More streams problems - slow apply process

Managed to hit a bug in 10.2.0.2 streams a few days ago. Bug number 6163622, two sources users into one destination user with a large number of transactions causes the apply process to go V.E..R..Y V..E..R..Y......S..L..O..W.

It's a spill issue, if the number of transactions in one commit is too high (number of rows committed > TXN_LCR_SPILL_THRESHOLD value in the table dba_apply_parameters) then spilling occurs and the apply process then takes for ever to read back from the spill table and apply the change to the destination database. Look in v_$streams_apply_coordinator and you will see the queue with the problem stuck in appyling mode.

TXN_LCR_SPILL_THRESHOLD is set via :-

BEGIN
dbms_apply_adm.set_parameter(
apply_name => 'APPLY_E1STRM',
parameter => 'TXN_LCR_SPILL_THRESHOLD',
value => 20000 );
END;