jailmaker VS helm-charts

I run docker on scale using a Debian jail. It works great, and has basically no overhead or networking/device issues like a VM.

I decided not to go the Truecharts or even TrueNas charts way for running my apps and instead used this script to set up what's effectively a low-overhead VM to run Docker in: https://github.com/Jip-Hop/jailmaker

Personally, I just switched from Openmediavault to Truenas. Truenas handles storage much better. For docker apps I run a Debian jail, rather than a VM. Jails are more performant and easier to manage IMHO. This is what I used to create the jail: https://github.com/Jip-Hop/jailmaker

An alternative is this. I have migrated all my apps to a “jail” (basically a Debian vm that runs on top of the scale os). It gives you full control of your containers. You could use the same compose or docker run that you used on unraid to recreate the container.

I've been using the jail maker script https://github.com/Jip-Hop/jailmaker in bridge networking mode and it works great. Host networking (which is the default) is a bit annoying because you can use any of the normal ports 22/80/443 since they're all used by TrueNAS.

Either I'll spin up a VM and pass through a GPU or I'll try this out.

Check out https://github.com/Jip-Hop/jailmaker. I love it. Have all my containers and docker compose running through there. No need to recreate the wheel, mess around with gui apps and the like.

Also recommend you take a look at Jailmaker, which runs docker in a SystemD-based jail and will save you from a lot of potential grief in the future when Ix Systems changes something again.

100% OP. iX' path with SCALE and Truecharts is heartbreaking. Fencing themselves in to this ecosystem, deleting the Linux -open way of doing things with like docker or else. Hopefully they get the community vibe against k3s and will build something useful into truenas in the future. For now there is this LXC-like linux jails: https://github.com/Jip-Hop/jailmaker

Honestly, just use the bjw-s app-template for it. It's not complex enough to warrant a from-scratch chart.

I just spoke with one of the k8s-at-home guys, and he hadn't heard of Helmet either, but he recommended this: https://github.com/bjw-s/helm-charts which has a bit more miles on it, formerly it was published as: https://github.com/k8s-at-home/library-charts – this is in thousands of charts, from what I can tell.

What about your charts being a 1:1 copy of Bernd Schorgers' app-template, formerly kah-common?

Spinning up workloads in kubernetes is much different than just spinning up a container in docker or even with docker compose. If someone has not already packaged it in a helm chart or some other kubernetes workload you'll have to develop one yourself. There are some nice library charts you can use as a base that should handle just about any random docker image you want to deploy. https://github.com/bjw-s/helm-charts/tree/main/charts/library/common there is also a repo of pre developed charts for common images. https://github.com/k8s-at-home/charts but be aware it was recently deprecated so it won't be receiving any updates.

I'm using this chart - it's a swiss army tool for any self-hosted docker containers lacking own helm files.

It seems to have some overlap with a similar library that has been developed over at the k8s-at-home community and later migrated to my personal repo: https://github.com/bjw-s/helm-charts/tree/main/charts/library/common (shameless plug :P).

It very much depends on how you structure and setup your applications. We always try to use Helm charts which means within each environment you have your Helm values and some generic stuff. In some cases we created kinda a blueprint chart which was generic enough to configure whatever we needed. A good example of that could be https://github.com/bjw-s/helm-charts. This keeps the „copy-pasting“ relatively organized and low effort.