← Kubernetes Podcast from Google
Kubernetes Podcast from Google
Kubernetes v1.32: Penelope, with Frederico Muñoz
00:44:14
Release Lead walks us through the 1.32 release: new feature highlights, deprecations and removals, and the release theme: Penelope. Do you have something cool to share? Some questions? Let us know: - web: - mail: - twitter: News of...
Loading summary
Transcript104 lines
- [00:00]
Kaslan Fields
Hello and welcome to the Kumarese Podcast from Google.
- [00:03]
Abdel Sigiwa
I'm your host Kaslan Fields and I am Abdel Sigiwa.
- [00:17]
Kaslan Fields
Welcome to the 1.32 release episode. In this episode we speak to the release lead, Frederico Munoz. We talked about the theme, what makes this release unique and why you might be interested in upgrading your clusters to this new version. But first let's get to the news.
- [00:36]
Abdel Sigiwa
Kubernetes Special Interests Group Network announced the general availability of the Gateway API version 1.2. The new version brings a number of new features to the standard release channel like HTTP route timeouts, infrastructure labels and backend protocols. Some new experimental features has been introduced like name rules, retries and percentage based mirroring. There are also two breaking changes that you will want to be careful of. Check the link in the show notes for more details.
- [01:05]
Kaslan Fields
AWS announced the general availability of EKS Auto Mode Elastic Kubernetes Services. Auto Mode is a fully managed Kubernetes offering where AWS manages the cluster lifecycle including selecting the right compute nodes, auto scaling and cost management.
- [01:21]
Abdel Sigiwa
AWS Re Invent took place last week on December 2 to 6 in Las Vegas. AWS announced many new features across infrastructure, machine learning and of course AI. Check the link in the show notes for a detailed list of top announcements.
- [01:35]
Kaslan Fields
And that's the news. Hello everyone and welcome to our 1.32 release episode. I'm excited today to be speaking with the release lead, Frederico Munoz. Thank you for being on the show. Frederico, would you like to introduce yourself?
- [01:50]
Frederico Munoz
Thank you for having me, Kazeline, of course. So I'm Frederico Munoz. Fred is perfectly fine and as you mentioned, I'm the release lead for Kubernetes 1.32. I mean, I've been on the release team since 1.25 and I've played and done different roles in the release team up until my nomination as release leads for this one. And I'm quite happy to have been involved and looking forward to the rest of the interview of course.
- [02:24]
Kaslan Fields
Excellent. And one reason I'm so excited to talk to you today is because we actually work with each other a lot in the Comms team in open source. So Release team is not the only thing you do in open source Kubernetes.
- [02:37]
Frederico Munoz
No, no, not at all. And it's very interesting this interview because it's a bit of a full circle for me because actually I don't know if you know this, but the very first contribution that I had in the Kubernetes project was with you because it was when I ended up helping out with some tweets that we needed for Kubecon Nice America 2021 or 2022, something like that. That was the very first contribution that I did. So obviously being with my continuous involvement in the Kubernetes project, obviously we worked together in the sick contributor experience comms, where I also do some work on the blogging and obviously making the overall community and the world aware of the great work that our contributors do. And I started there in the Kubernetes projects and it's still a substantial part of my involvement. And to a large extent, I think that very first step was instrumental in me giving the other steps and then joining the release team as a Release Shadow in 2025, et cetera.
- [03:40]
Kaslan Fields
It's kind of how it works in open source, right? You find one place that you kind of start getting involved in and then it just kind of builds as you find more activities that are interesting to you. So I'm so excited to hear that you got to do the Kubecon tweets. Everyone out there call for help. If you ever want to help out with the community's Kubecon coverage, we always welcome new contributors to help out with that. We have a little program where we define all of the social media posts that we need to make about what the contributor community is doing at Kubecon and we welcome anyone to contribute to that. So thank you, Frederico, for doing that. And I also want to call out that Frederico has also received a contributor award for his work as a blogging lead in the Comm Sub project.
