In Redshift, how do you add multiple constraints to a table after creating the table? - foreign-keys

Question

In Redshift, is there a way to add multiple constraint to an already created table?
Step ONE, created the table (say month ago)
CREATE TABLE test_user
(
account_id BIGINT DEFAULT NULL encode mostly32,
ingest_timestamp BIGINT DEFAULT NULL ENCODE mostly32,
ingest_date TIMESTAMP DEFAULT NULL ENCODE delta32k,
user_id BIGINT NOT NULL encode mostly32,
group_id BIGINT NOT NULL encode mostly32,
department_id BIGINT NOT NULL encode mostly32,
name VARCHAR(255) DEFAULT NULL ENCODE LZO,
mobile VARCHAR(15) DEFAULT NULL ENCODE LZO,
phone VARCHAR(15) DEFAULT NULL ENCODE LZO,
user_language VARCHAR(15) DEFAULT NULL ENCODE runlength,
deleted INT DEFAULT NULL ENCODE mostly8,
created_at TIMESTAMP DEFAULT NULL ENCODE delta32k,
updated_at TIMESTAMP DEFAULT NULL ENCODE delta32k,
PRIMARY KEY (user_id)
)
sortkey (account_id)
Step TWO, created more tables like 'test_group','test_department'
Step THREE, after a month, I am planning to add multiple constraints to 'test_user' table like this(but in separate query)
ALTER TABLE test_user ADD CONSTRAINT FK_1 FOREIGN KEY (group_id) REFERENCES test_group (group_id);
ALTER TABLE test_user ADD CONSTRAINT FK_1 FOREIGN KEY (department_id) REFERENCES test_department (department_id);
The above succeeds.
However, Is there a way to add multiple constraints in single query?
Tried like this,
option 1 -> ALTER TABLE test_user ADD CONSTRAINT FK_1 FOREIGN KEY (group_id) REFERENCES test_group (group_id),ADD CONSTRAINT FK_1 FOREIGN KEY (department_id) REFERENCES test_department (department_id);
option 2 -> ALTER TABLE test_user ADD CONSTRAINT FK_1 FOREIGN KEY (group_id) REFERENCES test_group (group_id), FK_1 FOREIGN KEY (department_id) REFERENCES test_department (department_id);
But got this error for both case,
An error occurred when executing the SQL command:
ALTER TABLE test_user ADD CONSTRAINT FK_1 FOREIGN KEY (group_id) REFERENCES test_group (group_id), FK_2 FOREIGN KEY (department_id) REFERENCES test_department (department_id); REFERENCES helpdesk_user (user_id)
ERROR: syntax error at or near "FK_2"
Is there a way to add multiple constraint to an already created table?
Please help

Answer 1

It doesn't matter when you define them, or even if you define them at all.
Redshift ignores constraints. Defining them has no effect, except that the optimizer may use them as additional information when creating the access plan.
However, I have created and used massive (billions of rows) databases in redshift, and I didn't define a single constraint of any kind and it all ran amazingly fast.
Answer: Don't bother.

Answer 2

The ALTER TABLE documentation does not indicate that it is possible to specify multiple constraints in one statement:
ALTER TABLE table_name
{
ADD table_constraint |
DROP CONSTRAINT constraint_name [ RESTRICT | CASCADE ] |
OWNER TO new_owner |
RENAME TO new_name |
RENAME COLUMN column_name TO new_name |
ADD [ COLUMN ] column_name column_type
[ DEFAULT default_expr ]
[ ENCODE encoding ]
[ NOT NULL | NULL ] |
DROP [ COLUMN ] column_name [ RESTRICT | CASCADE ] }
where table_constraint is:
[ CONSTRAINT constraint_name ]
{ UNIQUE ( column_name [, ... ] ) |
PRIMARY KEY ( column_name [, ... ] ) |
FOREIGN KEY (column_name [, ... ] )
REFERENCES reftable [ ( refcolumn ) ]}
On a side note, Redshift Pitfalls And How To Avoid Them has an interesting examination of the impact of constraints on Redshift queries (see "Beware of Constraints" section).