@unteem is not alone We work together.
We do this, because we host several discourse instances.
We started on docker, and now we run on kubernetes.
We moved our work under https://lab.libreho.st/ which is a community effort (@hellekin is also part of it ) . We want to advertise more our work in the coming weeks/months.
It is a real pain to maintain this… I literally spent hours, if not days for this commit that fixed my builds:
Any way. We are actually working on kubernetes operators. We start on Nextcloud, then RocketChat, and the next will probably be Discourse.
In the mean time, you can find the code for the docker images we use here:
The images themself are here:
https://hub.docker.com/r/libresh/discourse
https://hub.docker.com/r/libresh/discourse-web
As you can see we have been putting some time on it lately. So we have tags and pipelines. We need to add automation to have weekly builds.
We have an helm chart there:
But as you can see, it is not really maintained.
What I can say is that, it works for us If you want to share the ride and you feel adventurous, feel free to join us We have fun
We don’t really provide support, not much time for it, but if you PR, it will be welcomed. I really wish we could do this work under official discourse umbrella, this would be so much easier.
But at the end of the day, I start to understand discourse team. They have only one tool that they support for the community, and it works well. They provide good support for not so technical users, and this is really nice. If there is a problem, git pull && rebuild, it solves 99% of the problems I understand that supporting another tool is a big risk, and if not supported or not well done, it will harm the project in some ways. Again, thanks a lot discourse team for the hard work
My only problem is that we are probably many to develop our own solution, and the only way to collaborate is to collaborate here upstream.
Actually one way to do it, would be to have another image in discourse_docker that would be called super_base? without runit/anacron/nginx/postgres and the base would be based on super_base, and we could reuse it for our k8s deployment? I think this would work
What do you think?
Is discourse using kubernetes? I’m curious