MySQL full unicode support
UTF-8 is a character encoding that most websites use. It encodes each of the 1,112,064 valid code points. To store all of this information, four bytes is required. The most popular values are in the three byte region. MySQL by default only uses a three byte encoding and so values in the four byte range (e.g. Asian characters and Emojis) can not be stored. Any attempt to enter a text that contains four byte characters will result in a Moodle database error.
MySQL does provide full four byte UTF-8 support, but it requires certain database settings to be configured. From version 3.3 on Moodle uses full UTF-8 for both MySQL and MariaDB by default. Existing databases will still run with partial support, but it is recommended to move over to full support.
Moodle comes with a Command Line Interface (CLI) script for converting to full UTF-8 for MySQL (and MariaDB). Before Moodle versions 3.1.5 and 3.2.2 this conversion tool would only change the Collation to some variant of 'utf8_bin'. 'utf8_unicode_ci' was the recommended Collation. We now recommend using 'utf8mb4_unicode_ci' which supports four byte characters (utf8_unicode_ci only supports three).
This script will attempt to change the database Collation, Character set, default table settings and column definitions.
- Fresh installs of Moodle 3.1.5 and 3.2.2 onwards will use utf8mb4 by default, if the database server is configured appropriately (see below).
- Sites upgrading to Moodle versions beyond 3.1.5 or 3.2.2 can use the script to update to utf8mb4. In Moodle 3.3 onwards a warning will show that the database isn't using full UTF-8 support and suggest moving to 'utf8mb4_unicode_ci'.
- You may choose to keep using 'utf8_*' unless you have MySQL 8.0.28 or later, or MariaDB 10.6.1 or later, in which case you must update to utf8mb4 (because utf8 was renamed to utf8mb3 in these versions, see MDL-74616).
To allow for large indexes on columns that are a varchar, a combination of settings needs to be set. The file format for the system needs to be using "Barracuda". This allows for the row format to be set to "Compressed" or "Dynamic". To enable this setting see the upgrade steps listed below. Moodle will not install if you have large format enabled without the Barracuda file format.
File per table
To enable this setting see the upgrade steps listed below.
This in conjunction with the row format being either "Compressed" or "Dynamic" allows for large varchar indexes above 191 characters. To enable this setting see the upgrade steps listed below.
Steps to upgrade
Most important: Please backup your database before making any changes or running the CLI script.
- Change configuration settings for MySQL (exactly the same for MariaDB). This step is optional. You can run the script and it will try and make these changes itself. If errors occur then try manually changing these settings as listed below.
- On Linux based systems you will want to alter my.cnf. This may be located in '/etc/mysql/'.
- Make the following alterations to my.cnf:
Important Change; InnoDB: The following InnoDB file format configuration options were deprecated in MySQL 5.7.7 and are now removed:
[client] default-character-set = utf8mb4 [mysqld] innodb_file_format = Barracuda innodb_file_per_table = 1 innodb_large_prefix = true character-set-server = utf8mb4 collation-server = utf8mb4_unicode_ci skip-character-set-client-handshake [mysql] default-character-set = utf8mb4
- Restart your MySQL server.
- Run the CLI script to convert to the new character set and collation (requires Moodle 3.1.5, 3.2.2 or newer):
$ php admin/cli/mysql_collation.php --collation=utf8mb4_unicode_ci
- Make sure to repair and optimize all databases and tables.
mysqlcheck -u root -p --auto-repair --optimize --all-databases
NOTE: On very large sites this may take a long time to run. You should probably establish how long on a test install before taking your live site offline. In some cases you might consider dumping and re-importing your data.
- Adjust the $CFG->dboptions array in your config.php to make sure that Moodle uses the right collation when connecting to the MySQL Server:
$CFG->dboptions = array( … 'dbcollation' => 'utf8mb4_unicode_ci', … );
If you only have access to the database command line (or something like phpMyaAdmin) you can try the following SQL commands:
SET GLOBAL innodb_file_format = barracuda; SET GLOBAL innodb_file_per_table = 1; SET GLOBAL innodb_large_prefix = 'on';
- Try adding some Emojis (e.g. 😂💩) to your Moodle site to verify that the upgrade was successful.