(k3s) Finding a starting point... #8959
-
|
Hello there! First, for a little bit of context: I started my journey into Kubernetes late november of last year and I have now finally gotten around to deploy my first node with k3s; in total, this will span three arm64 and one riscv64 node. However, some images that I want to use don't have a riscv64 version - and some other applications I wish to deploy don't even have their own CI/CD for container images at all. So I need to make my own! And after crawling ArtifactHub and many other k8s related resources, I ended up here. I read through the documentation and I believe to have (finally!) found what I was looking for: A way to specify a remote, not owned by me, Git repository and then run arbitrary commands, in order, to do something with it. But I am a little bit lost on a few details that I either didn't really get from the docs or that I might have misunderstood. Hence why I found my way here, hoping to find a few pointers. :) There are two main pipelines that I want to run for starting out:
Both of them are straight forward; just resources pointing to Git repositories to trigger a set of jobs when ready. Here is what I haven't managed to pick up from the docks yet:
Thank you a lot and kind regards, Ingwie |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
|
Hey, sorry for the late reply. So one issue I'm seeing right off the bat, assuming you want to run Concourse on your k3s cluster, is that your k3's cluster is running on arm64 machines. Sadly Concourse doesn't officially release arm64 binaries at this time. I know some community folks build and publish some binaries though. Now for your questions:
So how this typically works is you make a job that downloads a git repo that contains your shell script, and then execute the shell script in a More concretely, you can see here:
You can see other ways to configure resources. In this case you want to read the docs for the git resource: https://github.com/concourse/git-resource/ You'll see that it has
Concourse doesn't use the k8s scheduler when you run Concourse on k8s. Concourse schedules and manages its own containers. The guidance for running Concourse workers on k8s is that each worker has the entire capacity of a single k8s node. You can ignore this guidance if you're very careful and ensure you don't run anything that's too cpu/memory intensive. This is a bad idea in general though.
Concourse does not have built-in artifact collection. You need to push your artifacts somewhere before your job finishes, otherwise they will be deleted from disk shortly after your job finishes. Hope that helps. Feel free to ask more questions. |
Beta Was this translation helpful? Give feedback.
-
Oh that is very unfortunate... Yes, my cluster is indeed arm64 and will soon be joined by riscv64 as well. So I will have to look into either building my own images or see what the community has made so far. Oof... more custom builds, yay. x)
I took a look and although this is quite a heavy hitter for an example, I think I figured out what you ment. I just haven't fully found out how the individual files in
So it does not actually spawn new pods and runs everything in the same context? That's interesting. I would have thought that even just running a simple
Fair enough, there's a plethora of tools like Sharry, Zipline or even more specialized ones out there. :) Hey, it could even be a resource I can Thank you for the detailed answer! I have gotten Tekton to install in the meantime, but it is much more complicated and I like the Concourse approach more in terms of layout. I will look into the images and see what I can put together. If I come across anything, I will let you know! Thanks and kind regards, |
Beta Was this translation helpful? Give feedback.
Hey, sorry for the late reply.
So one issue I'm seeing right off the bat, assuming you want to run Concourse on your k3s cluster, is that your k3's cluster is running on arm64 machines. Sadly Concourse doesn't officially release arm64 binaries at this time. I know some community folks build and publish some binaries though.
Now for your questions:
So how this typically works is you make a job that downloads a git repo that contains your shell script, and then execute the shell script in a
taskstep. You can see ho…