Why does my server show 100% disk usage during peak casino game wiki traffic?

If disk spikes to 100% during busy hours, it's usually due to image-rich content. In your scenario, your wiki with casino screenshots will be the culprit. Forums generally aren't very resource-intensive unless they're being flooded by tons of media. When visitors browse through pages full of high-resolution images, particularly from their phones, your disk I/O gets hammered. If people are continually reloading dozens of pages, caching won't help much. And, because you don't have a CDN in place, each image request taxes your I/O even more. Check if your database also has grown significantly, but I would start with optimizing the serving of images. Try to compress them, or use a CDN so the burden is taken off your server.

And wiki images and forum attachments cause 100% disk utilization at high load. The media rush kills read/write performance. Try isolating static content to a CDN, and try caching live threads in the forums. Additionally, examine your database writes on every forum post. See what’s chewing up disk IO with iotop. It’s probably both.

That spike in 100% Disk IO activity is not related to CPU usage, but to data reads and writes (IO). So is probably caused by your wiki (especially large images) or the forum, maybe both. If you watch the logs around that time, you’ll see when people are accessing an image, uploading one, or the image cache is being updated. Also, if you have a busy forum, with attachments or logging enabled, that can trigger I/O. Using Iotop to track processes, or Cloudflare to find out where all that traffic came from would be a good start. If you have a lot of people accessing images, you need to serve those from a CDN or optimize them so they’re smaller. Or if your forum is causing it, use a caching plugin to reduce the number of queries, or migrate to something less resource-intensive. Bottom line: It’s a combination of what content you have and how many people access it. Observe and try things from there.