- [04:23]
Frederico Munoz
Yeah, that was actually very well. I obviously loved it. And it was one of the ways that I think the Kubernetes community and the community work that we do end up being, in my experience, slightly different from the work that I started doing in free software and open source when I began my involvement in it in the late 90s, is that I think that there is a bigger understanding of the importance of community recognition and community involvement now than back then, which obviously is an improvement. So I really cherish that. Awards in a lot of ways, by the same measure that I take the nomination as release lead, it's mostly because it involves a recognition of the community and of the people that I work. I mean, in my own hours, in the night shift, let's call it like that, and like me, many others. So you end up building this sort of extended family that, I mean, the months go on and you start to depend on the work that others do and you start to know them. And obviously when you get this sort of recognition, for me it means that you're trusted by your peers, and that in itself is one of the best feelings that one can have. So absolutely, that contributor award was one of those examples. Just like the release lead nomination, building.
- [05:50]
Kaslan Fields
Open source software takes a village and one of the most exciting outcomes perhaps of all of that work is the releases. So you do a lot of work with talking about the work that contributors do in the blog and in comms throughout the year. But time you are the Release lead for 1.32. So can you tell us a little bit about what's going on with the 1.32 release?
- [06:14]
Frederico Munoz
In a way, I think that the best way to describe the 1.32 release is that it's been so far reasonably uneventful in terms of the way that it goes, which is exactly like we like it. The last release cycle in the year is always a bit more complex than the others because it's a bit shorter. And also it generally overlaps with three very important and relevant and impactful periods. One, it's Kubecon Art America, the other one it's Thanksgiving in the United States and also the Christmas season. So this makes the timeline arrangement a bit more of a challenge. And in this release we actually made some changes which up until this point have proved successful. For example, we've merged the code freeze dates and the test freeze dates, which previously the test freeze was a bit a couple of weeks after the code freeze. But the community consensus, and specifically of Sigrelease and all the contributors was that this was actually not yielding the right results in terms of the test coverage. So we ended up just merging them together. And this is one of the changes that this release has. I feel that each release, and I've been on this release since 1.25, as I mentioned, as I went to several different teams, comms enhancements, release notes, both as a shadow and as a septim leads each release cycle. I think that the actual maturity model and the improvements that we have makes the next release a bit easier for the people participating in it, and much more so to the release lead. I think that the way that for example the SIG release sub project works specifically with Kat and Grace helping the release lead and the rest maintains a certain continuity. And also at each release we document and improve the things that weren't working as well because of the retros that we do, which means that overall we are having a very peaceful release. And by the time that this interview Airs we will have the Kubernetes132 out. Also, I think that each release has a different flavor to it because although processes are important, each individual and each team is a composite of the release lead, obviously, but also of the sub team leads and of the shadows. And I have only good things to say about the team that I have. So I'm extremely lucky in being able to depend. We go back to the discussion that we had before. It's all a matter of trusting and depending people that you've. Some of them, you know for quite a few times, the others are just joining in, but all of them end up lending their time and passion and involvement in a way that makes a release not only a technical achievement, but also an achievement of joint community work, which I find very, very uplifting.
- [09:26]
Kaslan Fields
And rewarding.
- [09:27]
Frederico Munoz
And rewarding, yeah, indeed.
- [09:28]
Kaslan Fields
Yeah, for sure. And speaking of releases having their own flavor, I'm going to go ahead and ask what is the Release theme for 1.32 and what was your inspiration for it?
