Shulou(Shulou.com)05/31 Report--

How to expand Linkerd 2.10, I believe that many inexperienced people do not know what to do, so this paper summarizes the causes of the problem and solutions, through this article I hope you can solve this problem.

Linkerd is the smallest and simplest service grid in the world. Since Linkerd 2.0, we have followed the philosophy of minimalism, combinability, and built on existing ecosystems. This idea may seem a little crazy, and I'll describe a feature of the upcoming Linkerd 2.10 that will make Linkerd smaller and simpler: extension.

In Linkerd 2.10, we split the default control plane installation of Linkerd into a basic deployment, excluding Prometheus, Grafana, dashboards, and other non-critical components that were previously released by default. As a result of these changes, the size of a basic Linkerd control plane at startup has now been reduced from ~ 500mb of Linkerd 2.9to below 200mb.

These components are now available as an optional extension, as well as several other components that are not strictly necessary for basic operations. The initial collection of Linkerd extensions includes:

Viz, which will contain all the cluster metrics stacks: Prometheus,Grafana, dashboards, etc.; multicluster, which contains all machines for cross-cluster communication; and jaeger, which will contain distributed trace collectors and UI.

Using extensions serves two purposes. First, it allows users of Linkerd to choose exactly which part of the Linkerd they want to install on their cluster. Second, it allows the Linkerd community to build Linkerd-specific operators and controllers without having to modify the core Linkerd CLI. There are more related content below.

How does it work?

Installing an extension is as simple as you would expect. For example, to install the viz extension, you can run:

Linkerd install-f-| kubectl apply-# install the core control plane

Linkerd viz install-f-| kubectl apply-# install the viz extension

(for Helm users: each extension will have a corresponding Helm chart. )

We also make third-party extensions as easy as possible to connect to the same system. For example, if the linkerd-foo binaries are found in the user's search path, calling linkerd foo automatically invokes and passes parameters to the linkerd-foo binaries. In addition, after installation, linkerd check automatically runs checks for all installed extensions and connects the output to a report.

No matter where it comes from, the extension should "feel" like the rest of the Linkerd.

Why would you do that?

As the adoption of Linkerd continues to grow dramatically, so does the set of use cases it has to deal with. For some users, out-of-the-box observability is a key reason for adopting Linkerd. For others, it is secure cross-cluster communication. Others are Linkerd's transparent default mTLS. This diversity of use cases is good, but it also puts pressure on the project-- especially when we focus on simplicity.

So far, we have dealt with this issue in a relatively ad hoc manner, including the custom installation process for multi-cluster components, the special "bring your own Prometheus" feature, and so on. Consistency can be achieved by transferring all of these mechanisms to the extension framework: these feature extensions can now be treated in exactly the same way.

Finally, the exciting idea is to allow the features of Linkerd to feel like the rest of Linkerd, but without the need to modify the core project.

After reading the above, have you mastered how to extend Linkerd 2.10? If you want to learn more skills or want to know more about it, you are welcome to follow the industry information channel, thank you for reading!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.