They didn't use good, standard practices for handling the exception. It's an example of them not taking their own advice. Advice I'm sure you could find on the website, if it wasn't down.

Wouldn't a decent load-balancer/proxy server (e.g.: haproxy) in front of the application servers be a good idea to redirect traffic to a graceful "Ooops" page/server when something like this happens?

I often here this but it seems disingenuous - something is printing an error message, why could whatever that is not recover?

Why is it that people love to use the word "disingenuous" so much around here? I almost never see it used correctly.

If someone accused me in person of being "disingenuous" during an honest conversation, I would find it pretty rude.

The error message could represent defective hardware. If we did our job right, that's about all it would ever indicate. There's few sane actions one could take in an exception handler to remediate the problem. I will agree that a better overall system design could detect this problem and rotate the defective nodes, though.

But in this case for all we know it could be a software defect from which there is no recovery. IMO it's bad to overdesign recovery mechanisms that just end up masking design errors. Clever staged rollouts of changes are good ways to mitigate the impact of a new regression.

Hardware failure is one thing but in my experience it's just devs not taking the time to properly handle errors and exceptions. If something fails, restart the process and report the error silently in a log, not facing the customer.

The error is coming from IIS. There are several additional layers which could go down inc. CGI, Application Pool(s), Database(s), parts of the LAN (interconnecting different servers/services), or anything else the ASP.NET site relies upon (e.g. handle depletion, disk space, etc).

In IIS this is a "last resort" error when even the custom error handler cannot respond.

Microsoft themselves not changing the default IIS error page strikes me as funny.

Most companies have some sort of static 'technical difficulties' page to redirect you to instead.

That page can be shown if you set up custom error pages and those too fail to work correctly, then it falls back to this as a last resort.