Don Gollahon wrote:
<<
... about interminable garbage collection ...

>>

Have you looked at the selectivity of your indexes? Maybe I'm just hung up
on this, but anyway:
If you have deleted a lot of records where there are low selectivity
indexes, the cleaning up of these WILL TAKE A LOOOONG TIME. This is
because IB may/will have to search long lists of duplicate entries. Take a
look at the statistics for the indexes for the tables.
Or, if you know which of the affected indexes have low selectivity you
could just redefine them (drop the old ones, and define new ones where you
add the primary key (hope this is a single short field!) to the end of the
field lists. This could do wonders to your delete (ie. garbage collect) times.

I've also experienced the 'freeze at close time'. Unsettling. And
inconvenient.
I often do a commit and 'select count(*) ...' after massive deletes in
order to take the 'hit' there and then. However, IB's low priority garbage
collecting thread sometimes waits for a 'very low activity' period to do
its work. Charlie Caro wrote about it yesterday(?) on one of the Borland NG's.


--
Aage J.