that. So once we figure out, I don't think you can assign like you, can do the assign a person to um that.

B

I don't think, there's a slash label function in github um to do it without privileges, so this is we're gonna start working through as many of the issues, I think bruce. We found like there's 20 references to our cause in the okd documentation, and so we're going to use that to flush out all the issues and then move towards doing that. So what the ask is for this group is, if you find something especially right now, we're focusing on the docs.okay dot io.

B

If you could use this approach um to tagging them and keep an eye out for them, and we will do this and try and work with the docs team to make sure that, with each release, there's a process in place for these things, these patches to be applied correctly.

B

I'll stop sharing. If anyone has any questions, but that's that's what that issue is so you'll see a bunch of them pop up on the issues list um and right now they aren't properly labeled and that's that's my issue to figure out this week.

A

How many issues all right.

D

And is that the the okd repository, because I seem to be able to uh edit labels, you have more, you have more yeah yeah. I don't know why that don't straight, though right.

B

Yeah, I I don't, um I can add stuff, but I can't edit that so I'll figure it out and maybe in the interim, if you can take those ones and just add okd only label and the documentation label to all of them.

B

That would be helpful, but mustapha is going to keep motoring through and working with bruce and if other people want to get paired up with someone um mustafa, if you recall, is one of the newbies to the the red hat um openshift team and what we're trying to do is pair people, the newbies up with external folks who have used and deployed okd, so that there's some um life experience a lived experience going through this and then once we figure out this process and get a few of them through there.

B

um What I want to do is set up a hack, the docs you you don't have privileges either see it's not just me uh christian doesn't can't add them.

B

So what I'd like to do is sometime this summer, when everybody's feeling fine is to do a hack, the docs day and hopefully we'll catch everything in the docs.okd.io and then move to like the guides that that everybody's created and the actual readme files and take on each of those using the hop-in platform.

B

Because that's what I want to do on my summer. Vacation.

A

Excellent all right, I put a little note there um mentioning that uh diane is going to organize the hack the docs day, um so we can look forward to that. um The next thing I want to move to is vadim. If you could give a quick couple sentence, explanation of the um good first issues label and what that means and and what your hope is.

I

uh Right so we've been looking for easy bugs for community to get started for quite a long time and when we wrote uh a log bundle issue, I've noticed that.

I

Things are missing there, for instance, we're using additional uh services to update from fedora cross to ocd specific um look at this specific uh payload.

I

Normally, those are not a problem, but when things break, we would like to get logs from them as well. So a few issues to fix that have been tagged, and these are mostly touching the installer and our fork of it. So it would be very useful for community to get started with building their own installer.

I

uh That would effectively spin up a new guide how to build your own ocd installer until it merges in upstream, and we would add some more services to collect using uh log bundle and.

I

Whoever picks up this item to verify, meaning break their cluster, collect like bundle and ensure that all the logs are in place. So this kind of a small things would help us write better guides and explain how things are interacting with each other.

I

uh Hopefully this would affect the troubleshooting, the bootstrap and log bundle analysis, documentation and so on sports.

I

um I think that would be a great um starting point for folks to understand the complexity of how we built the system and how we analyze the problems, meaning block, bundles and batteries are essential for us to collect all the information. We need to find out all the details about the bug.

A

This might be something that the group could take on as a regular section of our meeting or a subgroup could meet to basically like go through these and sort of bootstrap each other into you know, because a lot of these involve little bits of knowledge that maybe one person has a little bit here. One person has a little bit there, um and maybe someone needs a jump start just to to be able to dig a little deeper and help out.

A

So it might be worth us thinking about how we could, as a group, sort of approach these um and share a little bit of information to get people um bootstrapped into being able to help, uh as as sort of group developers or group support people or whatever, and this sort of runs into that concept of we talked about at the last meeting of of having basically people who have who commit themselves to being supporters.

A

Basically, like supporting um uh you, know, questions and whatnot, and that that specifically make themselves available um or or say that they will specifically be active right to do that.

A

I

Thoughts about yeah, perhaps.

I

Perhaps we could suggest um we could have an open, floor-ish uh style of suggest your own favorite paper cut or an issue so that we could file it. I could help with if it touches some um ocp part that could help with filing a proper bugzilla ticket and driving it to the team, so that we could see how it gets fixed in master of release uh for eight then gets here. He picked 247 and explain how uh how soon do we get it and how to how to interact with other teams.

