So we have a stage server running Windows 2012 R2 server (It was upgraded from 2008 R2) It's running a bunch of asp.net MVC sites.
We used to run SSL sites on different ports to allow different domains before SNI. Now that we can use SNI with 2012 R2 we are moving the bindings from their old ports to port 443.
However now when we try to serve content up from the scripts folder over http on port 80 we get a 404 and over https on port 443, we get a sudden disconnect (ERR_CONNECTION_RESET in Chrome)
Here's the kicker: Adding another binding to a random port lets the content get served. It's only the scripts folder, making a copy of it to a different name or serving the same content out of the content folder works too. This was reproducible on a newly created (from scratch) web site.
Folder | http:80 | http:81 | https:443 | https:23457
content | works | works | works | works
scripts | (404) | works | (reset) | works
scriptsx | works | works | works | works
It feels like someone set something up somewhere deep in the bowels of IIS and then never told anyone else.
Turning on failed request tracing at the server has not shown any log entries.
Any advice would be appreciated.
Found this: https://forums.iis.net/t/1191833.aspx
Turned out we had qlikview server installed from a pilot phase long ago. It was not even enabled anymore. Uninstalling it restored normal functionality.
Seems to have happened to someone else recently too: https://community.qlik.com/t5/QlikView-Deployment/QlikView-blocks-Scripts-folder-for-asp-net-pages-on-same-server/m-p/113060