The seamless integration is a bit of an issue. I would like to see more direct connectors and stuff like SQLAlchemy. It's there but isn't available as a standard, so it involves going to GitHub to access it.

Whilst our server up-time is in the > 99%, we still have to reboot the DBMS periodically to allow us to archive the log file. The ability to reset the log file without stopping the DBMS would be helpful for us.

We have created our own error management/reporting process within our database procedures. Unfortunately, if there is a database error, such as a duplicate key, the default action is to abort. This means that our information is generally lost. An option to not abort on failure would be great. Of course, this only really occurs in testing.