I

Probably that should help us fix a bunch of things which are left behind, but very, very easy to fix. Actually.

B

Okay, I would also suggest that what we're doing with the docs the pair programming or pair reviewing process might be a good way to flush out the workflow around these.

B

um What we need- um and they may all be different because they may be- you know something maybe a fedora coreos, but I think if we do, if we have those tagged now, we could use the same team of group of folks that mustafa is coming from the newbie red hatters to pair them with external resources who are willing to help mentor them to do some of these first first paper cuts, um maybe motor through it. That way to get the process flushed out and I'd be happy to do facilitate that.

B

So do we have a bunch of first cuts listed well first, I I'm calling them paper cuts, but because you refer to them that, but as a label.

A

Yeah, there's a label there's like five of them right now, yeah.

B

Four or five of them, so maybe what the thing is is to reach back out to forget who the team lead is for mustafa and see if we can get a couple of those five looked at by that team, one-on-one and pair them with whomever is appropriate externally and then see who's missing. Because that's what that's where, like michael burke, came in with the docs, he understood the workflow for docs wow.

A

Some of these are pretty like, like so, for example, I'm looking at the one. That's that's ownership of um the um authorized key like that's something that we should be able to look at pretty quickly. You know and just figure out, well, okay. How is this? How could this happen? But you know I don't know um all right. So, let's put this on, we've got we're 18 minutes left. I do want to make sure we have time um to get to all the other things without rushing the other folks that are reporting in.

A

So let's put this on the agenda for the next meeting and sort of flesh out ideas um and sort of be prepared to to move forward on that at the next meeting. um In the meantime, if folks can chip in on those um again, these are these are issues that are tagged um in the okd repository um with a good first issue label.

A

All right um is there anything else in terms of issues um that anyone wanted to bring up, vadim or christian, or anyone else that you want to bring our attention to in the in the submitted issues.

J

A

Okay and what about discussion items so we talked before about subgroups um and do do we want to start documenting the proposed we've got these ideas, for you know people that would be um supporter like dedicated to doing support and whatnot. Should we create a spreadsheet of some kind and start having people basically submit their names and what it is that they'd specifically like to help with?

A

Does that sound like a good way to start um just as people log in each time uh put who you're affiliated with and then maybe underneath that um put what your interests are in contributing in the community or is that too much? Do you think for people to to put into into the document thoughts.

A

A

Wow, okay, I guess no one has any particular opinion about it. So let's do that at the next meeting, we'll ask folks to put not just who they're affiliated with, but also what part of okd they might be interested in. So maybe at the top of that document we can have a list of um of ways of like some labels.

A

That would be, you know, ways to participate, and then people can just sort of list those under their name in ways in which they might be interested in participating, we'll give it a shot, we'll see how it goes. um It just seems like a good way to start forming groups and knowing who wants to do what and keep track of that right.

A

uh We already talked about system omd, uh any other discussion items that folks want to bring to our attention anything else that would submit it. That was submitted in the discussion section of the repo um that you think we should pay attention to. In particular,.

D

uh Actually, jamie, I just have a comment on kind of the the whole uh sub group um plan, which I think is absolutely great. I I do think that that's gonna help focus people on a specific task and get get that task done more quickly.

D

So, in our charter, which we uh wrote at the very beginning of this working group, we already kind of have the idea of sub uh working groups, um and there is a little section in the chart about that um just to link, I linked it here in the chat um so yeah I'll put it in the dark as well um so yeah. Just uh I just wanted to voice uh my support for the idea in general and there already is kind of a process.

D

Although we haven't flashed it out, I think um of how to create one of these groups or get it started. um I I definitely appreciate this thanks for bringing it up.

A

Absolutely and so I'll reference back to that and then build on it um and sort of weave that into the next meeting, starting with the next meeting excellent and if folks want to chip in ideas and and hit me up between now and then have discussion about it. Please do this is a group effort by all means um any other discussion items uh that folks want to bring to our attention anything that was submitted in the discussion section. It doesn't look like there was that many, a handful.