- [09:40]
Frederico Munoz
I mean, this has both a short and a long answer, but as I never want to use one word when two will do, so I think I'll give you both. So the theme of this release is Penelope. And it's Penelope because of the Odyssey, right? The ancient Greek myth with Odysseus or Ulysses in the Greek and Latin name. And this is, I mean, for one thing, a theme doesn't have to mean anything, right? This is like a small little thing that it's left up to the decision of the release lead just to add a certain flavor, right? And it's completely detached from the actual version numbers that end up on the cloud providers, et cetera. So it's not very much used outside of this. But I still think it's important because it gives great T shirts and good swag. So. And in a way, it's a reflection of the community work that we do. So why Penelope? I think I need to go back for my own path in life to give a better answer. So I actually joined technology or IT or computer or programming, as I mentioned, in the 90s. Why? Because I started with Linux, I installed Linux and I started to learn with Linux. This will be a story that many of our listeners will understand very well. So you get this like black screen in front, you don't know anything about it, but you have this passion to learn and you end up learning by yourself. Programming in C in common Lisp. I mean, that was my door into technology. And why am I mentioning this? Because my actual Academic path was archeology, so that's what I did. My bachelor's and master's it was in archaeology while pursuing a career. So it was a bit of an. I'm a bit of an oddball in that my career was entirely in technology. I started as a programmer, then went into Linux and Unix administration, then joined IBM for technical solutions from IoT quantum computing ended up in Kubernetes because Linux obviously has a huge importance. When cloud native started, all my background in open source and Linux was right there. And currently I'm working at the SAS Institute as a cloud and architecture lead for Europe. But going back to Penelope, my knowledge of obviously the ancient Greek myths made me. I was a co organizer of Kubernetes Community Day Sportu a couple of months ago and there was a cybernetics exposition just on the room next to the KCD one. And I mentioned in passing that cybernetics and Kubernetes actually come from the same Greek root which is pilot, helmsman, et cetera. And that made me think and by that time I still hadn't the theme completely settled. Made me think, well actually this is interesting because a release cycle is a voyage, is a sort of an epic voyage where a journey where teammates help each other. But I still wasn't absolutely settled until I remember that actually the voyage from Troy to Ithaca of Odysseus took 10 years. And that like I remember that we are on the 10th anniversary of Kubernetes and after that I just actually introduced this in my release comms to the mailing list. I started with a small quotation of the Odyssey which is applicable to the specific release week that I was in. And in the last third of the cycle I settled on Penelope because I also on top of the 10 years and on top of the Kubernetes and journey theme also Penelope ended up with a loom, right? And she weaves by day but unweaves by night because she has all these suitors that they want to marry her, but she wants to wait for Odysseus obviously. And that actually reminded me of the way that the Kubernetes release cycles work because we have these deprecations and then we have the new alpha, beta and stable features. So we end up as a community building this very large project, the second largest open source project in the world. We end up doing a bit like Penelope. We wove things in and unwove others as we progress. So yeah, I mean several different layers here from the very simple it's just a cute Thing that I end up liking because of my past in archaeology, but also because I think it makes Sense on the 10th anniversary of Kubernetes and plays a certain homage to the Greek roots of the Kubernetes name.
- [14:26]
Kaslan Fields
Excellent. There are so many layers there and I always love to hear about the release theme. I think it gives such a human angle to the release. It is all people behind the scenes and I love how it ties to your own background, coming from archeology and going into technology, just because you were passionate about it. I wanted to call out that when we were getting this episode started, we did have some technical difficulties with the audio and we were worried for a moment there that perhaps it was not going to be the year of the Linux desktop. But don't worry folks, we fixed the audio issues.
- [15:01]
Frederico Munoz
Yeah, that's a recurring joke. But I've been on the Linux desktop for 20 years now, so it's not that everything works at first, it's just that I've learned to live with whatever limitations exist. They still are not enough for me to use something else.
- [15:21]
Kaslan Fields
So, speaking of limitations, releases really are all about limitations for our users. They are about hopefully solving some limitations and perhaps introducing others. So what do we have going on in the 1.32 release? Can you give us some highlights of some of the enhancements that are happening?
