A foreign key is a column (or combination of columns) in a table whose values must match values of a column in some other table. FOREIGN KEY
constraints enforce referential integrity, which essentially says that if column value A refers to column value B, then column value B must exist.
For example, given an orders
table and a customers
table, if you create a column orders.customer_id
that references the customers.id
primary key:
- Each value inserted or updated in
orders.customer_id
must exactly match a value incustomers.id
, or beNULL
. - Values in
customers.id
that are referenced byorders.customer_id
cannot be deleted or updated, unless you have cascading actions. However, values ofcustomers.id
that are not present inorders.customer_id
can be deleted or updated.
Details
Rules for creating foreign keys
Foreign Key Columns
- Foreign key columns must use their referenced column's type.
- Each column cannot belong to more than 1
FOREIGN KEY
constraint. - A foreign key column cannot be a computed column.
Foreign key columns must be indexed. This is required because updates and deletes on the referenced table will need to search the referencing table for any matching records to ensure those operations would not violate existing references. In practice, such indexes are likely also needed by applications using these tables, since finding all records which belong to some entity, for example all orders for a given customer, is very common.
If you are adding the
FOREIGN KEY
constraint to an existing table, and the columns you want to constraint are not already indexed, useCREATE INDEX
to index them and only then use theADD CONSTRAINT
statement to add theFOREIGN KEY
constraint to the columns.If you are creating a new table, there are a number of ways that you can meet the indexing requirement:
- You can create indexes explicitly using the
INDEX
clause ofCREATE TABLE
. - You can rely on indexes created by the
PRIMARY KEY
orUNIQUE
constraints. - New in v19.1: If you add a foreign key constraint to an empty table, and an index on the referencing columns does not already exist, CockroachDB automatically creates one. For an example, see Add the foreign key constraint with
CASCADE
. It's important to note that if you later remove theFOREIGN KEY
constraint, this automatically created index is not removed.
Tip:Using the foreign key columns as the prefix of an index's columns also satisfies the requirement for an index. For example, if you create foreign key columns(A, B)
, an index of columns(A, B, C)
satisfies the requirement for an index.- You can create indexes explicitly using the
Referenced Columns
- Referenced columns must contain only unique sets of values. This means the
REFERENCES
clause must use exactly the same columns as aUNIQUE
orPRIMARY KEY
constraint on the referenced table. For example, the clauseREFERENCES tbl (C, D)
requirestbl
to have either the constraintUNIQUE (C, D)
orPRIMARY KEY (C, D)
. - In the
REFERENCES
clause, if you specify a table but no columns, CockroachDB references the table's primary key. In these cases, theFOREIGN KEY
constraint and the referenced table's primary key must contain the same number of columns. Referenced columns must be indexed. There are a number of ways to meet this requirement:
- You can create indexes explicitly using the
INDEX
clause ofCREATE TABLE
. - You can rely on indexes created by the
PRIMARY KEY
orUNIQUE
constraints. - New in v19.1: If an index on the referenced column does not already exist, CockroachDB automatically creates one. It's important to note that if you later remove the
FOREIGN KEY
constraint, this automatically created index is not removed.
Tip:Using the referenced columns as the prefix of an index's columns also satisfies the requirement for an index. For example, if you create foreign key that references the columns
(A, B)
, an index of columns(A, B, C)
satisfies the requirement for an index.- You can create indexes explicitly using the
Null values
Single-column foreign keys accept null values.
Multiple-column (composite) foreign keys only accept null values in the following scenarios:
- The write contains null values for all foreign key columns (if
MATCH FULL
is specified). - The write contains null values for at least one foreign key column (if
MATCH SIMPLE
is specified).
For more information about composite foreign keys, see the composite foreign key matching section.
Note that allowing null values in either your foreign key or referenced columns can degrade their referential integrity, since any key with a null value is never checked against the referenced table. To avoid this, you can use a NOT NULL
constraint on foreign keys when creating your tables.
A NOT NULL
constraint cannot be added to existing tables.
Composite foreign key matching
New in v19.1: By default, composite foreign keys are matched using the MATCH SIMPLE
algorithm (which is the same default as Postgres). MATCH FULL
is available if specified.
In versions 2.1 and earlier, the only option for composite foreign key matching was an incorrect implementation of MATCH FULL
. This allowed null values in the referencing key columns to correspond to null values in the referenced key columns. This was incorrect in two ways:
MATCH FULL
should not allow mixed null and non-null values. See below for more details on the differences between comparison methods.- Null values cannot ever be compared to each other.
