Hello everyone.
Thanks to @thetarkus we have a deployment method called All-In-One Docker Container. This basically builds one image containing all the services which are needed to run Funkwhale: Funkwhale itself, a database, Celery and probably a few other tools.
There are two problems with this setup which is why @egon0 and I discussed to deprecate this setup method. This would mean we would stop building those images, we don't update this anymore, we do not suggest this method in our docs. We are aware there are a lot of people running the All-In-One deployment and we don't want to leave them behind: This step wont be made without providing a migration path and guide and provide support for the migration if there are any problems.
The two big issues with this deployment methods:
- This setup is technically problematic. By having everything in one container we are unable to detect failure of single services and restart them if needed, which does docker for the multi container setup.
- Its additional workload to maintain this setup method. Shortly we had the issue we are unable to reach out the maintainer of the Dockerfile, but the build was broken. We forked the project now, did the needed updates and provide those builds, but maintaining yet another way of deployment is an additional tasks. We need time we could spend building Features, Infrastructure or support our community.
I think from a technical perspective the situation is quite clear: This deployment is not ideal and we have a better one. But if you have some arguments about this, let me know!