C

No okay, just a short question, say inside feature that shows up in okd's overview on the web console. Is this uh yeah? Does this feature do something- or this is just a remainder of uh ocp feature- that's turned on okay,.

I

It's an inside separator which, if you have uh redheads full secret, would collect data from your cluster periodically and save it in red. Hat's uh servers just a small sub of data, which is very useful for us to identify problems early again. If you don't have a valid full secret, there is nothing to upload. Rather, you cannot authenticate with the red hat server.

I

If, if there has been some data uploaded, it can also give you your recommendation. There's our 8gb, which is too small, and things like that.

I

It kind of does not apply to okiti, but that gives us a valuable stats on how people are using the okd and that's the only stuff we have meaning we are very tempted to use it to make valuable decisions. For instance, there is a bunch of four five clusters um and barely any for eight clusters live.

I

That means we need to push for eight uh advertisements on every single meeting like that, so that people would give it a try. uh Is it representative of what people without the pull secret are using? I don't know we don't have any dates for that. um That's that's! Basically what we've got.

K

Is that thanks providing the telemetry you talk about when we do upgrades, or is that something else.

I

It complements telemetry because telemetry is actively sending things rapidly, but insights makes a snapshot every couple of hours, two or maybe four hours and telemetry is able to send metrics, while insights is actually making a snapshot of uh control, plane objects, anonymizes them and sends us here. So we get some information based on metrics. For instance, we can kind of find out your operator state based on metrics, but it's a bit tricky and if we want some more details, we look into insights archives because those are actual snapshots of the project.

K

Okay and I'll stop, but it doesn't work if you don't have a full secret.

I

Yeah again, everything flops if you're right.

D

Yeah, the the insides operator is essentially just the client side of this. Obviously, um then send all that data. It collects to um the backend that red hat has, um but I'm not sure, I'm not even sure, there's a an open source repository for that back-end. um So I'm not sure how how you would use this, uh if not for the for the only use case it has sending the data back to red hat um and obviously that only works.

D

If you have the uh the right pull secret on the official red hat full secret, I mean.

L

D

Not sure I mean it's definitely helpful for our sre team. I think um that's a big data source for them, uh but yeah I'm I I was I've always been wondering whether we could kind of uh make that warning of that note notification, that you could connect the or install the insights operator and set more data that we kind of don't have that show up on okd.

D

um But on the other hand you could also use a red headphone secret and install it on okd cluster and then send the data to red hat. Although probably nobody has done that so far.

H

Yeah christian, I think in vadim. It actually would be helpful. I think if somewhere it was written down the precise boundary of where you could use the red hat, pull secret and not because I've heard many things informally and certainly like you know, anybody that has a developer red hat developer account, which is anybody, uh can get one that's valid for 60 days and then.

M

H

M

H

To be valid after that, and there's some maybe mysterious way where you can say that you're self-supporting or not anyway, the uh like that.

I

That touches the topic, which I don't really know a lot about, which is subscriptions, so all of that is outside of our um kind of reach. There is a service in in front of all thread heads um servers which we maintain internally called dual booth and what it does is it effectively validates the pull secret and tells us hope this data proceed further or it needs to be thrown off.

H

I

That might be actually.

H

I

I

Well, there was a nice talk about subscriptions recently on openshift, and it's slightly different from engineering standpoint like uh um are not set in style. They can, if you're, a educational for instance facility. You can ask for a subscription which enables you, with self-supported stuff.

H

uh Good luck with that.

M

A

I actually actually put in a request a year ago hold on let's, let's we have seven more minutes and we do want to get report from the docs folks. So, let's we will investigate, I think, subscriptions and that boundary for pull secrets we'll see what we can find out. Let's add that to the list of to do's okay, so we have joseph on the call- and I don't see mike, but I know joseph and diane both have things related to docs and outward communications uh to bring to our attention so take it away. Folks.

B

Joseph you want to go first, and I don't know what you have so go for it.

C

I have to admit that I was not um was not in the docs meeting the last time, so no news from my side.

B

Okay, no not to worry- and I think I pontificated a little bit earlier in the issues section about um what we're doing around docs and um and trying to tag things and create um a flow around the docs.okd.io.

