MySQL: Inserting Rows Only If They Don't Exist - Techniques and Considerations

  • InnoDB Locking: InnoDB uses row-level locking to ensure data consistency during concurrent access. When inserting a new row, it acquires locks to prevent other transactions from modifying the data.
  • Gap Locking: For tables with unique indexes, InnoDB also uses gap locking. This ensures that no rows can be inserted between existing values.
  • The Challenge: If two connections try to insert the same non-existent row concurrently, they might each lock the "gap" where the row would be inserted. This creates a deadlock, where neither connection can proceed.

Here are approaches to achieve "Insert if not exists" without deadlocks:

  1. SELECT ... FOR UPDATE: This approach involves:

    • Running a SELECT ... FOR UPDATE on the desired row (even if it doesn't exist).
    • Checking the results. If no row is found, you can insert.
    • However, SELECT ... FOR UPDATE on a non-existent row might not prevent another connection from attempting the same insert, leading to a deadlock later.
  2. Stored Procedure with Retry Logic: You can create a stored procedure that:

    • Checks for the existence of the row.
    • If absent, attempts to insert.
    • Implements retry logic in case of a deadlock error. This involves retrying the insert after a short delay.

Important Note: While option 1 seems straightforward, it has limitations and might not completely eliminate deadlocks. Option 2 with retry logic offers more control but requires additional code. Choosing the best approach depends on your specific needs and tolerance for deadlocks.

SELECT ... FOR UPDATE (Limited effectiveness):


SELECT * FROM your_table WHERE unique_column = 'your_value' FOR UPDATE;

  INSERT INTO your_table (column1, column2, ...) VALUES ('value1', 'value2', ...);



  • We start a transaction to ensure atomicity (all actions succeed or fail together).
  • We attempt to select the row with the desired unique_column value and lock it with FOR UPDATE. Note: This might not prevent another connection from attempting an insert before we check ROW_COUNT().
  • If no row is found (ROW_COUNT() = 0), we proceed with the insert.
  • We commit the transaction to make the changes permanent.

Stored Procedure with Retry Logic (More robust):


CREATE PROCEDURE insert_if_not_exists(
  IN value VARCHAR(255)
  DECLARE insert_successful BOOL DEFAULT FALSE;

  -- Retry loop with maximum attempts
  WHILE insert_successful = FALSE AND attempt < 3 DO

    SELECT * FROM your_table WHERE unique_column = value FOR UPDATE;

    IF (ROW_COUNT() = 0) THEN
      INSERT INTO your_table (column1, column2, ...) VALUES ('value1', 'value2', ...);
      SET insert_successful = TRUE;
    END IF;


    SET attempt = attempt + 1;


  IF insert_successful = FALSE THEN
    SIGNAL SQLSTATE '45000' SET MESSAGE_TEXT = 'Maximum retries exceeded for insert';
END //

  • This stored procedure takes a value to be inserted.
  • It uses a loop with a retry limit (attempt < 3) to handle potential deadlocks.
  • Inside the loop, it follows a similar logic as the previous example with SELECT ... FOR UPDATE and checking ROW_COUNT().
  • If the insert is successful (insert_successful becomes TRUE), the loop exits.
  • If retries are exhausted, it signals an error message.


  • These are examples, and you might need to adjust them based on your specific table structure and needs.
  • Option 2 offers more control but requires additional code.
  • Consider the trade-offs between these approaches when choosing the best solution for your scenario.


This is the recommended approach for most cases. It utilizes a single statement and avoids complex logic:

INSERT INTO your_table (column1, column2, ...) VALUES ('value1', 'value2', ...)
ON DUPLICATE KEY UPDATE column_to_update = 'update_value';
  • This statement attempts to insert the new row.
  • If a row with the same unique key combination already exists, it updates the specified column (column_to_update) with the provided value (update_value).
  • This approach avoids deadlocks as it doesn't involve separate checks and inserts.

Note: This method is most effective when you have a defined UNIQUE or PRIMARY KEY constraint on the columns that determine uniqueness.


This approach offers a simpler syntax but can be less desirable:

INSERT IGNORE INTO your_table (column1, column2, ...) VALUES ('value1', 'value2', ...);
  • However, unlike a regular insert, it ignores any errors that might occur during the insert, including duplicate key violations.
  • This can be useful if you don't care about updating existing rows and just want to insert if it doesn't exist.

Be cautious with INSERT IGNORE:

  • It ignores all errors, not just duplicate key violations. This can lead to unexpected behavior if there are other potential errors during the insert.
  • It doesn't provide any feedback on whether the insert was successful or not.
  1. Check Constraints:

While not directly related to inserts, you can define a CHECK constraint on your table to enforce data integrity and prevent duplicate inserts:

ALTER TABLE your_table
ADD CONSTRAINT check_unique UNIQUE (column1, column2, ...);
  • This constraint ensures that no two rows can have the same combination of values in the specified columns.
  • When attempting to insert a duplicate row, MySQL will throw an error, allowing you to handle the situation appropriately in your application code.

Remember, the best approach depends on your specific needs.

  • INSERT ... ON DUPLICATE KEY UPDATE is generally preferred for its efficiency and clarity.
  • INSERT IGNORE can be useful in specific scenarios where ignoring errors is acceptable.
  • Check constraints offer another layer of data validation but require handling errors in your application logic.

mysql sql innodb

Alternative Approaches to Grasping Identity Values in SQL Server 2005

Challenge: Unlike inserting a single row, directly retrieving multiple identity values generated in a single INSERT statement is not possible in SQL Server 2005...

Bitwise Operators Demystified: Flipping Bits with Confidence

Here are three common ways to flip a bit in SQL Server:Using the Bitwise NOT Operator (~):This operator inverts each bit in the operand...

MySQL to CSV: Beyond the Basics - Exploring Different Export Methods

CSV (Comma-Separated Values) is a file format where data is stored in plain text. Each line represents a record, and values within a record are separated by commas...

Optimizing Performance or Diving Deep? Demystifying Rails Raw SQL

Ruby on Rails (Rails) is a popular web framework that simplifies web development. It provides a powerful abstraction layer called ActiveRecord that interacts with your database using Ruby code...

Is MariaDB 5.5 Really Slower Than MySQL 5.1? Understanding Database Performance

Here's the breakdown:MySQL and MariaDB are both relational database management systems (RDBMS) used to store and manage data...

mysql sql innodb