Multiple indexes using same columns on one table - is this doable?

Question

Hi
I have a table like MyTab(a int, b int), and I am required to create a primary key index and a non-unique index on this table using columns (a,b) in a specific table space.
The back end database is Oracle 10g.
Here's what I have tried so far, needless to say, unsuccessfully.
Alter Table MyTab
Add Constraint c_1 primary key (a, b)
Using Index (Create index mytab_idx on MyTab(a, b))
Using index tablespace results_index
So my question are:
1. is this is possible? if so, what is the correct syntax.
2. assuming it is possible, has anyone used this sort of construct before? it appears to be conflicting and inconsistent to me.
Many thanks, and kind regards,
Naran

Answer 1

Naran H wrote:
Hi
I have a table like MyTab(a int, b int), and I am required to create a primary key index and a non-unique index on this table using columns (a,b) in a specific table space.
The back end database is Oracle 10g.
Here's what I have tried so far, needless to say, unsuccessfully.
Alter Table MyTab
Add Constraint c_1 primary key (a, b)
Using Index (Create index mytab_idx on MyTab(a, b))
Using index tablespace results_index
You can't have the unique and non-unique indexes in place at the same time, but you can have the non-unique index supporting the unique (PK) constraint.
I guess your problem is that you don't want to have an interval of time where you have dropped the existing unique index while you create the replacement.
The only current way I can think of for getting from:
unique index(a,b)
to
non-unique index(a,b) with PK constraint (a,b)
without having a point in time where bad data can appear, and with minimal locking is:
create unique index (b,a) online;
drop index (a,b);
create non-unique index(a,b) online;
add primary key using index (a,b) enabled, novalidate
validate primary key
drop index (b,a)Regards
Jonathan Lewis