To correct these issues, all composite key matches defined prior to version 19.1 will now use the MATCH SIMPLE
comparison method. We have also added the ability to specify both MATCH FULL
and MATCH SIMPLE
. If you had a composite foreign key constraint and have just upgraded to version 19.1, then please check that MATCH SIMPLE
works for your schema and consider replacing that foreign key constraint with a MATCH FULL
one.
How it works
For matching purposes, composite foreign keys can be in one of three states:
Valid: Keys that can be used for matching foreign key relationships.
Invalid: Keys that will not be used for matching (including for any cascading operations).
Unacceptable: Keys that cannot be inserted at all (an error is signalled).
MATCH SIMPLE
stipulates that:
Valid keys may not contain any null values.
Invalid keys contain one or more null values.
Unacceptable keys do not exist from the point of view of
MATCH SIMPLE
; all composite keys are acceptable.
MATCH FULL
stipulates that:
Valid keys may not contain any null values.
Invalid keys must have all null values.
Unacceptable keys have any combination of both null and non-null values. In other words,
MATCH FULL
requires that if any column of a composite key isNULL
, then all columns of the key must beNULL
.
For examples showing how these key matching algorithms work, see Match composite foreign keys with MATCH SIMPLE
and MATCH FULL
.
CockroachDB does not support MATCH PARTIAL
. For more information, see issue #20305.
Foreign key actions
When you set a foreign key constraint, you can control what happens to the constrained column when the column it's referencing (the foreign key) is deleted or updated.
Parameter | Description |
---|---|
ON DELETE NO ACTION |
Default action. If there are any existing references to the key being deleted, the transaction will fail at the end of the statement. The key can be updated, depending on the ON UPDATE action. Alias: ON DELETE RESTRICT |
ON UPDATE NO ACTION |
Default action. If there are any existing references to the key being updated, the transaction will fail at the end of the statement. The key can be deleted, depending on the ON DELETE action. Alias: ON UPDATE RESTRICT |
ON DELETE RESTRICT / ON UPDATE RESTRICT |
RESTRICT and NO ACTION are currently equivalent until options for deferring constraint checking are added. To set an existing foreign key action to RESTRICT , the foreign key constraint must be dropped and recreated. |
ON DELETE CASCADE / ON UPDATE CASCADE |
When a referenced foreign key is deleted or updated, all rows referencing that key are deleted or updated, respectively. If there are other alterations to the row, such as a SET NULL or SET DEFAULT , the delete will take precedence. Note that CASCADE does not list objects it drops or updates, so it should be used cautiously. |
ON DELETE SET NULL / ON UPDATE SET NULL |
When a referenced foreign key is deleted or updated, respectively, the columns of all rows referencing that key will be set to NULL . The column must allow NULL or this update will fail. |
ON DELETE SET DEFAULT / ON UPDATE SET DEFAULT |
When a referenced foreign key is deleted or updated, respectively, the columns of all rows referencing that key are set to the default value for that column. If the default value for the column is null, this will have the same effect as ON DELETE SET NULL or ON UPDATE SET NULL . The default value must still conform with all other constraints, such as UNIQUE . |
Performance
Because the foreign key constraint requires per-row checks on two tables, statements involving foreign key or referenced columns can take longer to execute. You're most likely to notice this with operations like bulk inserts into the table with the foreign keys. For bulk inserts into new tables, use the IMPORT
statement instead of INSERT
.
You can improve the performance of some statements that use foreign keys by also using INTERLEAVE IN PARENT
, but there are tradeoffs. For more information about the performance implications of interleaved tables (as well as the limitations), see the Interleave tables section of Performance best practices.
Syntax
Foreign key constraints can be defined at the table level. However, if you only want the constraint to apply to a single column, it can be applied at the column level.
You can also add the FOREIGN KEY
constraint to existing tables through ADD CONSTRAINT
.
Column level
Parameter | Description |
---|---|
table_name |
The name of the table you're creating. |
column_name |
The name of the foreign key column. |
column_type |
The foreign key column's data type. |
parent_table |
The name of the table the foreign key references. |
ref_column_name |
The name of the column the foreign key references. If you do not include the ref_column_name you want to reference from the parent_table , CockroachDB uses the first column of parent_table 's primary key. |
column_constraints |
Any other column-level constraints you want to apply to this column. |
column_def |
Definitions for any other columns in the table. |
table_constraints |
Any table-level constraints you want to apply. |
Example
> CREATE TABLE IF NOT EXISTS orders (
id INT PRIMARY KEY,
customer INT NOT NULL REFERENCES customers (id) ON DELETE CASCADE,
orderTotal DECIMAL(9,2),
INDEX (customer)
);
CASCADE
does not list objects it drops or updates, so it should be used cautiously.
