I started a coexistence upgrade from 2010 to 2013 a couple weeks ago, had 2013 CAS in front proxying OWA and web requests to 2010 and 2013 users alike, all running properly. Then "something" happened, not sure what, and suddenly all Exchange 2013 web requests to login to OWA or to ECP generate a 500 error on the server, no matter if I use localhost, IP address, or FQDN for the internal or external URLs.
Because this is production, I moved back to the 2010 CAS array as the primary and tried to tinker with 2013. Nothing would work, nothing would allow me to access mailboxes hosted on 2013, so I had to disable the mailboxes and re-enable them on 2010 (migration commands wouldn't even work to move them back to 2010). In the end I uninstalled all 2013 servers from my organization and checked ADSI Edit to ensure there weren't any pointers to the 2013 servers I installed.
Fast forward to this week, I wanted to try again, see if the uninstall of the servers fixed the issues. I built two new Windows Server 2012 servers (one CAS and one MBX) and accepted all default settings coming off of the CU2 installer (except one I chose the CAS role and one I chose the MBX role). Fresh install, no users running, I cannot get to the ECP or OWA once again, 500 error no matter what URL I use to access the servers. And proxying to a 2010 mailbox still fails as well.
I've checked out everything I can online, I've removed and re-added the Windows Identity Framework, I've recreated the OWA and ECP directories on CAS and MBX server, and I've run the get-serverhealth command against both servers for the OWA and ECP health sets without errors. Any ideas?
Environment: two Windows servers running AD services and the Hyper-V role, one Windows 2008 R2 SP1 with two virtual guests running Windows 2008 R2 SP1 servers with Exchange 2010 SP3 RU2 and one Windows 2012 with two virtual guests running Windows 2012 servers with Exchange 2013 CU2. Image