Why SQL_MODE is important

Today was another example of where a correct SQL_MODE saved customer data from being corrupted. By default, MySQL does not enforce data integrity. It allows what is called silent truncations where the result of what you INSERT or UPDATE does not represent truth. NOTE: I see very few customers ever have this correctly configured, those that do have actually listened to my advice.

If you do not read any further, your production MySQL environments should be running with at the bare minimum of SQL_MODE=STRICT_ALL_TABLES however I would also advocate for additional SQL_MODE settings.

For this example, some modified undesirable code attempted to reduce a counter by 1, however because of an UNSIGNED data type and a correctly set SQL_MODE, the application produced an error and data was not corrupted.

This is what should happen with your SQL.

mysql> update stats set loss_count=loss_count - 1 where user_id=42;
ERROR 1690 (22003): BIGINT UNSIGNED value is out of range in '(`db`.`stats`.`loss_count` - 1)'

It is interesting to note that the error message actually is misleading. The datatype for the column is SMALLINT however the error message prompted an unnecessary schema verification. Even with the calculation I would have assumed the inherited data type would be the column definition before the subtraction. If you try to set the value automatically to negative you get a better message.

mysql> update stats set loss_count=-1 where user_id=42;
ERROR 1264 (22003): Out of range value for column 'loss_count' at row 1

So why is this a big deal? If your MySQL instance is not using SQL_MODE then the result would have been a value of 0 and this would never highlight the bug in the code.

I should point out that only Drizzle of the MySQL forks and variants has tightened up this data integrity, however I have to agree to disagree with the decision to drop UNSIGNED. While I under internally the code simplification behind the decision, without check constraints in MySQL, UNSIGNED is a saving grace. In this example, Drizzle in this case would not have reported an error.

Tagged with: Databases Drizzle MySQL

Related Posts

Why Being Proactive Is Always a Winning Approach

Many companies manage production infrastructure using a reactive model rather than a proactive one. Organizations typically react to warnings and alerts, then implement corrective actions in response. While some companies have well-designed architectural patterns—such as feature flags and rate limiting—that can quickly mitigate the impact of issues, these are merely temporary solutions, not resolutions.

Read more

AWS CLI support for Aurora DSQL and S3 Tables

If you were following the AWS Re:invent keynote yesterday there were several data specific announcements including Aurora DSQL and S3 Tables . Wanting to check them out, I downloaded the latest AWS CLI 2.

Read more

Migrating off of WordPress - A Simplified Stack

The ongoing drama between Wordpress v WP Engine continues to cross my reading list, but I have permanently removed WordPress from my website. I have finally transitioned away from the complex Linux/Apache/MySQL/PHP (LAMP) stack required for self-hosting WordPress on my professional website.

Read more