This is probably not a big deal to most people, this is probably not even a big deal in general. But last night I had the highest number of served request per second of 40.83 on my cheap VPS that is also running half a dozen other sites including this one and a decent sized NextCloud instance that will absolutely cause the server to overload when it gets too busy. Somehow the server was able to keep up with the requests with very few errors. This was not always the case, but after months of optimizing and caching, it keeps up well.
My previous high was about 35k hits per hour (about 9.7 requests per second) and this easily eclipses that. This is 99.9999% AI bot traffic that is being used to train AI models so nothing that can really be monetized (yet) — but it is more or less a fun science project that I setup to learn how to create a dynamic Laravel controller that can handle sub-domains (41k+ sub-domains). I have been encouraged to write about how I am doing this, especially on such a low end VPS. I wish I had the time to really dive into it. I will say that since it has been such an un-expected side project I am actually planning on doing it even bigger with what will likely be 100k+ sub-domains routed with a single router on a very tiny VPS.
I also noticed my queue system is absolutely obliterated. To make page loading and analytics processing faster I use a queue system that processes out of band. I sometimes see this get to 8k or 9k… not 96k. Each hit generates about 3 queued jobs (page ranking, analytics, cache performance).
I have no idea how long it will take my tiny VPS to recover from this wave. I have never seen this many queued jobs… The wave does seem to have calmed down, but it is struggling to get that number down. This VPS has 1 shared CPU core and 2GB of RAM, you can see the load values are not actually that high (though kind of high for a single CPU VPS). That being said. The site loads fine, even the heaviest pages load in under a second.
