Archives of the TeradataForum

Message Posted: Tue, 29 Jul 2008 @ 20:15:54 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

Hi Michael,

RPT_MEMBER_UPD was created by joining rpt_member and another table that had consumer_counts. The rpt_member_id was pulled from rpt_member into the new table. This insert statement create 64,576 rows. I have verified that the rpt_member_id is unique in each table. After creation, if I join the two tables I get 2910 rows.

The NOT IN query returns about 61K rows, but every rpt_member_id returned by that query actually is in rpt_member_upd. The sample set of 7 records was meant to show that whatever sample I pull of records that are not supposed to be in rpt_member_upd are in rpt_member_upd.

I have also tried to cast both ids as integers during the match, but that has had no effect.

Could space issues or row locking cause only a portion of records in an update to succeed? More generically, does anyone have experience with joins missing data elements that should be captured in the join? Could this be related to distribution of data across the AMPS?

Thanks everyone for the help so far.

Thanks,

Ben

	<Prev	Next>		<<First	<Prev	Next>	Last>>