MySQL VARCHAR(255) UTF8 is too long for key, but max length is 1000 bytes

Question

I know there have been plenty of questions about this, but I think my math is right.

• MySQL reserves 3 bytes per UTF8 character.
• MyISAM allows keys of leng

Answer 1

Anyone who does need a larger key length should look at innodb_large_prefix

visit http://dev.mysql.com/doc/refman/5.5/en/innodb-parameters.html#sysvar_innodb_large_prefix

Answer 2

If you're using utf8mb4, and you have unique indexes on varchar columns that are greater than 191 characters in length, you'll need to turn on innodb_large_prefix to allow for larger columns in indexes, because utf8mb4 requires more storage space than utf8 or latin1. Add the following to your my.cnf file.

[mysqld]
innodb_file_format=barracuda
innodb_file_per_table=1
innodb_large_prefix=1
init_connect='SET collation_connection = utf8mb4_unicode_ci'
init_connect='SET NAMES utf8mb4'
character-set-server=utf8mb4
collation-server=utf8mb4_unicode_ci

---

More info about the why and future from MySQL 5.7 documentation:

If innodb_large_prefix is enabled (the default in MySQL 5.7.7), the index key prefix limit is 3072 bytes for InnoDB tables that use DYNAMIC or COMPRESSED row format. If innodb_large_prefix is disabled, the index key prefix limit is 767 bytes for tables of any row format.

innodb_large_prefix is deprecated in MySQL 5.7.7 and will be removed in a future release. innodb_large_prefix was introduced in MySQL 5.5 to disable large index key prefixes for compatibility with earlier versions of InnoDB that do not support large index key prefixes.

To sum up, the limit is only there for compatibility and will be increased in future versions.

Answer 3

MySQL reserves the max amount for a UTF8 field which is 4 bytes, so that is why you are blowing past the 1000 byte limit. My recommendation is to create the varchar at less than 255 or create it without UTF8.

Both of those solutions are probably not right for you or you would have already tried that.

The only other solution I can think of is to split the column into 2 small columns and create an unique index on both of those fields, but I believe that you would get the same error as above.

Since you probably need UTF8, I would seriously look at reducing the varchar(255) column down a little to 250 (or 249) to make this work.