This library plugin defines a common abstraction to connect to relational database. By doing so, it serves three purposes:
- It allows database driver plugins (such as H2 Database Plugin) to be developed, improving the user experience for administrators.
- It simplifies other plugins that use RDBMS by eliminating the need to code up a configuration UI to let administrators select database.
This plugin is not meant to be used by end users by itself. It's supposed to be included through the dependencies of other plugins. If you are wondering how to store your job configuration etc. in a database, the answer is that you cannot.
The key class in this plugin is the
Database
class, which acts as a holder for JDBC DataSource instance. The
Database class is an extension point to be implemented by database
driver plugins.
The calling code can use this DataSource instance to obtain a
connection to the database.
This plugin adds a system configuration entry to let the administrator
configure the database used by Jenkins to store miscellaneous stuff. So
the easiest way for plugins to start storing data to the database is to
use this Database instance. This global database instance is kept in
the GlobalDatabaseConfiguration class. You can inject this via
@Inject, or you can call GlobalDatabaseConfiguration.get() to
retrieve it, and then use the getDatabase() method to obtain the
Database instance.
Because the database is shared by all the plugins, please use table names that include your plugin name as a prefix to avoid collisions.
In addition to the global database, this plugin also creates a database
local to
TopLevelItem.
These databases are normally backed by embedded database that stores
data under $JENKINS_HOME/jobs/NAME, and this simplifies the backup,
copying, deletion of the data that's local to jobs (such as test
reports, coverage data, and so on.)
This information is kept in the PerItemDatabase class, which you can
obtain by PerItemDatabaseConfiguration.findOrNull().
Sometimes it makes sense to store data to an entirely different database. Users may already have a database with data in it, in which case he'd want to just connect to that.
A plugin that wants to do this should define a field whose type is
Database (see
example).
In the config.{groovy,jelly}, use the f:dropdownDescriptorSelector tag
to allow the user to select a database (see
example).
This plugin exposes it through JPA 2.0
API (internally, it
uses Hibernate but please do not rely on this fact if you can as it may
change.) The entry point to the JPA support is the PersistenceService
class, and this exposes methods for obtaining EntityManagerFactory for
both the global database as well as arbitrary per-item database of your
choice.
Because there are several different databases, involved, @Entity
annotation alone is not sufficient. For persisted classes meant for the
global database, please put @GlobalTable in addition to @Entity.
The following code shows how to use this to persiste a new row:
public class Push {
@Inject
PersistenceService ps;
public void go(int n) throws IOException, SQLException {
Jenkins.getInstance().getInjector().injectMembers(this);
EntityManager em = ps.getGlobalEntityManagerFactory().createEntityManager();
em.getTransaction().begin();
TestRow row = new TestRow();
row.buildNumber = n;
row.x = "foo";
em.persist(row);
em.getTransaction().commit();
em.close();
}
}
@GlobalTable
@Entity
public class TestRow {
@Id
@Column
public int buildNumber;
@Column
public String x;
}MySQL Database plugin and PostgreSQL Database plugin are good examples of typical database driver plugins. For other "unusual" drivers that doesn't use the canonical host+database+username+password+properties combo, see H2 Database plugin source code as an example.
Refer to our contribution guidelines
Licensed under MIT, see LICENSE