Carefully performing invalid operations to get the wrong error and the right result
The RavenDB setup process happens in the browser, and the last part involves restarting the server and then redirecting you to the new server. Along the way you have also specified a certificate to use and the other configurations.
We got a bug report about that when the admin configured us with a self signed certificate. During the time the server restarts, the browser will ping the server periodically, waiting for it to come back up with the new configuration. That can be a problem when using self signed certificates, because the browser will reject them as untrusted.
From the point of view of the client side running in the browser, there is no way to tell the difference from a server that is down and a server that is using a self signed certificate. But we wanted to get the nice feature of showing you when you can move to the new URL. So how can you do that?
Remember that RavenDB have auto detection for invalid HTTP access when using HTTPS? And that this error is raised at the HTTP level?
That means that we can carefully construct an HTTP request such as “http://my.ravendb.sever:443” and check what the result is. If the server is up, the request will fail with a bad request error, and that means that we can distinguish between the server being down and the server being up (but maybe with bad cert).
In fact, once we know the server is up, we can check if the certificate is valid, and show something about it.
This is convoluted, requires us to do things at several places at once at very different levels of the stack. But it is quite amazing to see, it just works!