Table level
Parameter | Description |
---|---|
table_name |
The name of the table you're creating. |
column_def |
Definitions for the table's columns. |
name |
The name of the constraint. |
fk_column_name |
The name of the foreign key column. |
parent_table |
The name of the table the foreign key references. |
ref_column_name |
The name of the column the foreign key references. If you do not include the column_name you want to reference from the parent_table , CockroachDB uses the first column of parent_table 's primary key. |
table_constraints |
Any other table-level constraints you want to apply. |
Example
CREATE TABLE packages (
customer INT,
"order" INT,
id INT,
address STRING(50),
delivered BOOL,
delivery_date DATE,
PRIMARY KEY (customer, "order", id),
CONSTRAINT fk_order FOREIGN KEY (customer, "order") REFERENCES orders
) INTERLEAVE IN PARENT orders (customer, "order")
;
Usage examples
Use a foreign key constraint with default actions
In this example, we'll create a table with a foreign key constraint with the default actions (ON UPDATE NO ACTION ON DELETE NO ACTION
).
First, create the referenced table:
> CREATE TABLE customers (id INT PRIMARY KEY, email STRING UNIQUE);
Next, create the referencing table:
> CREATE TABLE IF NOT EXISTS orders (
id INT PRIMARY KEY,
customer INT NOT NULL REFERENCES customers (id),
orderTotal DECIMAL(9,2),
INDEX (customer)
);
Let's insert a record into each table:
> INSERT INTO customers VALUES (1001, 'a@co.tld'), (1234, 'info@cockroachlabs.com');
> INSERT INTO orders VALUES (1, 1002, 29.99);
pq: foreign key violation: value [1002] not found in customers@primary [id]
The second record insertion returns an error because the customer 1002
doesn't exist in the referenced table.
Let's insert a record into the referencing table and try to update the referenced table:
> INSERT INTO orders VALUES (1, 1001, 29.99);
> UPDATE customers SET id = 1002 WHERE id = 1001;
pq: foreign key violation: value(s) [1001] in columns [id] referenced in table "orders"
The update to the referenced table returns an error because id = 1001
is referenced and the default foreign key action is enabled (ON UPDATE NO ACTION
). However, id = 1234
is not referenced and can be updated:
> UPDATE customers SET id = 1111 WHERE id = 1234;
> SELECT * FROM customers;
+------+------------------------+
| id | email |
+------+------------------------+
| 1001 | a@co.tld |
| 1111 | info@cockroachlabs.com |
+------+------------------------+
Now let's try to delete a referenced row:
> DELETE FROM customers WHERE id = 1001;
pq: foreign key violation: value(s) [1001] in columns [id] referenced in table "orders"
Similarly, the deletion returns an error because id = 1001
is referenced and the default foreign key action is enabled (ON DELETE NO ACTION
). However, id = 1111
is not referenced and can be deleted:
> DELETE FROM customers WHERE id = 1111;
> SELECT * FROM customers;
+------+----------+
| id | email |
+------+----------+
| 1001 | a@co.tld |
+------+----------+
Use a Foreign Key Constraint with CASCADE
In this example, we'll create a table with a foreign key constraint with the foreign key actions ON UPDATE CASCADE
and ON DELETE CASCADE
.
First, create the referenced table:
> CREATE TABLE customers_2 (
id INT PRIMARY KEY
);
Then, create the referencing table:
> CREATE TABLE orders_2 (
id INT PRIMARY KEY,
customer_id INT REFERENCES customers_2(id) ON UPDATE CASCADE ON DELETE CASCADE
);
Insert a few records into the referenced table:
> INSERT INTO customers_2 VALUES (1), (2), (3);
Insert some records into the referencing table:
> INSERT INTO orders_2 VALUES (100,1), (101,2), (102,3), (103,1);
Now, let's update an id
in the referenced table:
> UPDATE customers_2 SET id = 23 WHERE id = 1;
> SELECT * FROM customers_2;
+----+
| id |
+----+
| 2 |
| 3 |
| 23 |
+----+
> SELECT * FROM orders_2;
+-----+--------------+
| id | customers_id |
+-----+--------------+
| 100 | 23 |
| 101 | 2 |
| 102 | 3 |
| 103 | 23 |
+-----+--------------+
When id = 1
was updated to id = 23
in customers_2
, the update propagated to the referencing table orders_2
.
