The server sometimes stops working for some reason, the reason might be nginx not being able to handle a lot of connections at once. Which may be caused be the old UNMS?
Well after some time it pulls everything down with it, and everything except the webBIOS stops working, which is expected and clearly shows it's not an DNS issue.
Further monitoring will be required to understand this problem. I've reactivted the UNMS, so that the requests get passed through and not all go into nothing at the nginx, and as it will keep the connection open, they'll just pile up.
Let's wait and see...