Table names in MySQL are case sensitive. This behavior can be adjusted as seen here but it is best to assume that the DB will be case sensitive for table names. Consequently, make sure your @Table annotation matches your Liquibase createTable element.
Prior to version 5.6.4, MySQL did not store fractions of a second in temporal data types. (See here). To be cross-database compatible then, we can not rely on straight comparisons of a new Date() to time values read from the database. Instead we must make the time values coarser by rounding down to the nearest second. You can use the following:
```
long now = new Date().getTime() / 1000;
long someDBtime = foo.getTime() / 1000;
if (now == someDBtime) {
...
```
Additionally, since MySQL doesn’t store fractions of a second, it may be necessary to insert a “sleep 1” in any spec test that is testing that something happened after something else. Since the tests run so rapidly, it is possible that MySQL will record them as occurring in the same secord.
String searches in MySQL are case insensitive. See here.
The Candlepin table and column definitions expect UTF8 collations. While they are explicitly set in most cases now, older Liquibase changeset files may not specify the collation, so it’s a good idea to ensure that the default collation won’t cause problems when creating new tables, columns and foreign keys.
Some of the current locking mechanisms within Candlepin expects the behavior of READ-COMMITTED transaction isolation. Most things may work properly without changing this, but certain spec tests may fail indeterministically without it.
We are using DATETIME to store date information in MySQL. In MySQL, DATETIME does not store any time zone information. This is consistent with our Postgres and Oracle schemas which use TIMESTAMP WITHOUT TIME ZONE. If we ever decide to store time zones, we have a couple of options.
- Add a column to store the time zone.
- Switch to TIMESTAMP which stores the value in UTC. Note that the TIMESTAMP type can only store values up to the year 2038. Additionally, MySQL has an annoying “feature” where the first TIMESTAMP column in a table is set to an automatic default and to be updated with the current time every time the record changes. See here.

Running Candlepin on MySQL

Getting Ready

Install MySQL (or MariaDB which is already in Fedora repos). For MariaDB the packages are named mariadb-server and mysql-connector-java.
```
$ sudo yum install mysql-server mysql-connector-java
```

Create /etc/my.cnf.d/isolation.cnf and open it in an editor. Add the following to set the default character collation and transaction isolation mode:

[mysqld]
# Required for Candlepin
collation-server=utf8_general_ci
character-set-server=utf8

# Required for Candlepin
transaction-isolation=READ-COMMITTED

Enable MySQL/MariaDB with systemd. For MariaDB use mariadb.service.

$ sudo systemctl enable mysqld.service
$ sudo systemctl start mysqld.service

Create the Candlepin user.

$ mysql --user=root mysql --execute="CREATE USER 'candlepin'@'localhost';   GRANT ALL PRIVILEGES on candlepin.* TO 'candlepin'@'localhost' WITH GRANT OPTION"

Create the Candlepin database

$ mysqladmin --user="candlepin" create candlepin

Deploying

Add the necessary configuration to /etc/candlepin/candlepin.conf. Make sure you comment out any options that might conflict with the options below!

jpa.config.hibernate.connection.driver_class=com.mysql.jdbc.Driver
jpa.config.hibernate.connection.url=jdbc:mysql:///candlepin
jpa.config.hibernate.dialect=org.hibernate.dialect.MySQL5InnoDBDialect
jpa.config.hibernate.connection.username=candlepin
jpa.config.hibernate.connection.password=

org.quartz.jobStore.driverDelegateClass = org.quartz.impl.jdbcjobstore.StdJDBCDelegate
org.quartz.dataSource.myDS.driver = com.mysql.jdbc.Driver
org.quartz.dataSource.myDS.URL = jdbc:mysql:///candlepin
org.quartz.dataSource.myDS.user = candlepin
org.quartz.dataSource.myDS.password =
org.quartz.dataSource.myDS.maxConnections = 5

Deploy to MySQL.
```
$ bin/deploy -m
```

Last modified on 25 November 2024

Table of Contents

MySQL Tips for Developers

Running Candlepin on MySQL

Getting Ready

Deploying