For the longest time, this site's had some stability issues where it would silently crash from time to time. Before, I didn't really look into it too much since the problems seem to be rather deep, and I had the option to just restart the server when I needed it. Sometimes it would be down when I wanted to show it to people, so that wasn't great. Now, I have a bot that references a page of this site, and ultimately I decided to start looking into the issue.
I say the problems were rather deep because there were no console errors that reported back what was going on when the site crashed. It would just close itself without warning. After researching the issue, I found a clue when looking through the system logs. Apparently, the server was automatically shut down because it used too much memory. Sure enough, when using top to keep track of the process, the memory usage would just keep on going up and up upon requests, and eventually everything would blow up.
I tried switching servers, from the default WEBrick to thin, but that didn't fix the issue. At some suggestions, I tried looking into memory debugging gems and other methods to help locate the issue, but, probably at my impatience, none of this worked out. Some of the gems were deprecated, and other ways seemed too convoluted. In the end, I never figured out why that memory was increasing. It might've been a problem with sessions. I glossed over my code and didn't see anything out of the ordinary.
One potential fix was to use worker killers as a bandaid. Unicorn and Puma support worker killers, and from my understanding what they basically do is restart the server when it uses memory pass a certain limit. Not the most pleasant of solutions, and in Puma Worker Killer's own words, it's like putting a bandaid on a gunshot wound.
That's probably my last resort, but I didn't have to go so far. I stumbled on a Digital Ocean article about setting up a Rails app with Puma and Nginx, and I looked into that. Took me a bit to get through some humps, but things seem to be fine right now. I don't know if Nginx or Puma has some automatic stabilizing feature built in like the killers, but for the past day or two, my site seems to be stable. Whatever, I'll take it.