ArcGIS Server federated with Portal installed with Builder: problem with FQDN in URLs.

[AlessandroValra]

I have installed ArcGIS Enterprise 10.5.1 with the automatic procedure of the Builder.

Everything went fine, all green checks.

I came up with having two working web adaptors, one for portal, one for server.

Everything is working when used from within the server where everything was installed.

The problem is the web adaptor took the FQDN of the server machine, so I was not able to sue it from another pc.

So I uninstalled and reinstalled both webadaptor pointing to a domain which is accessible from outside the server (and for which I have a SSL certificate for HTTPS connection).

I was then able to access portal and server from outside.

However, when I tried to logging in, a page with the Portal for ArcGIS layoutshowed up, but instead of the iframe to insert my credentials, I got a 404 error page.

Everything seems caused by the fact that during login, the authorization goes through the Portal and the URL somehow is still set to the FQDN, causing the request to time out (becasue, again, it is not recognized from outside the server).

I have changed the tier from ARCGIS_PORTAL to GIS_SERVER in the server administratrion page.

I was then able to login but then then I had issues with validating the federated server.

I would need to have the server federated with the portal because I would need to publish services to it from ArcGIS Pro.

How can I overcome the problem of the FQDN/EXTERNAL DOMAIN?

Should I give up using the Builder and do a clean install from scratch by soing thins manually?

Also, if you want to see what I have already been trying, please refer to this other thread.

Thanks

[ThomasPuthusserry]

Hi Alessandro / and other experts

I have come across your question after publishing mine. Basically in the same situation and struggling for a week.

Simply: have you figured out how to sort this? I managed to change the alias for the portal, but still unable to do this for the server. Any help / steps involved would be appreciated.

Thanks Thomas

[TomTurchioe2]

This also happens when folks install on, say AWS, and create an alias for a machine (like "frontend") but the hostname is something like WIN-A8EHN1234NK.  Enterprise Builder will "helpfully" grab the hostname and the person installing using Enterprise Builder specifies "frontend.dummy.nodomain" or some such make believe FQDN instead of WIN-A8EHN1234NK.ec2.internal or something like that.  Net net is the Enterprise Builder config that runs after the install will fail.  I always use the real hostname + ec2.internal and make sure the hosts file has "<real-hostname>.ec2.internal" in it in addition to <real-hostname>.  But since the doc doesn't call this stuff out, many people walk into this issue and can spend an inordinate period of time (as someone above mentioned) trying to troubleshoot.  I voted yes on the Ideas page link above.

[MuneerMajid]

Try and capture some of the calls being made using something like a fiddler or even just the chrome developer tools. I think that the /portal/sharing url here is still using the machine name and hence you might be getting a not found. You might be better off deploying this without using the builder.