Golang backend, some light vanilla Javascript frontend. postgres db. running on podman (docker-compose), fedora server.
I have some background in C programming, so Go was easy to pick up. the standard libs are great for building server-side webapps with a sql backend. templates lib and some glue javascript make front-end pages a breeze.
I'm a linux guy thru-and-thru, so running fedora server on the latest kernel. latest podman from RedHat and docker-compose to orchestrate.
I'm one guy so I don't do any HA or cloud or any other funny business. My one server is racked nearby and I've had good history with the colo.
I already use Fedora Workstation and am familiar with the environment, install, and setup process. CoreOS' "happy path" seems to be cloud (or at least hosted in a hypervisor) -native, and that was not the target here. There seem to be external dependencies on installing it to bare metal I didn't want to mess with[1]. I installed Fedora Server from .iso almost exactly the same way I've installed Fedora (workstation) many times.
My "day job" uses rhel, and with the recent CentOS end-of-life debacle I'd like to stay both FOSS and .rpm-compatible with rhel. I don't know for sure, but I think CoreOS is a different "vertical" than the fedora/rhel/centOS family.
I'm generally wary and suspicious of "automatic" updates, preferring to control when/how updates are applied to my machines. $ sudo dnf system-upgrade is a well-known quantity to me.
I have a complicated LVM setup, and significant time/effort already invested into LVM. As of F34 LVM/ext4 is still first-class.
CoreOS seems to be targeting cloud/kubernetes as implementation characteristics, and my "one man shop" gig is the anti-that. I'm pretty old school - only just started learning Docker about a year ago. I've recently bought into containers as packaging/deployment aids, and using docker-compose to keep deployment in version control. I guess all of this is to say I'm trying to "keep it simple / stupid" and re-use what I already know. Unless CoreOS can give me order-of-magnitude performance gains, I don't know the value prop beats the cost to pick it up.
| If CentOS Stream is going to be just a FOSS upstream for RHEL, will you still pick Fedora Server over it?
Yes, but only because Fedora Server is (also/even more) a FOSS upstream for RHEL. Doing something that isn't Fedora (Workstation/Server) is "one more thing", however small, and I'd rather spend the brain tokens elsewhere. If it 10x'ed my app I'd do it, but I can't imagine it would.
Thanks for the links, I wasn't so mad at CentOS Stream the product itself per se, I think it's a valid product / change to the old model. My biggest issue is the changing of the promises - the supported life of the product changed in-production. That's a big no-no to me.
Fedora (upstream) requires you to keep up, but at least it doesn't make promises that it won't keep.
I have been using Fedora Workstation for a while now, and I am trying to pick the best single-server OS for all my future projects.
Since Fedora is now a direct upstream for CentOS Stream – and CentOS Stream 9 is basically a LTS version of Fedora 34[1] – I am leaning towards CentOS Stream.
In fact, CentOS Stream might now even make sense as a stable desktop OS, and as a better alternative to Ubuntu LTS.
I have some background in C programming, so Go was easy to pick up. the standard libs are great for building server-side webapps with a sql backend. templates lib and some glue javascript make front-end pages a breeze.
I'm a linux guy thru-and-thru, so running fedora server on the latest kernel. latest podman from RedHat and docker-compose to orchestrate.
I'm one guy so I don't do any HA or cloud or any other funny business. My one server is racked nearby and I've had good history with the colo.