Project Lightning Talk: FluxCD - Gitops For All Sizes - Matheus Pimenta, Maintainer

This talk covers the features and capabilities of Flux, a GitOps tool designed for managing Kubernetes clusters. The speaker, Mateus Pimenta, explains how Flux automates CI/CD pipelines by reconciling the state of clusters from Git repositories, ensuring performance and security. Flux is designed as a lightweight set of Kubernetes controllers and integrates well with various Kubernetes features, including server-side apply. He highlights Flux's advanced capabilities such as drift detection, health checks, and the use of cell expressions for resource health status. Additionally, the speaker mentions a new open-source project called Flux operator that simplifies the management of Flux and introduces a user interface. Other enhancements include support for OCI artifacts and OpenTelemetry integration for tracing Flux resources.

Hello everybody. Thanks for being here today. My name is Mateus Pimenta. I am a Flux maintainer. And I work for Control Plane. So we're here today to talk about Flux and the scenario is uh you do DevOps, MLOps, AIOps and you need tools that will give you the the proper best practices, right? Everything has to work. So with Flux you can achieve that. You can automate your

CI/CD pipelines. It's a GitOps tool so you reconcile the state of your cluster from from Git, right? it's designed for performance, security. It integrates very well with the Kubernetes security model. It's designed as a set of Kubernetes controllers. So it has very comprehensive APIs, CRDs, right? So and it's also very lightweight. Um and Flux is trusted by several large orgs. For example, Azure and AWS. Azure even

has like Flux extension in their marketplace that you can install in your Azure clusters. Also for example in the telco space many telco companies adopt Flux. They even have this open source project on top of Flux which is called Sylva. It's like it provisions for you this management cluster that allows uh managing workload clusters. They they give a talk about it today in FluxCon. Really cool. Also

very large financial institutions adopt Flux for its ability to scale, right? Like because obviously manual operations don't scale, right? And Flux makes it possible for you to manage hundreds of clusters declaratively via GitOps. So that's why it's trusted by all these organizations. So yeah. We put the the effort so you don't have to. so a few features and abilities that Flux has, for example, I believe it

has the best Helm support in the landscape. We have a deep integration with the Helm SDK. We just released support for Helm 4, which came out late last year. And yeah, we work with the with the Helm to to get this done. They were really helpful. Thank you. also because Yeah, so it like the the way you can perform drift detection for you, it's really good in

in how Helm was designed. So Also Flux uses server-side apply natively. So this is a feature from Kubernetes that allows you to play nice with other tools, right? It's always a chaos when multiple tools are trying to manage the same resources inside the cluster. And Kubernetes built server-side apply for for making this nicer. And and Flux has this natively. Also we are really strong with health checks,

right? So when you have GitOps pipelines, you can't just apply everything and expect everything to work. Especially for example, if you're bootstrapping a cluster, right? things have to happen in a specific order. So, that is achieved by by a combination of a few features which are depends on, right? Like you can declare dependencies between different Flux resources like customizations and Helm releases and combined with health checks,

then you're sure that a dependency is ready so you can move on. And this is particularly really powerful in Flux with support from cell expressions. That That's a feature we have introduced very recently. So, you can like um in the ecosystem uh there are like there are many different tools that you have to deal with and they don't like various Kubernetes controllers and they the way they

um express the the healthy of their resources in the Kubernetes status field and in the in the state can vary, right? So, with the power of of cell expressions you can build expressions for each different resource kinds you have in your cluster. So, you can really express how um that resource is signaled as healthy for for the pipeline to move forward, yeah. Also, a lot more things.

Uh for example, we also have now at control plane introduced the open source project Flux operator which makes Flux management much easier. Like now you have this you can manage Flux declaratively. So, you can declaratively manage the tool that manages things for you in a declarative way. So, uh in Flux operator we have also built a a UI for Flux. This is a feature users asked for

a long time and it's having a lot of success. People are loving it. Also, we have in Flux operator the MCP server. So, yeah. Um type coding is inevitable. So, uh we are obviously following the trends and we have also just recently released agent skills. As well. So, yeah. Also, we have OCI artifacts which is the the way that we believe software distribution should go forward, right?

Like you you ship your container images with in in OCI registries and you can ship configuration as well. And the all the security tooling that that is built around it is useful for both things. And also recently we released open telemetry integration. Now you can have traces for the the flux resources. And there's more to come as well with you're going to be able to see a

deep tree of the traces like with proper correlation between the flux resources and everything. I guess that's it. Um Scan our code and I'll be there to talk to you if you're interested. Yes, thanks. All right. Thank you very much.