Kubernetes KubeBuilder Controller Runtime and Controller Tools, 28 Jul 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Demo: New plugin grafana/v1-alpha by Tony (Google Summer Code 2022)

Description

Record from the Kubebuilder, Controller-Runtime, & Controller-Tools meeting on Jul 28, 2022

To know more about see:https://master.book.kubebuilder.io/plugins/grafana-v1-alpha.html
If you would like to propose changes or report issues over the plugin, please raise an issue in the repo: https://github.com/kubernetes-sigs/kubebuilder

Author: https://github.com/Kavinjsir

A

Okay, so hi everyone, so my name is Tony and I'm. Currently, a university student uh studying for like computer science uh and I was getting lucky to get admitted by uh the group Builder project in this summer for the gsoc project and I'm, currently working with camiller and brochure on developing the grafana plugin, which is a pretty new one.

A

So the attempt for this is we're trying to simplify uh the process for users to visualize the running status of the operators, so they don't necessarily need to do the invention or the creation of the Manifest on their own, because sometimes maybe, like the peoples from different teams, might have like uh duplications on creating the panels, because some panels might be in common, and so the other attempt for us to create this plugin is also so like things like uh by default.

A

The control runtime has provided a rich number of metrics for different perspectives, which is pretty good. So when you go to our documentations, so you can see that in the Matrix references uh you can see that there's a lot of numbers of metrics, that being provided by default from the control runtime. This is a good thing, um so also we think that maybe it's for some author of the operator. Maybe they don't necessarily need to get that much familiar with these default metrics.

A

Also, they don't might need to learn too much of Prometheus like the syntax of the prom ql or the functions, uh but they do have the same like expectations or the attempts to see what's happening to their running operators, and so, if our plugin can directly bring them the Manifest of the grafana Manifest and they can directly import it into their own website UI and they don't necessarily need to do the learning things and then they can directly see what's happening for their Simplicity, what's happening to their operators.

A

uh So ultimately, we hope that our we hope that our plugin can uh help users to visualize the status of the controller, help to improve their control, to make it more easier, more safe, more robust. So that's the basic idea, uh the original idea to have this to introduce this with a plugin, so I'm going to give a very brief Showcase of the current usage of this plugin.

A

It is not very complicated and it's pretty new, so it's in a very early stage, so the basic usage of it is to trigger by the init sub command or the edit sub-command. So as the definition of these two sub commands by init, it means when you initialize a new project. uh You can introduce this for final plugin and then you will have some manifest given IELTS. So, similarly, if you have the uh you, you have you directly have an existing project.

A

You can use the edit sub command and also with the plugins attached that you can have the uh the dashboards. So here I'll give an example of using the plugin in the memcache operator, which is a sample under the operator SDK.

A

uh uh So here uh so I'm gonna paint this to the Bro very.

A

So since this memcache operator is already an existing project scaffolded by operator SDK, we can use this edit sub commands with uh to introduce the new plugin I'm gonna. Do a computer edits the plugins and then you would have the a grafana folder inside the project and we give the Json file, which is directly the Manifest of the profana. So here we have the control resource, metrics, Json file and the control runtime Matrix Json file, so indicated by the name. So they're like trying to give the dashboards for different aspects. One is the resource.

A

One is the runtime, so I'm gonna uh copy the contents for one of them. So here I'm gonna copy the runtime Matrix.

A

And then in my uh grafana, I can just directly click the import and paste the content of the Json file and I can choose the relative folder for it. So here I just simply to use the operator. Ops availability and I'll choose the relative data source, which is Prometheus and I import it, and now we have the we will have the panels for uh the uh uh the status of the memcache operator, so I can switch to different time space uh time period to see what's happening in history.

A

So, as you can see uh in the current status of the plugin, uh the runtime Matrix Json file will bring you to Raw. One is the reconciliation, one is called over queue and in this reconciliation we simply give the total count of the red constellation and the error color of the reconciliation and scan the next rule. For the word queue we bring four panels on the left side.

A

uh We give the statistical uh result of the seconds of the items, stayed in the work queue uh before they've been requested, and, alternatively, we also give the seconds of the items get processing by the work queue and on the left side we have the ad rate for The Virtue and the retress rate inside of work queue. So that's basically uh what the control runtime metrics Json file can bring you about the status of the uh uh uh the your operator, uh and so we can do simply the same things for another Json file.

A

I'll quickly, do the Showcase uh again uh just copy the announcements this time the resource and then uh sorry.

A

Again, we can click the import and paste the content of the Json, and similarly we choose the folder and select the data source and click import. And now we have this uh new dashboard, which is the control resource metrics.

A

And basically you can see that we simply bring the CPU and memory usage of the operator so uh similar as like the the similar as the previous dashboards. You can click the time range to see what's happening in different time and that's four.

A

Okay. So that's basically the uh thank you.

A

So that's, basically the uh uh the current usage of the plugin, and so we have a currently running PR for uh the dashboards and that's exactly uh what is the the output I showed uh previously, and uh we have some ideas. It's like.

A

We want to uh we'll hope to bring more panels that for different perspectives, because sometimes my users may may want to see different aspects happening for the operator, but we currently just bring those simple panels, because we also want to receive the feedbacks from you and the community to see what kind of metrics is the highest priority, especially in a case that the current metrics given by control runtime is pretty like a pre-number.

A

So that's we want to see like what is the most important one so that we can bring in the plugin to show and also and the previous one in the current uh status. It's like. We try to uh separate the panels into different dashboards. Also, we try to separate the panels into different roles, so we also want to see like if different dashboards is what you expect or you think that's. The role is good so that we can have those different panels integrated into the same dashboard but separated by different roles.

A

So that's the basic idea that uh we would like to see what you and Community want to see and yeah. That's, that's the that's! That's the current things. That's that's all I can have for this. So if there's any questions or if there's something, that's anyone want to talk about.