What DBMS jobs and database links required for Materialized view replication?

Hi Gurus,

I have 2 sites, a master site on A and an MV site on B using primary keys on the replicated tables. The set up, briefly, and in no particular order, is:

On A (master):
* repadmin user

• schema owner
• master replication group
• snapshot logs for each table
• each table has a primary key defined
• each table belongs to the master replication group
• purge job
• a dbms_repcat.do_deferred_repcat_admin job - what does this do? I have read the manual but still didn;t understand what this does in real terms

On B (Materialized view)
* mvadmin

• 3 database links - 1 * public, 1 * mvadmin, 1 * schema owner
• Have updateable materialized views for each table
• Materialized view group
• all snapshots added to MV group so updatable
• refresh group
• have purge job * 1 hr
• have a refresh job * 1 hr
• push job from the MV to the master

The set up was a mix of using the Oracle master/mv site set up wizards and manually executed statements - hence in part my confusion.

Both sites are in a test environment so I have room to experiment.

My problem is that I am unsure about the relationships between the links and the jobs in an MV environment. Also, while I have no problem pulling the data down with the refresh job, the mlog$ tables are not cleared.

Even tho' I have set up a writable MV environment, this is really being used as one way replication.

I have hunted around the web but most articles refer to master to master replication which all specify a push job on each master site and database links.

My questions:

• do I need a push job from the master to the snapshot site?
• if so: do I need to create a database link/s between the master and the snapshot? who owns the database link? repadmin, public or the schema owner? how do I ensure that the purge job starts right after the p