Agree strongly. I think we have the best of both worlds in this topic too. I’ve already done a dummy migration from the centos server to the ubuntu one and I’ll likely move production shortly. Honestly, (for me at least) the install of centos I had was a pain. Docker “broke” in some way, shape or form everytime I updated Discourse. I’m sure someone who’s more into the nuaces with centos and docker would do better.
For me, I’d rather spend time continuing to build our community.
I used to think that, but now many problems that I solve will never happen again because the software will have changed. The likelihood that any Docker/CentOS problems will be the same in a year, or even a month, seems fairly slim.
The answers that I give again and again are being out of disk space or ram. Sadly, people don’t seem to find these on Google or anywhere else.
Software with fast release cycles is really hard to bring into enterprise grade stable distributions, where package updates are mainly for stability. My full respect for every developer who backports for example Kernel patches into such long term releases.
I had a chat a while ago with an engineer at RedHat, and especially in terms of security patches, Docker isn’t really fun to keep with. Still, for example the OpenShift platform relies on that.
I am coming from years of Debian unstable, Kubuntu, Fedora, and now macOS on my desktop. For servers, I’ve found the RedHat world more reliable when building my own RPM packages. I used to do a lot of packaging in the past, but that changed too. Debian and Ubuntu LTS do work fine for me as well. Different paths and commands are sometimes disturbing, but not really hindering.
Since we support many platforms with Icinga, we also do a lot of testing. I’ve seen Solaris, *BSD and other Linux distributions. They are all good for what they do, my preference for enterprise servers still is either RHEL/CentOS or Debian/Ubuntu. That also makes it easy to write Puppet manifests for better deployments.
On the long run, Ubuntu is probably the best for running Discourse. It also includes systemd as de factor standard nowaways. If you do need a more recent Nginx (e.g. as proxy for offline maintenance), there’s external repos for that too.
Agreed. Docker and many of the modern container tools like Kubernetes, Terraform, etc. are really hard to follow and keep track. I am always happy when Google leads me to a possible solution, which then brings me to another source, putting it altogether.
Rephrased: My main “investments” are over at monitoring-portal.org. In here I need to learn and chime into threads where I could help others with admin tipps. I like that a lot.