I’m thinking of building a home server with the following setup:
Intel NUC
Fedora CoreOS
Nextcloud HUB (Docker image)
When discussing this on the Nextcloud forum, some people suggested me to use the Nextcloud virtual machine instead of the Docker image. According to them, the virtual machine is easier to setup and to backup.
What do you think? Should I use the Nextcloud virtual machine or should I use the Nextcloud Docker image? What is easier to setup and to maintain? And is Fedora CoreOS suitable for my setup or is it overdone for a personal home server?
I have 2 instances of Nextcloud setup…a production version in KVM/Qemu and a pod-container version that I’m testing.
To be honest, the container version is a bit of a hassle. Nextcloud isn’t a great candidate for an “application container”. It operation has lots of dependencies on os-level components (ie. cron, filesystem, copying / delete files, Libreoffice components, applications that require other OS components like Maps, etc…).
To be clear, it is possible to get a functional NC installation up and running as a pod container…however in my experience the cost-benefit analysis is worse than the experience of running in it in a VM.
You write that you use Nextcloud in a virtual machine. Did you use the official Nextcloud virtual machine image for this? So it does run on KVM? Or did you create your own virtual machine and installed Nextcloud on it using the archive file or the web installer?
I just used a Centos-8 virtual machine (but Debian / Fedora / Nextcloud VM would all would work fine). Yes, I am also using Cockpit to manage…using nginx for a similar reverse proxy setup like you mentioned.
Would be nice if Cockpit could configure/manage reverse proxy setups for VMs managed by cockpit-machines automatically, but sadly doesn’t exist AFAIK. To be honest though, each reverse-proxy VM exist as a separate file under /etc/nginx/conf.d…so I have /etc/nginx/conf.d/site1.domain.com, /etc/nginx/conf.d/site2.domain.com and each of those files are pretty much boilerplate only specifying the specific VM to reverse-proxy towards…so its pretty simple to manage from cockpit ‘terminal’.
Thanks again for the information! I think I’m going to build the same kind of setup as you did. Not sure yet which operating system I’m going to use (Fedora, CentOS Stream or Rocky Linux, etc).
It would be great if Cockpit included a reverse proxy to forward requests to the specific virtual machine. If someone has more than one virtual machines this is always necessary (AFAIK).
Now that CentOS is going to be converted in CentOS Stream: to which Red Hat distro are you going to move?
I think there are four options:
Fedora Server
But is it stable enough? Do I not get to much updates? The release cadence of 6 months is not a problem to me but a lot of small inter between updates could be.
CentOS Stream
The same for CentOS Stream (but doesn’t have the 6 months release cadence of Fedora Server).
RHEL
Stable but needs a individual developer subscription (16 installations). This of course not a problem but needing a subscription to install software is against my principles. Of course this is something that I could overcome.
Rocky Linux
Stable but doesn’t contribute to the development of the Red Hat products (as far as I know). They ‘only’ compile the code and create a RHEL clone. But it is without a subscription.
I dismissed Fedora Server and CentOS Stream (because of release cadence / too many updates).
I’m now only in doubt between RHEL and Rocky Linux. Do I want to go official and with a subscription? Or do I choose absolute freedom and go for Rocky Linux?
In the end it’s Red Hat that is doing most of the work (developing almost the complete Linux stack) so I think I will go for that. It’s a small ‘burden’ for supporting open source software and their documentation is very good.
Fedora is my daily workstation OS. CentOS was my server OS, outside of the office. I hope Rocky fills the gap left by CentOS. But I am currently looking AlmaLinux as well. Another RHEL rebuild that has a clean and polished ‘feel’ to it and has been faster to market with patches than most of the others. I am running a few systems with Rocky and Alma. Both are great, drop-in alternatives.
Thanks for sharing!
AlmaLinux also seems very interesting!
I only wander what they (AlmaLinux, Rocky Linux, etc.) give back to Red Hat?
For example: the company behind AlmaLinux develops KernelCare but doesn’t make it open source. So they they take all the open source software of Red Hat and then add their proprietary software to it. That doesn’t seem fair.
Not everything that benefits RH can be measured in opensource products released. RH themselves doesn’t opensource a bunch of their products. I would say that ANY RHEL rebuilder provides RH (and in turn, the opensource community) with significant benefit. By simply providing the service of rebuilding and making it available to the developer of the next ‘killer app’ and the thousands of engineers and hobbyists. I argue RH sees this benefit by the Developer offering they are making available now.
What did the CentOS project itself, prior to becoming part of the RH/IBM mega-olopy, provide back to the RH beyond what I stated above? I am in no way attempting to disrespect CentOS by that question – I believe they provided a great service. Just making my point.
Hoping for Rocky to become the next CentOS - until then I am keeping my mind and eyes open.
I read a nice article about this (see link below) and I think you are right. The new structure (being CentOS an upstream project) has the benefit that now project like Rocky Linux and AlmaLinux can more easily contribute back to the development of RHEL.
Could you share your configuration?
I mean how you configured Nginx to reverse proxy request to the two virtual machines?
Or do you know any good documentation about this (using Nginx or Apache)?