I have a successful single-node install of ICP 3.1.0 CE. I want to access the console using a fully qualified DNS name instead of an IP address, and have a public wildcard certificate which I wish to use to secure console access.
I was able to add the myhostname-only and myhostname.mydomain.com variants to the console and change the console to use my public certificate, so that is all working properly. But when I log into the console using myhostname.mydomain.com and look at the URLs associated with the interface items, some refer (correctly) to paths anchored at myhostname.mydomain.com... and some (e.g. Catalog, some items under Platform, etc) refer to paths anchored off the IP address.
Is there a way to change this behavior, such that FQDNs are used consistently throughout, without reinstalling ICP?
If not, and if the mixed results I see are because I did something boneheaded during install, can someone clarify what I should do to ensure that all paths post-installation are FQDNs instead of IP addresses?
Thanks!
I was unable to find a way to correct the discrepancies in URLs, so I followed Justin's lead... I deleted the cluster, explicitly set cluster_lb_address to the desired FQDN in config.yaml, and reinstalled. Now all URLs are FQDNs.
I don't know if this is the recommended way to fix this issue, or if it is a bug, or if I was simply missing something... but setting cluster_lb_address achieved the desired result.