I have a table that I have registered for propagation and have set up a CD table with a subset of columns.

I also have defined a subscription, again with just that subset of columns, and am propagating to a target table that is a condensed , noncomplete CCD table (I'm just after capturing updates to those columns).

The target table has a couple of propagator columns on it - ibmsnap_operation and ibmsnap_logmarker (approx commit time).

Trouble is that when a source table column that is not in that subset is updated, it still results in a row in the target table, with the operation and logmarker columns updated accordingly.

The manual seems to suggest that 'column subsetting' is a piece of cake. Is propagator doing what it's supposed to or have I got something wrong - any ideas ?