> Have you checked the number of rows in those 6GB mlog$ tables?
> Did you realize that the number of rows maybe did not really
> add up to 6GB, no matter how big each row piece could be? In
> that case, much of the 6GB is not occupied by real data, and
> you can safely get rid of much of the "fat" by moving the mlog$
> tables:
>
> alter table mlog$some_table move;
>
> optionally specify a new tablespace. After the move, check the
> bytes in dba_segments for those mlog$. I bet they will be much much
> smaller than 6GB. Quite often huge mlog$ are result of
> the snapshots not been refreshed for a long time, and that
> really pushed the segment size up. The extra space allocated to
> those mlog$ will not be reclaimed by subsequent succesful refresh,
> but a "alter table ... move" can.
>
thanks for the answer,
I did a move but the size is the same after that
Is there a timestamp I can use to delete some rows ?
Received on Wed Sep 29 2004 - 14:15:51 CDT
