Unable to get machine name as FQDN in Portal for ArcGIS

[ashasuryawanshi]

Hi @cameronkroeker,

I was deploying Portal for ArcGIS v11.3 using Chef. After the successful execution of the Chef script, I noticed that the portal's FQDN URL, https://portal.doamin.com:7443/arcgis, was redirecting to the IP address. Upon checking the machine name in the portaladmin, it displayed the IP address instead of the hostname.

To troubleshoot this issue, I followed these steps:

1. I cleaned and reinstalled the portal.
2. I set the parameter ['configure_cloud_settings'] to false (which defaults to true).
3. I added the hostname in the hostname.properties file.
4. I manually installed the portal.
5. Changed hostname.properties file instead of IP, replaced the FQDN

I have followed all the above steps, but the issue is still not resolved. All the machines are the same, yet we are only experiencing this problem with the portal. I tried to manually change the IP address to the hostname from the backend configuration file, but it automatically reverted back to the IP address when the portal service gets restarted for reference, I have attached the portal chef script, portal chef logs, and portal admin machine name screenshot

portal.json

Portal_log.txt

[cameronkroeker]

Hi @ashasuryawanshi,

Is this issue happening in Amazon AWS EC2 instances? I am wondering if Chef is setting the hostname properties correctly, but then when the portal service is restarted the Node Agent EC2 observer is detecting that its in amazon and is overriding it to use ip.

Can you share the following file contents?

C:\Program Files\ArcGIS\Portal\framework\etc\NodeAgentExt.xml

It may look something like this:

<NodeAgent>
  <Observers>
    <Observer class="com.esri.arcgis.portal.observers.EC2StartupObserver" platform="amazon">
    </Observer>
   </Observers>
  <Plugins>
  </Plugins>
</NodeAgent>

Next, let's check to see if there is an environment variable named arcgis_cloud_platform and if so is it set to aws?

echo $arcgis_cloud_platform

Thanks,
Cameron K.

[ashasuryawanshi]

Hi @cameronkroeker,

I've checked both the configuration file and the environment variable as you requested.

Here are the screenshot of the C:\Program Files\ArcGIS\Portal\framework\etc\NodeAgentExt.xml file on our instance:

And here is the output from checking the environment variable:

Thanks,
Asha S.

[cameronkroeker]

Hi @ashasuryawanshi thank you for providing those. Looks like the NodeAgent EC2 observer will automatically use ip instead of hostname if it detects the environment variable arcgis_cloud_platform set to aws. Furthermore, the arcgis_cloud_platform environment variable likely got set to aws due node['cloud']['provider'] defaulting to ec2.

I think we can try to work around this by adding node['arcgis']['cloud']['provider'] attribute to the json config file:

{
   "arcgis": {
      "cloud": {
         "provider": "" <- leave this blank to see if it will override the default of ec2
      },
      "version":  "11.3",
      "run_as_user":  "arcgis",
      "portal": {
       ...
      },
      "run_list": [
        ...
       ]
   }
}

Before re-running with the newly updated json config file, ensure to remove the existing arcgis_cloud_platform environment variable. Then after the chef run completes, check to ensure it is not being re-created. This should hopefully allow the hostname values to stick and not be overridden by the EC2 observer within NodeAgent.

Thanks,
Cameron K.

[ashasuryawanshi]

Hi @cameronkroeker,

Thank you so much for the detailed information and the suggested workaround. We implemented the changes you recommended, and I’m happy to report that the issue is now resolved.

We appreciate you pointing us in the right direction. This has been a significant help.

Thanks again,
Asha S.