- [15:42]
Frederico Munoz
Absolutely. So let me start by what we don't have in this release. So by the deprecations, we actually don't have a major deprecation. Like we had some releases in which we had some very impactful deprecations. That's not the case here. We have just one deprecation, which is not something that we will need to write blog articles about to warn people, etc. I mean, it's nothing like the removal of Docker Shame and what a time. Yeah, exactly. So that's one of the aspects of this. I Mean, we have 44 enhancements in new enhancements in this release, 18 Alph, around 12 in beta, and 13 in stable, which is a good number, just around the average for this. Obviously the release article will identify the highlights that we think are most significant. And amongst those highlights, for me specifically, I think that the dynamic resource allocation changes are worth mentioning, both because of the deprecation and because of the things that enter. So this goes back actually to the theme and the idea of we continuously move things in and move things out. So what happens in this release is that we have several new enhancements that improve dynamic resource allocation very simply and without going into the nitty gritty details. DRA is a part of Kubernetes that allows the resource allocation for workloads that require specialized hardware. Obviously in this day and age with AI and data intensive workloads, etc. This can mean GPUs, FPGAs, network adapters, whatever. So it's obviously a very important aspect of the way that Kubernetes evolves to meet the demands that are expected of it. And we have several different DRA enhancements which are very, very useful. Namely, I think that in terms of the deprecations, actually it's also the array, right? What happens is that we add things to dynamic resource allocation like structured parameter support, for example, that allows the cluster autoscaler to actually work directly without needing a third party driver. You can read this on the article and there's a link to the caps and kubernetes and descriptions, et cetera. At the same time, we remove, we deprecate the old DRA implementation, right? So why do I think this is noteworthy? Because it shows the way that features, the lifecycle of features and the way that Kubernetes release end up being this dynamic process in which things are added, they are tested, they are applied in the field, they get reviewed, we get the feedback from the users, we end up, in some cases it's perfectly applicable and they get more incremental improvements. In other cases like the ra, we end up going to in a different direction to better meet these new challenges and we remove some parts of the previous implementation. So if I had to choose one, I think that this is an interesting one because it shows that Kubernetes is a very, very dynamic piece of software. And the way that the cycle works in terms of features has this ability to respond to the way that features are used or not used and incrementally or more radically improve and replacement. So that is certainly one of the major ones. There are others. I mean now it's all a matter of taste and some people will like a bit more the ones around POD level resource specifications, specifically some of the changes that made it in 2032 in terms of being able to set resource requests and limits at a POD level. Some of the work also done, but unfortunately not completely completely done in this release, but some of the work made it. But in the next one I think this will likely be even better, which is the ability to have POD resources resizing dynamic while it's running. I personally like for example, the ability to take snapshots of volumes. So as a Group, right? It's volume group Snapshot API is introduced in 1.32, so this is obviously very interesting for backups in order to maintain data consistency. So for me, these are a couple of the good examples of what 1.32 brings to the table.
- [20:51]
Kaslan Fields
I feel like these major enhancements that are happening in 1.32 kind of have a common theme in Kubernetes growing in terms of the different types of workloads that it supports. We mentioned AI, of course, but also stateful. When Kubernetes first came out, it was very focused on stateless workloads, but running stateful workloads on Kubernetes is commonplace now, I would dare to say. And so we've seen more and more features added into Kubernetes to enable stateful workloads, which inherently involves managing new kinds of hardware for the cluster. And so AI also needs hardware management type of capabilities, so we have the dynamic resource allocations. And of course with stateful you need backups as well. So we've got a couple of kind of thematics, major updates in 1.32 there. And I did want to call out. You mentioned there was one deprecation. It's actually a removal, right, of a beta API.
- [21:49]
Frederico Munoz
Yep, exactly.
- [21:51]
Kaslan Fields
The V1 beta 3 API version of Flow Schema and Priority Level configuration.
- [21:57]
Frederico Munoz
Yeah.
- [21:58]
Kaslan Fields
So if you use the V1 beta 3 version of Flow Schema and Priority Level configuration, watch out and make sure that you change that API version for 1.32.
- [22:09]
Frederico Munoz
Exactly. Yeah. That is the one API removal that exists in this release. So it's that removal of the Flow Control, Flow Schema and Priority Level configuration. Then we also have the withdrawal of the old DRA implementation and actually also the deprecation of the git repo volume types to respond to a security issue that was identified. So it was deprecated as well.
- [22:34]
Kaslan Fields
Excellent. And there's a few others that were called out early in the sneak peek at 1.32. And of course as this is coming out, there should be new blogs about all of the features that are going to be in 1.32. At some point I will stop being completely amazed perhaps in the number of features that go into these releases. I still feel like 44 is quite a lot, but I guess it is fairly standard in terms of recent releases at least.
- [23:02]
Frederico Munoz
The idea of trying to make a comparison by release of this, that would be interesting, the amount of enhancements we have that information, obviously in terms of the articles and of the features, there will be feature blogs Coming out in the weeks after the release, we already have most of them closed and ready to go. So that's another aspect of the release which I find very interesting. So we not only guarantee the documentation of the features, but we for those that warrant specific highlighting by the sigs, we will have feature blogs that will go into the more technical details, use cases, why it is important. And one aspect of this is that I actually learned a lot as well. I mean the reality is that it's impossible to have full knowledge of everything that goes on in Kubernetes. I personally have a superficial knowledge of a part of Kubernetes, both from a codewise perspective. But even as a release team lead there are lots of things that pop up which I wasn't aware existed. So I think that part of the reward of working in a release cycle is actually being sort of forced to learn the interesting stuff that's getting in. And sometimes you end up finding things that will pique your interest and you will delve a bit more into them. And I think that's a very interesting side effect of being elite, but more in general a member of the release team.
