Zev Berezin wrote:
> Hi All,
>
>
>
> Tech Specs: O/S Solaris 9 ' IDS 7.31.FD7
>
>
>
> I was playing around with fragmentation recently when I came across a
> couple of interesting questions.
>
>
>
> Question 1)
>
> I came across a seemingly inconsistency when fragmenting a table that
> contained
>
> a Primary Key using a fragment by expression clause. Oncheck showed the
> P.K. got fragmented across multiple dbspaces same as the table however
> the underlying unique index that is system created upon
>
> creation of the P.K. remained in the original dbspace and was not
> fragmented.
>
> This was not the case however for a P.K. that was build upon an existing
> unique index that I had previously created.
>
> In that scenario the unique index got fragmented along with the rest of
> the table.
>
>
>
> Question 2)
>
> How could in the above case an unique index get fragmented upon
> fragmentation of table if creating it with the
>
> Same fragment by expression clause after the table gets fragmented
> results in a 872 error.
If you don't specify a fragmentation expression, the index will follow
the table, but since the engine no longer creates attached indexes, the
index will be semi-attached, ie following the table, but with independent
extents.
Art S. Kagel
> finderr 872
>
> -872 Invalid fragment strategy or expression for the unique index.
>
>
>
> The round-robin method cannot fragment unique indexes. If the
>
> expression method fragments the indexes, all the columns that are used
>
> in the fragmentation expressions must also be part of the index key.
>
>
>
> Any assistance is greatly appreciated.
>
>
>
> TIA,
>
> Zev Berezin
>
>
>
>
>
>
>