Re: Using sqlldr chews up extra space?

From: Martin Mr <oa43_at_dial.pipex.com>
Date: 25 Jun 1999 22:04:00 GMT
Message-ID: <01bed6e8$dae12ee0$914995c1_at_dnyeaqvf>


The direct apth in sqllldr adds data only above the high water mark.So either truncate the table before the reload or don't use direct path.

Martin

Rick Osterberg <osterber_at_fas.harvard.edu> wrote in article <7l0m64$7t6$1_at_news.fas.harvard.edu>...
> Situation:
> Digital Unix 4.0x, Oracle 8.0.5. I'm using sqlldr to load in a ~25000
> record text dump file into a database.
>
> SQL> drop table userlist
> SQL> create table userlist (....)
> <run sqlldr>
> SQL> select count(*) from userlist
> returns about 25,000 (which is correct)
> SQL> select bytes from user_segments where segment_name = 'USERLIST'
> returns about 16 megs
> <run sqlldr again>
> SQL> select count(*) from userlist
> returns about 25,000 (still correct)
> SQL> select bytes from user_segments where segment_name = 'USERLIST'
> returns about 32 megs
>
> The .ctl file for the load specifies
> DIRECT=TRUE
> UNRECOVERABLE
> REPLACE
>
> But somehow, if I do the load twice in a row, the table size doubles,
even
> though the same data is in there. Anyone have any ideas why this
happens?
>
> -Rick
>
> --
>

+--------------------------------------------------------------------------+


> | Rick Osterberg osterber_at_fas.harvard.edu
 |
>
+--------------------------------------------------------------------------+

>
Received on Sat Jun 26 1999 - 00:04:00 CEST

Original text of this message