- [24:40]
Kaslan Fields
Yeah, that flow control, flow schema removal, I don't even know what that is. I don't know that I've seen that my entire time in Kubernetes. So there's always something new to learn and the release is a great place to find a whole bunch of new things to learn. And you have been involved with releases for at least two years, right? We have three releases per year for Kubernetes. And you've been through what, seven then?
- [25:08]
Frederico Munoz
I mean I joined in 1.25 so seven. So yeah, and I joined, I wouldn't say by accident, but it was through the obviously normal process which is just a couple of some months after we started working in seek contrabacks comms I applied for the shadowing program, which by the way will be. By the time the release gets out, there will be a shadowing questionnaire out for the 1:33 release.
- [25:38]
Kaslan Fields
So if you're interested in being part of the release team, make sure you fill out that form.
- [25:42]
Frederico Munoz
Exactly. So I filled it out to the best of my ability without any great expectations. But fortunately Kat Cosgrove was the comms lead for 125 and she picked me for being a shadow in comms. And then I was the comms lead and then it was my foot in the door in the. In the release. That initially experience actually was very, very interesting because it allowed me to. I Mean in a way, go back in time in terms of community involvement and again feel what it means to do this kind of community based development with a constant rhythm.
- [26:25]
Abdel Sigiwa
Right.
- [26:25]
Frederico Munoz
Because what I'm trying to say here is that you release as a start, as an end, and it requires you to actually do things in a certain pace. It's not something that you can easily, or you shouldn't at least just stop doing it because it has a timeline and there's a certain amount of pressure there. So I found that pressure interesting in the way that it made my weekly life slightly more interesting. So even if sometimes it can seem a bit too much, obviously with the day job, family, other stuff going on, I always come out in the end thinking that I've been fortunate to be able to participate. And even when sometimes I think, nah, this is too much, then I just need one week without doing it. And at the end of that week I'm already ready to go. So yeah, it's from 1:25 to 1:32. It's a substantial amount of time which allowed me actually to learn different things in the release cycle and in a way prepared me for the role of release lead, which is mostly around being able to depend on others in a gracious way. So there's a certain amount of humility that you need to learn in actually in order to be the lead of anything. So I think that I've learned that and I'm thoroughly enjoying the experience.
- [27:53]
Kaslan Fields
As his contributor award says. Frederico is an absolute boss.
- [27:58]
Frederico Munoz
You're very kind. Yes, I still have that pizza.
- [28:03]
Kaslan Fields
Yeah, he got one of the pizza awards. The Pizza Netties.
- [28:05]
Frederico Munoz
Yeah, exactly.
- [28:06]
Abdel Sigiwa
And.
- [28:06]
Frederico Munoz
But it's carry woods. No.
- [28:11]
Kaslan Fields
Oh, Chop wood, Carry water award. Yeah, but it's chop pizza, carry boxes. Yes, exactly.
- [28:16]
Frederico Munoz
Chop pizza, carry boxes. Exactly. That's it.
- [28:19]
Kaslan Fields
We like to have a little bit of fun with our awards. But in all seriousness, Frederico has wonderful not just technical skills but also management skills that you've brought to your contributions in open source kubernetes. So it sounds like as makes sense. That was an important part of you coming up to this release lead nomination.
- [28:39]
Frederico Munoz
I mean, I find it very hard to speak about that. It goes back to what I've said before. I mean, partially, obviously each one of us has their own life experience and some of the things that I lack are made compensated by others that I bring. And the good thing is that we all have this. So in the end, working together ends up working pretty well. Given my role and my professional experience. There are certain things that come more naturally to me, others less so. So we end up trying to put into action those things that can happen, help the projects and at the same time depend on others to cover the things that we're not. So experience in and also learn. And that's I think that learning for me, it's one of the fundamental things that keeps me going in this projects because I think I'm still very far away from saying I've learned everything that I could from the Kubernetes project. And so that's what kept me reapplying to the release team. And I'm sure that I'm just a this is a typical experience for most people that keep contributing because one thing that I found in common with others that keep contributing because that's a problem with open source projects in general. You have a very peaks of participation but sometimes not as much a steady flow of contributors that stick around. But the ones that end up sticking around, you end up even without knowing exactly how you end up picking up more responsibilities. And after a while you see others looking to you like they recognize something. But you I've just joined yesterday, but well, it was a couple of years and in the end it's the recognition of the work that we do that puts us in a position of being trusted.
