Done it again this morning. Service re-started and all okay. I'll try to find out what is causing it. Perhaps it happens after the backup. I don't think its load related unless we get a huge rush of people in the early hours of the morning. I have done load testing in the past and the website recovers. I'm sure the answer is there somewhere in the huge log file.
Well reports back from the server engineers was a bit of a dead end. They have continual ping tests and it didn't show up it went down, nor in the server logs. We know the httpd service has stopped as it showed it had stopped, and starting it, it would come to life.
Website can use Apache or Nginx, it was on Apache, so trying the alternative for at least the static files. I'll run some checks. Fingers crossed its an improvement. Also changed some DNS records so its more of a direct route, perhaps a ping test was bouncing off the content delivery network as okay, but wasn't really coming from our web server.
Name servers updating in the next 24 hours. This is to not use Cloud flare. Cloudflare is a content delivery network, so much traffic like images and css files would be cached by them. However this seems to obstruct diagnosis if something goes wrong. Cloudflare was something I activated a long time ago as a test. Now I'm going to test without it. The server dosn't need it and more than capable of delivering pages.