Re: Incorrect Migrated/Chained rows...

MAK wrote:

> I am trying to reduce nos of migrated/chained rows from our database.
> I did following things to get them identified and corrected. The
> database is using 8K block size.
>
> 1) analyze table s_contact list chained rows into chained_rows;
> 2) select count(*) from chained_rows where table_name = 'S_CONTACT'
> count(*)
> ----------
> 488080

Crikey.

> 3) I got all the migrated/chained rows in following temp table.
>
> create table temp_s_contact_03
> pctfree 0

Well, that explains a lot. It isn't merely the fact that a table is subject to updates that causes row migration, but the fact that updated rows have nowhere to grow into. And you provide no room for row growth when you set pctfree to zero. If you're looking to cure row migration in the long term you need to increase this setting as well as move the affected rows.

>pctused 5
> inittrans 5
> unrecoverable

Why are you using Oracle 7 terminology? If this *is* Oracle 7, then fair enough. But if it isn't, stick to 8.0+ terms, which in this case would be "nologging".

For the rest of your post, please bump pctfree to something more sensible and try again.

Regards
HJR Received on Mon Apr 26 2004 - 22:34:29 CDT