On Sep 11, 2013, at 9:16 AM, hogart wrote:
Hi Nate, Yes, db schema is old, but the upgrading (manage_db.sh upgrade) to the current version resulted to the error, as mentioned above. I don't remember the release number of the previous version of Galaxy, but it was installed in the March this year. A didn't restored the dump yet, now I am playinig with the newly created database. In my newbie point of view, it seems like, that dropping of the old database (drop database galaxy; as superuser) was incomplete, - is it possible? Thanks for letting me know, I will try to upgrade the PostgeSQL (the current version is default version of CentOS 5.6).
Sorry, I missed the bit where you were starting with an empty database and not reloading your dump of the old one. It does seem possible that your database is not empty on startup, as the username column should not already exist at the time that this migration script runs. When you drop and then create the database, Galaxy should create all the necessary tables and run through all of the migration scripts the first time that you start it. Is it doing this? If it thinks that the database is already at revision 13, I suspect the database is not empty when it starts up. Check your database_connection string in universe_wsgi.ini and make sure it is pointed at the correct database. --nate