TL;DR: Managing 7 servers to run a project that no one uses is a bad (but funny) idea. Especially when you have no idea about what you’re doing.
Note: You’ve probably noticed that I don’t rewrite CS Realms/CS Servers in site updates. For some reason, I absolutely love the way it was written. Although my writing is much better now, I’ve decided to leave this page alone. It’s sort of a time capsule to a time where the site read a lot like this article. Enjoy!
At “CS” (not telling you the EXACT name because uh, privacy), I had an idea. More secure, and supercharged LAN worlds were needed. I exactly made that, being CS Realms. And then I did it again with CS Servers.
Both projects followed the same logic. Set up a half-a-dozen desktops with software to run and manage servers across all the “nodes”, and then get people to use them.
In the first rendition of this concept, known as CS Realms, I had about 3 nodes all running Debian 8.6 (this was a very different time of my life where I decided to use Debian instead of Ubuntu Server). Every node had a static IP, and power up after AC loss on. All nodes had an auto-startup script on them, and I think in the end there was something around 5 or so nodes running. I had developed forums and a website for it as well, so it was a big project. The backbone of this project was the now defunct PufferPanel, which had a nice web GUI, and allowed the making and managing of these servers across nodes.
The ambition of this project was great, and in my previous rendition of this page, I described this project like such:
“I absolutely love to work on this project”
“I want this to replace SCMC as a huge project I made and maintain”
“It’s a really gigantic project”
I can’t remember exactly why it failed, but I’m attributing it to that no one used it. Setup was painful from what I can recall. In January 2017, I decided to end CS Realms.
That is, until 9-10 months later, when CS Servers got started. This project was twice as ambitious as CS Realms, and much more annoying than CS Realms. This time, I gathered 7 nodes, slapped Ubuntu Server 16.04 on them with a GUI for not good measure, and started to get to work. This time, Pterodactyl was the backbone for this project. It did the exact same thing as PufferPanel, it was just under a new name, and had a nicer GUI.
Unfortunately, lots of problems plagued CS Realms, and managing 7 servers is easier said than done. At this point in CS, lots of kids really liked using mods, and Forge was a pain in the ass to run on Pterodactyl. Corruption issues that required manually renaming files was a huge annoyance, and the install time for just one server was easily 20 minutes. I tried to help bring that install time down, but it wasn’t an easy task.
Node 7 had lots of issues with its RAID setup, and every few weeks required a reinstall of Ubuntu Server. Doing updates and software upgrades took at least an hour each week, and the cron scripts sorta helped. Oh, not to mention that the daemon had issues starting up when the system booted, and that required more startup script changes.
Most damning was migrating people over to CS Servers, which required sitting in front of a kid for 30 minutes who just wanted to play Minecraft. Given the small amount of time I had at the start before all the little kids came, migrating to CS Servers would take a while. Oh, it isn’t like each kid makes new worlds every week however. That’s not a problem.
The worst part? With all the servers sitting idle, it overheated the server room, which actually shut down the internet equipment there, twice. I don’t want to imagine how hot it got in there, but at least we know in the future how to start up a freshly server-baked cookie business down the road.
In the end, I shut down CS Servers again mainly due to the fact that administering 7 servers is no easy task.