B

um The other thing and- and I will create use joseph your blog um on the site to create a little documentation around that and awareness. For that. um The other thing uh I think I added it in the lower section. Let me just popping between things um we did update, I'm not sure anyone noticed in slack um the community support text at the very top of the openshift dev and openshift users to point people to the okd working group for community support.

B

So that's key- and I mentioned in the chat on the side on june 25th post red hat summit, gathering we're going to take a number of the end user stories that peop that were pretty popular, including joseph's, um talk on okd to azure and restream them live on openshift tv with sort of a mystery science theater approach. So we may pause them and put in new um references and things like that and updates.

B

And so, if you're, if you're around and you want to join us, um the link is embedded into the notes here and we'll be streaming. I think it's about four hours long, but there were some awesome talks.

B

Interest identification of members um is the other thing here, um not quite sure what you meant by that jamie, but um I'm the okd working group's doc sub group meets every other tuesday, oh at the same time as this one. So if you are interested in joining that group um same time same place, one week later, sorry.

A

That was connected to what we were talking about earlier with people basically um denoting their interests.

B

A

The documentation.

B

Yeah, okay, cool, thank you um yeah! So and that's that's that and and if anyone has a talk that um they want to give, um I have a podium via openshift, commons and openshift tv to do that. um So I'm looking for other okd related talks over the summer that um if someone has one I would love to um hint john thornton um to to to do a you know, interview style or whatever style. You like uh get your story out there and get more people um coming to the okd working group.

F

I will try. I'm gonna have a very busy summer. We.

L

Thought we'll try to.

F

Try to convert some of that work into something I may be able to present.

B

Yeah no worries um yeah, and I there was one other thing that I was trying to think of um yeah, and I know this summer is going to be busy for everybody, and people are going to start taking vacations, but um you know coming up with uh a date for the hack the docks once we're there. I can see jamie's got his hands full, um but uh yeah with a good kind of hands.

B

Full, yes I'll, put something out with maybe a doodle or something on the working group list to see if we can come up with a date for that, and now I can see from my next meeting a few people popping in because this is my channel and maybe what we'll do is we'll get a separate channel for the okd working group jamie at some point, so that they don't overlap.

B

Yes, and that is definitely the okd's working group's youngest member ever and um he has not yet gotten a panda thing, so we're gonna have to figure out how to get panda diapers. I think our panda t-shirt. So uh this is leo. Why.

A

Yeah all right any last minute we have three more minutes left any last minute items that folks want to talk about before we sign off.

B

And shri I just wanted to mention. I saw your post in the working group about the messaging around okd and well. We'll have I'll have a conversation with you about that and see how we can move that whole conversation forward. So look for me in slack sometime when I'm I'm hanging out and you think, I'm I don't look busy and you don't look busy you're on european time or are you on the west coast? No.

L

B

Eastern eastern time, okay, I'll look for you and we can. We can move that forward. Birthing.

A

Mohammed, uh we can, if you go to the um slack channel and post your question there, lots of people can pop on there's lots of people doing um home, lab setups, um and we do actually have a couple of documents that we can point you to um that are on the website joseph. If you have that link handy, maybe throw down the um the home lab link um uh that we have up on the documentation um but yeah. If you go on the slack channel and ask your question, um we can definitely help you out.

K

Yeah I mean this is just for like value adds for okd, for you know, eventually we'll put a section of the readme try to flush that out, make it more clear what you get.

K

I know it is tiny bit unclear, but yeah I can definitely drop a line in there. I don't remember if I did before, but I'll put another one in now that it's more.

B

It it also might be a fodder for a good um talk as well, so think about that, like you know the way that, like yourself and neil or um and myself um talking through why what the what the value proposition is for you guys. I think that might be a good monday open shift. I do um upstream stuff, so I could do an open shift okd. You know what the value prop is for okd from dado's point of view um and have you.

J

What event would we be doing this in, or is this a dedicated event or something like I? I guess I missed a little bit because I'm juggling like three meetings at the moment.

B

That's okay! I I host. um I have on openshifttv a channel called openshift commons in which, on mondays, I do um upstream talks and okd qualifies as an upstream project so or a sibling stream or whatever. But oh.

A

B

A

Yeah sure can't explain it because there's there's this next meeting coming up and there's already people joining yeah. So let's, let's move it to the channel and um say thanks to diane and christian vadim and timothy and everyone else for contributing and bruce say.

B