Wildcard DNS for CF domains may lead to loops with server dns name

[metahertz]

Have just hit this one myself so thought it would be worth mentioning.

If you have deployed cloud foundry dns by pointing *.yourdomain.net to your GORouter. Then you move this to point *.yourdomain.net to your SSLProxy instead (sensible) AND the SSLProxy VM has access to real world DNS, then nginx is going to resolve 0.router.yourdomain.net as its own IP and you'll get a loop, ending in the following cryptic nginx error page and nothing in the error.log in /var/vcap/sys/log/

400 Bad Request
Request Header Or Cookie Too Large
nginx

Just a heads up.

As our Gorouters have static IP's on our CF deployment, we have put a IP in the 'server' definitions instead of DNS for the time being to mitigate this without needing to also manage a seperate (from the wildcard) DNS zone.

Matt