UPDATE
On this page
UPDATE is a DML statement that modifies rows in a table.
Syntax
UPDATE table_references[[LEFT|RIGHT] JOIN [table valued expression] [ON join_condition]]SET column_reference=expr [ , column_reference=expr , ... ][WHERE where_condition][LIMIT row_count];
Arguments
table_
One or more tables to reference during the update operation.table_.table_ clause may be a comma-separated list of tables, or a join expression.table.), on the left of assignments in the SET clause, or single identifiers.table_ clause.
column_
A column in the specified target table to update, designated by either a single identifier or table..
expr
An expression that evaluates to a valid column-type value for the specified column.
where_
One or more expressions that evaluate to true for each row to update.
row_
The maximum number of rows to be updated.
Remarks
The UPDATE statement modifies each specified field in a row with new values.SET clause indicates which columns to modify and the values they should be given.WHERE clause provides any conditions that identify which rows to update.WHERE clause is not specified, all rows are updated.LIMIT clause places a limit on the number of rows that can be updated.
Although UPDATE supports referencing multiple tables using either join or subquery, SingleStore only supports changing one table in UPDATE statement.SET clause, all columns must come from the target table only.
The first field in the target table of type TIMESTAMP or TIMESTAMP(6) has special behavior for update operations, where it is updated to the current timestamp value when not explicitly updated with a different value.
In addition, a field f declared with the ON UPDATE <value> modifier will be updated to <value> if any other field is updated, but f itself is not updated.
UPDATE queries will fail if the maximum_ limit has been reached.
This command must be run on the master aggregator or a child aggregator node.
Updating columns which are part of the SHARD key is unsupported.
Transactions that span more than one database are not supported.Feature 'Cross-database transactions is not supported by SingleStore.
SingleStore does not currently support updating a reference table based on a join with a sharded table and returns the following error: ERROR 1706 (HY000): Feature 'Multi-table UPDATE/DELETE with a reference table as target table' is not supported by MemSQL.t1 as a reference table, the following update will return an error:
UPDATE t1 JOIN t2 ON t1.b = t2.b SET t1.b = t1.b+1;
To avoid getting an error:
-
Use the same type of table for update.
For example, use both the target table and joined table either as a sharded table or as a reference table. -
Use
UPSERTwith a reference table that has a primary key.For example, if t1is a reference table with a primary key, then the following query runs successfully:
INSERT INTO t1 SELECT t1.* FROM t1 JOIN t2 ON t1.b = t2.b ON DUPLICATE KEY UPDATE t1.b.
Refer to the Permission Matrix for the required permissions.
Limitations on an UPDATE/DELETE Target Table
Some restrictions affect the performance of UPDATE/DELETE queries:
-
The
UPDATE/DELETEtarget table cannot be moved that is, broadcasted, reshuffled, or have a hash table built on it. -
The
UPDATE/DELETEtarget tablet is enforced to be the first table in the join tree. -
It is not possible to seek into the target table while scanning another table.
Examples
UPDATE a SET c1 = 0;UPDATE a SET c1 = 0 WHERE c2 = 100;UPDATE a, b SET a.v = b.v WHERE a.name = b.name;UPDATE a LEFT JOIN b ON a.name = b.name SET a.v = b.v;UPDATE looooooooong as a, b SET a.v = b.v WHERE a.name = b.name;UPDATE a, b, c SET a.v = 0 WHERE a.x = b.x and b.y = c.y;UPDATE a, b, c SET a.v = c.v WHERE a.x = b.x and b.y = c.y;UPDATE b, a SET a.v = b.v WHERE a.name = b.name;UPDATE dataset SET valid = false WHERE v = (SELECT MAX(v) FROM dataset);UPDATE dataset SET valid = false WHERE name IN (SELECT * FROM invalid_names);UPDATE dataset SET v = v - (SELECT MIN(v) FROM dataset);UPDATE records a JOIN(SELECT name, COUNT(*) as count FROM samples GROUP BY name) bSET a.count = a.count + b.count WHERE a.name = b.name;
Updates Where More Than One Value Maps to a Target Row
Depending on the data, some update statements may try to assign more than one value to a single target row.
Example Non-deterministic Update
The following example shows the behavior of a non-deterministic update.
CREATE TABLE t1(a int, b int);CREATE TABLE t2(b int, c int);INSERT t1 VALUES (1,2);INSERT t2 VALUES (2,3), (2,4);UPDATE t1, t2 SET t1.b = t2.c WHERE t1.b = t2.b;Rows matched: 1 Changed: 1 Warnings: 0
SELECT * FROM t1;
+------+------+
| a | b |
+------+------+
| 1 | 3 |
+------+------+The target row’s b field is set to 3, but it could have been set to 4 as well.
Updates Using Joins
The following example uses the UPDATE . statement to perform a cross-table update without joining to a subquery.
SELECT * FROM product;
+----+------+------------------------------------------------+
| ID | Tag | Description |
+----+------+------------------------------------------------+
| 1 | US | {"description":[{"Name":"Ark","UID":"ghx"}]} |
| 2 | UK | {"description":[{"Name":"Canon","UID":"yad"}]} |
| 3 | US | {"description":[{"Name":"Peak","UID":"cmk"}]} |
+----+------+------------------------------------------------+SELECT * FROM stock;
+----+-------+----------+
| ID | P_ID | Quantity |
+----+-------+----------+
| 1 | 12 | 2914 |
| 2 | 32 | 3532 |
| 3 | 64 | 2268 |
+----+-------+----------+UPDATE stock sINNER JOIN product pON s.ID = p.IDSET s.P_ID = p.ID;SELECT * FROM stock;
+----+------+----------+
| ID | P_ID | Quantity |
+----+------+----------+
| 1 | 1 | 2914 |
| 2 | 2 | 3532 |
| 3 | 3 | 2268 |
+----+------+----------+Update Using Limit
LIMIT is used with an UPDATE query to limit the number of rows that will be updated.UPDATE to work, it must run on a single partition; otherwise, it will result in an error.
CREATE TABLE lmt_exp(item_id INT, item_name VARCHAR(20));INSERT INTO lmt_exp VALUES(1, 'widget'), (2, 'lgr widget'), (3, 'xl widget');UPDATE lmt_exp SET item_id=2 LIMIT 1;
ERROR 1749 (HY000): Feature 'UPDATE...LIMIT must be constrained to a single partition' is not supported by SingleStore Distributed.In this example, the table was created using a shard key.
CREATE TABLE lmt_exp (item_id INT, item_name VARCHAR(24), SHARD KEY(item_id));
INSERT INTO lmt_exp VALUES(1, 'widget'), (2, 'lgr widget'), (3, 'xl widget');
SELECT * FROM lmt_exp;+---------+------------+
| item_id | item_name |
+---------+------------+
| 1 | widget |
| 2 | lgr widget |
| 3 | xl widget |
+---------+------------+To constrain the UPDATE to a particular field, filter on a specific key of the SHARD KEY.
UPDATE lmt_exp SET item_name='med widget' WHERE item_id = 2 LIMIT 1;
SELECT * FROM lmt_exp;+---------+------------+
| item_id | item_name |
+---------+------------+
| 1 | widget |
| 2 | med widget |
| 3 | xl widget |
+---------+------------+Warning
Two items to note: a SHARD KEY cannot be added to a table after table creation, and you cannot use UPDATE on a SHARD KEY column.
Last modified: October 8, 2025