Similarly, a deletion will cascade. Let's delete id = 23
from customers_2
:
> DELETE FROM customers_2 WHERE id = 23;
> SELECT * FROM customers_2;
+----+
| id |
+----+
| 2 |
| 3 |
+----+
Let's check to make sure the rows in orders_2
where customers_id = 23
were also deleted:
> SELECT * FROM orders_2;
+-----+--------------+
| id | customers_id |
+-----+--------------+
| 101 | 2 |
| 102 | 3 |
+-----+--------------+
Use a Foreign Key Constraint with SET NULL
In this example, we'll create a table with a foreign key constraint with the foreign key actions ON UPDATE SET NULL
and ON DELETE SET NULL
.
First, create the referenced table:
> CREATE TABLE customers_3 (
id INT PRIMARY KEY
);
Then, create the referencing table:
> CREATE TABLE orders_3 (
id INT PRIMARY KEY,
customer_id INT REFERENCES customers_3(id) ON UPDATE SET NULL ON DELETE SET NULL
);
Insert a few records into the referenced table:
> INSERT INTO customers_3 VALUES (1), (2), (3);
Insert some records into the referencing table:
> INSERT INTO orders_3 VALUES (100,1), (101,2), (102,3), (103,1);
> SELECT * FROM customers_3;
+-----+-------------+
| id | customer_id |
+-----+-------------+
| 100 | 1 |
| 101 | 2 |
| 102 | 3 |
| 103 | 1 |
+-----+-------------+
Now, let's update an id
in the referenced table:
> UPDATE customers_3 SET id = 23 WHERE id = 1;
> SELECT * FROM customers_3;
+----+
| id |
+----+
| 2 |
| 3 |
| 23 |
+----+
> SELECT * FROM orders_3;
+-----+-------------+
| id | customer_id |
+-----+-------------+
| 100 | NULL |
| 101 | 2 |
| 102 | 3 |
| 103 | NULL |
+-----+-------------+
When id = 1
was updated to id = 23
in customers_3
, the referencing customer_id
was set to NULL
.
Similarly, a deletion will set the referencing customer_id
to NULL
. Let's delete id = 2
from customers_3
:
> DELETE FROM customers_3 WHERE id = 2;
> SELECT * FROM customers_3;
+----+
| id |
+----+
| 3 |
| 23 |
+----+
Let's check to make sure the row in orders_3
where customers_id = 2
was updated to NULL
:
> SELECT * FROM orders_3;
+-----+-------------+
| id | customer_id |
+-----+-------------+
| 100 | NULL |
| 101 | NULL |
| 102 | 3 |
| 103 | NULL |
+-----+-------------+
Use a Foreign Key Constraint with SET DEFAULT
In this example, we'll create a table with a FOREIGN
constraint with the foreign key actions ON UPDATE SET DEFAULT
and ON DELETE SET DEFAULT
.
First, create the referenced table:
> CREATE TABLE customers_4 (
id INT PRIMARY KEY
);
Then, create the referencing table with the DEFAULT
value for customer_id
set to 9999
:
> CREATE TABLE orders_4 (
id INT PRIMARY KEY,
customer_id INT DEFAULT 9999 REFERENCES customers_4(id) ON UPDATE SET DEFAULT ON DELETE SET DEFAULT
);
Insert a few records into the referenced table:
> INSERT INTO customers_4 VALUES (1), (2), (3), (9999);
Insert some records into the referencing table:
> INSERT INTO orders_4 VALUES (100,1), (101,2), (102,3), (103,1);
+-----+-------------+
| id | customer_id |
+-----+-------------+
| 100 | 1 |
| 101 | 2 |
| 102 | 3 |
| 103 | 1 |
+-----+-------------+
Now, let's update an id
in the referenced table:
> UPDATE customers_4 SET id = 23 WHERE id = 1;
> SELECT * FROM customers_4;
+------+
| id |
+------+
| 2 |
| 3 |
| 23 |
| 9999 |
+------+
> SELECT * FROM orders_4;
+-----+-------------+
| id | customer_id |
+-----+-------------+
| 100 | 9999 |
| 101 | 2 |
| 102 | 3 |
| 103 | 9999 |
+-----+-------------+
When id = 1
was updated to id = 23
in customers_4
, the referencing customer_id
was set to DEFAULT
(i.e., 9999
). You can see this in the first and last rows of orders_4
, where id = 100
and the customer_id
is now 9999
Similarly, a deletion will set the referencing customer_id
to the DEFAULT
value. Let's delete id = 2
from customers_4
:
> DELETE FROM customers_4 WHERE id = 2;
> SELECT * FROM customers_4;
+------+
| id |
+------+
| 3 |
| 23 |
| 9999 |
+------+
Let's check to make sure the corresponding customer_id
value to id = 101
, was updated to the DEFAULT
value (i.e., 9999
) in orders_4
:
> SELECT * FROM orders_4;
+-----+-------------+
| id | customer_id |
+-----+-------------+
| 100 | 9999 |
| 101 | 9999 |
| 102 | 3 |
| 103 | 9999 |
+-----+-------------+
Match composite foreign keys with MATCH SIMPLE
and MATCH FULL
The examples in this section show how composite foreign key matching works for both the MATCH SIMPLE
and MATCH FULL
algorithms. For a conceptual overview, see Composite foreign key matching.