- [30:42]
Kaslan Fields
And speaking of recognizing the work and bringing it back to the 1.32 release, I'm going to ask you one final question which is for all the those folks out there listening who have to run Kubernetes clusters and unfortunately update Kubernetes clusters, what advice do you have for them on why they should update to 1.32? Upgrade to 1.32 I suppose is the right terminology.
- [31:07]
Frederico Munoz
Yeah, I mean there are a couple of different ways to address that. I can start by the practical one, which will eventually everyone will update to 1.32 because I would think that if we consider that most people use for example one of the clouds managed Kubernetes flavors, eventually all the others will be deprecated. So 132 will be the one that.
- [31:33]
Kaslan Fields
We will start to use quicker than you want.
- [31:35]
Frederico Munoz
Exactly.
- [31:36]
Kaslan Fields
Long term support. Something. Something. Exactly. Exactly.
- [31:39]
Frederico Munoz
On the other hand, I think that just like every other release, there's a choice that people can obviously do. I'm not saying that everyone should jump into the new release as soon as possible. I would say this. If any of the enhancements that we've just discussed around POD resources, around volume snapshots, around dynamic resource allocation, if they scratch a specific itch that you're feeling right now that that isn't being met, then I would say this is a good reason for updating as soon as possible. If not, what I would say is that you should take this release as another one in terms of the update cycle of Kubernetes, and hopefully most people using Kubernetes will have this more streamlined approach to updating and avoiding staying in old versions for a substantial amount of time because that accumulates technical debt, accumulates problems, issues. So it's a good idea to update as a regular practice in this case to 1.32. But yeah, I think these are the two and the two possible ways to reply to your question. And I mean, it's not specific to 1:32. It's something that I think will be applicable to just about any release and the decision of upgrading or not.
- [33:13]
Kaslan Fields
And specifically to 1.32, I think there's not a lot of breaking changes here. There's a couple of removals, there's that beta API, there's a deprecation which won't directly affect anything, and then there's the dynamic resource allocation stuff. So if you're using any of that, check on it. But I think those cases will be a little bit rarer. They're not going to be a huge impact generally.
- [33:38]
Frederico Munoz
Exactly. There's not a deprecation that completely breaks something that is being used by just about everyone. So I think that 1.32, it's a good incremental release that will allow people to update when they think and when they feel that the features that it brings to the table makes sense or as a part of the regular keeping. Because obviously there's a lot of fixes as well. I mean, it's not just this is an important aspect. We're all focusing on the enhancements on the features, but there's a lot of work going on in terms of actual bug fixes. So each new release of Kubernetes addresses problems that existed before which are not highlighted as a feature, but they actually should add stability and should address problems that if you're not updating to 1.32, you're more likely to be facing them in your usage of Kubernetes.
- [34:41]
Kaslan Fields
So there you go, folks. That's Kubernetes 1.32. Make sure you check out the release blogs for more information and we'll see you there. Thank you so much, Frederico.
- [34:52]
Frederico Munoz
Thank you very much, Catherine, for having me. It was a great pleasure.
- [34:57]
Kaslan Fields
I'm so glad that we got to talk to Frederico for This release episode and that we've continued doing our release episodes. These are always a lot of fun to do and we're all the way at 1.32 here at the end of 2024.
- [35:09]
Abdel Sigiwa
Yes. I just have to say before, I think probably Frederico have been one of the easiest guests to coordinate an episode with this year. It always takes like couple of emails back and forth, but it was just so easy to do it. It was like kudos.
- [35:24]
Kaslan Fields
As I say in the interview, Frederico is an absolute boss. That is what his award says for from the contributor awards and it is because he is very good at these types of program management type activities.
- [35:37]
Abdel Sigiwa
Yeah. Makes probably Frederico a good release lead.
