Change primary key to auto increment

Question

I have a table Player, and a primary key playerID which is character(7). I have some records in this table already, and also have some other tables that has playerID as a foreign key, and those tables also have some records already.

How can I set the playerID to auto-increment? After reading for a while I think that I should have done this from the beginning, but since I can't do that now, is there anyway I can do it?

For example, when I run this

ALTER TABLE player ADD COLUMN key_column BIGSERIAL PRIMARY KEY;

it returns an error:

ERROR: multiple primary keys for table "player" are not allowed

and if I drop the existing playerID, records in other tables that reference it will be dropped as well.

Is there a way to "change" the existing primary key playerID to auto increment?

Answer 1

I figure it out: just add an auto-increment default value to the playerID:

create sequence player_id_seq;
alter table player alter playerid set default nextval('player_id_seq');
Select setval('player_id_seq', 2000051 ); --set to the highest current value of playerID

Answer 2

DROP SCHEMA tmp CASCADE;
CREATE SCHEMA tmp ;
SET search_path=tmp;

   -- create som data to play with
CREATE TABLE bagger
        ( player_id CHAR(6)
        , tralala varchar
        );

  -- populate the table
INSERT INTO bagger(player_id,tralala)
SELECT gs::text, 'zzz_' || gs::text
FROM generate_series(1,10) gs
        ;

SELECT * FROM bagger;

  --
  -- create the sequence, change the datatype and bind it to the sequence
  --
CREATE SEQUENCE player_id_seq;
ALTER TABLE bagger
        ALTER COLUMN player_id TYPE INTEGER USING player_id::integer
        , ALTER COLUMN player_id SET NOT NULL
        , ALTER COLUMN player_id SET DEFAULT nextval('player_id_seq')
        ;
ALTER SEQUENCE player_id_seq
        OWNED BY bagger.player_id
        ;
   --
   -- reset the sequence to containe the maximum occuring player_id in the table
   --
SELECT setval('player_id_seq', mx.mx)
FROM (SELECT MAX(player_id) AS mx FROM bagger) mx
        ;
SELECT * FROM bagger;
\d bagger

Output:

DROP SCHEMA
CREATE SCHEMA
SET
CREATE TABLE
INSERT 0 10
 player_id | tralala 
-----------+---------
 1         | zzz_1
 2         | zzz_2
 3         | zzz_3
 4         | zzz_4
 5         | zzz_5
 6         | zzz_6
 7         | zzz_7
 8         | zzz_8
 9         | zzz_9
 10        | zzz_10
(10 rows)

CREATE SEQUENCE
ALTER TABLE

 setval 
--------
     10
(1 row)

 player_id | tralala 
-----------+---------
         1 | zzz_1
         2 | zzz_2
         3 | zzz_3
         4 | zzz_4
         5 | zzz_5
         6 | zzz_6
         7 | zzz_7
         8 | zzz_8
         9 | zzz_9
        10 | zzz_10
(10 rows)

                                 Table "tmp.bagger"
  Column   |       Type        |                      Modifiers                      
-----------+-------------------+-----------------------------------------------------
 player_id | integer           | not null default nextval('player_id_seq'::regclass)
 tralala   | character varying | 

Answer 3

I don't think you can have 2 primary keys in one table, and because playerID data type is character(7) i don't think you can change it to auto increment.

So i believe you would have to remove the primary key constraint on playerID if you want to be able to add a new primary key.

Since you already have data on your table and you use playerID as a foreign key in other tables, i would advise you to duplicate your player table and test these changes on your second table to avoid corrupting your data.

But before you try all this, make sure that you are trying to make this changes using the same db-user that created the tables you are changing

Answer 4

The following code will set auto increment to an existing column in PGSQL:

ALTER TABLE schema.table ALTER COLUMN id 
  SET DEFAULT nextval('schema.table_id_seq'::regclass);