First, let's create some tables. parent
is a table with a composite key:
> CREATE TABLE parent (x INT, y INT, z INT, UNIQUE (x, y, z));
full_test
has a foreign key on parent
that uses the MATCH FULL
algorithm:
> CREATE TABLE full_test (
x INT,
y INT,
z INT,
FOREIGN KEY (x, y, z) REFERENCES parent (x, y, z) MATCH FULL ON DELETE CASCADE ON UPDATE CASCADE
);
simple_test
has a foreign key on parent
that uses the MATCH SIMPLE
algorithm (the default):
> CREATE TABLE simple_test (
x INT,
y INT,
z INT,
FOREIGN KEY (x, y, z) REFERENCES parent (x, y, z) ON DELETE CASCADE ON UPDATE CASCADE
);
Next, we populate parent
with some values:
> INSERT
INTO parent
VALUES (1, 1, 1),
(2, 1, 1),
(1, 2, 1),
(1, 1, 2),
(NULL, NULL, NULL),
(1, NULL, NULL),
(NULL, 1, NULL),
(NULL, NULL, 1),
(1, 1, NULL),
(1, NULL, 1),
(NULL, 1, 1);
Now let's look at some INSERT
statements to see how the different key matching algorithms work.
MATCH SIMPLE
Inserting values into the table using the MATCH SIMPLE
algorithm (described above) gives the following results:
Statement | Can insert? | Throws error? | Notes |
---|---|---|---|
INSERT INTO simple_test VALUES (1,1,1) |
Yes | No | References parent (1,1,1) . |
INSERT INTO simple_test VALUES (NULL,NULL,NULL) |
Yes | No | Does not reference parent . |
INSERT INTO simple_test VALUES (1,NULL,NULL) |
Yes | No | Does not reference parent . |
INSERT INTO simple_test VALUES (NULL,1,NULL) |
Yes | No | Does not reference parent . |
INSERT INTO simple_test VALUES (NULL,NULL,1) |
Yes | No | Does not reference parent . |
INSERT INTO simple_test VALUES (1,1,NULL) |
Yes | No | Does not reference parent . |
INSERT INTO simple_test VALUES (1,NULL,1) |
Yes | No | Does not reference parent . |
INSERT INTO simple_test VALUES (NULL,1,1) |
Yes | No | Does not reference parent . |
INSERT INTO simple_test VALUES (2,2,NULL) |
Yes | No | Does not reference parent . |
INSERT INTO simple_test VALUES (2,2,2) |
No | Yes | No parent reference exists. |
MATCH FULL
Inserting values into the table using the MATCH FULL
algorithm (described above) gives the following results:
Statement | Can insert? | Throws error? | Notes |
---|---|---|---|
INSERT INTO full_test VALUES (1,1,1) |
Yes | No | References parent(1,1,1) . |
INSERT INTO full_test VALUES (NULL,NULL,NULL) |
Yes | No | Does not reference parent . |
INSERT INTO full_test VALUES (1,NULL,NULL) |
No | Yes | Can't mix null and non-null values in MATCH FULL . |
INSERT INTO full_test VALUES (NULL,1,NULL) |
No | Yes | Can't mix null and non-null values in MATCH FULL . |
INSERT INTO full_test VALUES (NULL,NULL,1) |
No | Yes | Can't mix null and non-null values in MATCH FULL . |
INSERT INTO full_test VALUES (1,1,NULL) |
No | Yes | Can't mix null and non-null values in MATCH FULL . |
INSERT INTO full_test VALUES (1,NULL,1) |
No | Yes | Can't mix null and non-null values in MATCH FULL . |
INSERT INTO full_test VALUES (NULL,1,1) |
No | Yes | Can't mix null and non-null values in MATCH FULL . |
INSERT INTO full_test VALUES (2,2,NULL) |
No | Yes | Can't mix null and non-null values in MATCH FULL . |
INSERT INTO full_test VALUES (2,2,2) |
No | Yes | No parent reference exists. |