Geoportal on an Amazon Virtual Machine
By default, metadata records that are harvested into the Geoportal are not editable. This includes any records in the Geoportal that were not created explicitly with interface provided by the web application ("Use dedicated editor to create metadata manually"). The SQL below creates a database trigger that allows harvested records to be edited.
Be aware that there is the potential for a lossy data translation when you do this. The harvested record is completely rewritten using the schema definitions that your Geoportal implements. If the original harvested record has any data that is not included in your schema implementation, that data will be lost after the original is edited. Another way to put that is, if you edit a harvested record, the edited version will only contain information that you see in the editor interface. If the original started with anything else in it, that will be lost.
ESRI's Geoportal wants an LDAP server, and suggests using Apache Directory Server.
I started a new VM using the image usgin2-atpms. Installation was as easy as:
wget http://apache.imghat.com//directory/apacheds/unstable/1.5/1.5.7/apacheds-1.5.7-i386.deb sudo dpkg -i /home/ubuntu/apacheds-1.5.7-i386.deb
It was easy to connect to via SSH Tunnel to port 10389. The next step would be to fix some issues with the instance and its data-store, remove the extra MySQL installation, and then move on to configuration of the PostgreSQL database.