We should consider adding a note/blurb in the Oracle documentation that in some way raises awareness that Oracle requires a bit of tuning in order to run the connector in an ideal performing condition.

For example, Oracle XE images spin up with redo logs configured with 50MB each which is severely too small to store the data dictionary. For some Oracle EE environments with lots of tables and columns, logs may need to be sized upwards of 1GB to 2GB each depending on the size of the database's data dictionary.

This is a bit tricky because it's not as simple as providing a raw number. There is a certain amount of nuance in considering the database's traffic volume, other integration points that may be involved such as DataGuard and GoldenGate, etc.

The goal with this note is to make users aware that using Oracle isn't like other databases such as MySQL or PostgreSQL where you can spin up a database instance, start a connector, and it "just works". There is unfortunately a bit of care to making sure that Oracle LogMiner has the right space/resources to do its job effectively and efficiently.

CC gmorling@redhat.com broldan@redhat.com