- [35:41]
Kaslan Fields
Exactly. Awesome. And so one thing that we started out talking about was the release theme and it was such a cool story that he had behind the release theme. It's always so interesting to see what the release leads do with it. Sometimes it's just silly for fun. Sometimes there's deep meaning behind it. This was one with some personal connections and also with deep meaning in a variety of ways. So it's super cool because it's Penelope, which is from the Odyssey. It's got the Greek connection with Kubernetes and connected to Frederico's own background in archeology.
- [36:17]
Abdel Sigiwa
Yeah, that actually was quite surprising as somebody who did archaeology for a part of their life, I guess, before switching to tech. Like, have Frederico been always on archeology or always in tech?
- [36:28]
Kaslan Fields
It sounded like always in tech. It sounded like he did his degree in archeology and then the call of technology was too strong. One of those folks got it.
- [36:38]
Abdel Sigiwa
Nice. Nice. Yeah. The theme is pretty cool, to be honest with you. When I was listening to the first portion of the episode, at least I heard Penelope and it pulled my attention because actually the name Penelope is a very popular name. Well, I'm not sure about popular, but it's a very common name in France. Like you have people in France that use that name for a female name. Right. And so when I heard Penelope, I was like, who is Penelope in this? I think it was even like the wife of a very famous politician in France was called Penelope something. I just don't remember the full name, but I was like, why are they talking about Penelope? And then I had to go Google it to figure out, oh, this is a mythology, a Greek mythology thing.
- [37:16]
Kaslan Fields
Oh, yes. You never read the Odyssey? You'll have to check that out sometime.
- [37:21]
Abdel Sigiwa
No, I have to check it out then. No, I have never read the Odyssey. No.
- [37:24]
Kaslan Fields
Yeah. We watched a movie on that in high school, one of those classics. Oh.
- [37:28]
Abdel Sigiwa
Oh, I see. Okay, cool. Nice.
- [37:31]
Kaslan Fields
It's got some really nice connections there. He even tied it in with the 10 years thing because of Odyssey's Odysseus 10 year journey home.
- [37:39]
Abdel Sigiwa
Nice.
- [37:40]
Kaslan Fields
So many layers. Very nice.
- [37:42]
Abdel Sigiwa
A lot of meaning.
- [37:43]
Kaslan Fields
Mm. And then we talked a little bit about the highlights of the release. 44 new features, I think Frederico said in this release, which every time it surprises me lately, I would really like to see a graph. We mentioned this in the episode. Love to see a graph of like the number of enhancements in each release over time.
- [38:04]
Abdel Sigiwa
Okay. That's pretty insane. 44 new features. Quite a lot, actually.
- [38:07]
Kaslan Fields
Yeah. I think we might have had one other release that we did an interview for which was higher. So I think it's like in the right realm for the last several releases at least. But it's still so many things when.
- [38:21]
Abdel Sigiwa
You have that many number of features. It's also a lot of work for the people that needs to get their code in before the cutoff date. Right. Like before the code freeze. So I assume it have been insane for a lot of maintainers.
- [38:31]
Kaslan Fields
Yep. And we always like to call out, especially the removals and deprecations that are in each release. So there was one removal of a beta API, so relatively low risk removal there. And then deprecations. There were a couple of course deprecations means that nothing is removed yet. It's just a warning that something is going to be removed in the future. So those should not be any kind of breaking change. One of them was an API and one of them was the first version of the dynamic resource autoscaler, the dra.
- [39:05]
Abdel Sigiwa
Interesting. Okay, so there was a new version of the API, basically that was introduced in 1.32.
- [39:10]
Kaslan Fields
Exactly.
- [39:11]
Abdel Sigiwa
Nice, nice. Cool. Sounds good.
- [39:13]
Kaslan Fields
And dynamic resource allocation, I think, is actually the a. Is.
- [39:18]
Abdel Sigiwa
Yeah, I think that that's what it means. Yeah.
- [39:20]
Kaslan Fields
Yeah. That is one of the core features, I would say, of 1.32, something we've been talking about a lot. I know I've talked about it with folks a lot recently. The ability to more dynamically manage your hardware resources within your Kubernetes cluster. It's very important in this day and age. We talked about its importance both for stateful workloads, which are increasingly commonplace on Kubernetes as well as AI.
