This message was deleted.

Not a bad idea

thanks for the reply. can you give more context why you think so?

Like every great answer in the tech space - it depends. Normally I'd advise against it because you want your k8s nodes to be as ephemeral as possible. But depending on the applications you could get away with it. The cool thing about containers is they are portable in a way java could only dream of 😛. So instead of running them side by side why node build and test your apps as containers then redeploy the entire box all at once. if you want to have a local way of testing rancher-desktop comes will all the tools you need for a workstation. once you get it working in 1 k8s cluster - it will work in all of them 🙂

this seems like a bot response 🙃

Yup you caught me I'm totally a bot 🤣

sorry if it’s a genuine response

So this all comes down to what you are trying to do. Are you migrating production apps? Or can you afford some downtime. There a lot of ways to 'migrate'

we are a team of two. there are around 50+ servers to manage. there are two ‘clouds’. one of which has kubernetes. don’t ask about the version, please 😄

Great k8s can literally replace your entire data center and consolidate everything

in the last year we migrated our dev env from vagrant/docker(-compose)

Ok so is the goal to eventually move all the apps to containers then? Or do you want to keep 2 infrastructure styles in play?

the goal is to consolidate the stack as much as possible. so yes. move everything to containers. for the past week I’ve been reading up on k3s and I think/hope k3s is the way to go (k3d for dev on macOS).

I'm gonna say don't worry too much about the 'distributions' of k8s manifests work the same in all of them. It's just about what comes baked in by default. K3s is lightweight at the cost of not being as flexible as some other distributions. As for dev k3d is one of my personal favorites just because its simple out of the box and it runs natively on my Linux box (working at SUSE I run Linux on everything :)). The reason I recommend rancher desktop is it's a bit bigger but comes with more tools and a gui.

the reason I’m worrying about k8s distributions is because I didn’t have a choice before I joined where I’m at. Now is the time I can make a difference. In my team, we are two. 🙂

I really feel like step one is to start getting your apps into containers. Then start bringing up k3s in your production environment. Small bites will help the most here

at this time, I have one cluster on bare-metal 🙃

You're not stuck with anything - it's all vendor neutral and even platform neutral

that’s the dream 🙂

Sealed secrets are pretty cool but not quite as flexible as vault imo. That again all comes down to exactly what you're storing and how.

to my original question:

can anyone tell me how bad of an idea it is to start running k3s along 'regular baremetal services' and slowly starting to migrate everything to k3s provided I have enough CPU/memory resources for both?

Yes - it's doable. Unless your apps start interfering with k8s services on the node. This is why I say it depends...

I get you. I can get around that 🙂

as a migration path - sure as long as you're careful. But you're playing with fire in a way... I would have a solid plan to start shifting things over and dedicate the entire node for k8s

yeah. at the start I just want to have a path to switch back if necessary. but I’m worried about https://github.com/k3s-io/k3s/discussions/7337

Yea that's gonna be an issue. You should not run swap on k8s nodes at all. The terrible thing about this is it's hard to tell what will happen just because Linux swap doesn't really know how to swap k8s services and things start to break in strange ways.

dude/dette!!! thanks for your all your ingisht! 🙂 to you and

iamkiloman

🙂

I would ask that you please keep asking in public channels I watch from time to time but I'm not very reliably watching this channel

👍