- [39:46]
Abdel Sigiwa
Nice. Nice. Yeah, it's pretty cool to see DRA kind of being pushed as fast, I would say, or as quickly. I think it has Been in the making for a while, but it's good to see people kind of pushing it forward because there is a need for the kind of things with AI, right?
- [40:00]
Kaslan Fields
Yeah.
- [40:02]
Abdel Sigiwa
Nice.
- [40:02]
Kaslan Fields
Immediate need.
- [40:04]
Abdel Sigiwa
Exactly.
- [40:05]
Kaslan Fields
And then the in place pod auto sizing resizing as well.
- [40:09]
Abdel Sigiwa
Oh yes. That will be leveraged by vertical pod autoscaler. So you know, you can live resize a POD without restarting it. That's pretty cool actually.
- [40:16]
Kaslan Fields
Yeah. I really want to look more into how accessible that is in different environments because something that I like to remind people about is that it actually does take some work for Kubernetes to be able to work on different environments. You don't see it if you use a managed service from a cloud provider because they take care of that for you. But realistically it does take some plugins used to be used to be in tree provider code. And we talked earlier this year about the removal of all of the cloud provider code from upstream Kubernetes. So it actually does take a little bit of extra work. There's something extra that you have to install in the Kubernetes cluster for it to work in different environments, which will of course impact your ability to dynamically autoscale things in a lot of cases. I'm very curious to see if there are any restrictions on how easy that is to use in different environments.
- [41:07]
Abdel Sigiwa
I would have to assume that one of the restrictions would be pretty much how much resources are available on the node where that POD is running. Right, of course. So that's probably the main one obviously.
- [41:17]
Kaslan Fields
Which it has to know about in order to do that, which it has.
- [41:20]
Abdel Sigiwa
To be aware of. Then the other one I would assume also has to do something with whether your application running inside the POD actually support that. Right. Like I know for example that a Java application will not be able to leverage more memory just because the GV has to be set up with certain parameters when it starts. So you cannot just like live resize the POD and expect the GVM to use more memory without restarting it. Or at least that's how it used to be. It might have changed. Now it would be interesting to see how what the impacts of that from the application also point of view. Right.
- [41:49]
Kaslan Fields
I remember some stuff I think at Kubecon about Java workloads on Kubernetes specifically. Maybe it was related to this. That is an interesting consideration that I hadn't even thought of.
- [42:01]
Abdel Sigiwa
Yeah, not all programming languages or not all applications will be able to just leverage more CPU and memory that is available. I mean cpu, I think it's fine because it's a compressible resource, but memory is kind of interesting. And then I think the interesting part would be what happens if you scale down the pod, what happens to your memory. Right. So I think that there is probably some experimentation that needs to be done on these kind of things, at least.
- [42:24]
Kaslan Fields
From the Kubernetes side. Kubernetes doesn't have to kill your pod in order to give it more resources, but what that means for your application may vary.
- [42:33]
Abdel Sigiwa
Yes, I suppose the feature is there, so go try it out.
- [42:37]
Kaslan Fields
Yes, I definitely want to explore that. And then of course we talked quite a bit about Frederico's path to becoming release lead. We discussed his background already, that he has a degree, I suppose in archeology. New information for me from this conversation and got into open source in the 90s, but started contributing to Kubernetes, I suppose during the pandemic, started contributing to the comms team, which I helped to lead and kind of immediately applied for release team after getting started with comms a little bit.
- [43:11]
Abdel Sigiwa
Nice.
- [43:12]
Kaslan Fields
Got selected as a shadow, did several different roles within the release team and then eventually was nominated for release lead.
- [43:19]
Abdel Sigiwa
Yeah, so thank you very much Kaslan for that interview.
- [43:22]
Kaslan Fields
Thank you Abdel for chatting with me about it and I look forward to 1.32. Hope you all check it out.
- [43:29]
Abdel Sigiwa
That brings us to the end of another episode. If you enjoyed the show, please help us spread the word and tell a friend. If you have any feedback for us, you can find us on social media Kubernetespod or reach us by email at Kubernetespodcastgoogle.com you can also check out our website at Kubernetespodcast.com where you will find transcripts and show notes and links to subscribe. Please consider rating us in your podcast player so we can help more people find and enjoy the show. Thanks for listening and we'll see you next time.