Configuration management

Properties

Property files are used for configuration during the startup sequence before the repository is available. These properties configure the instance for a particular use (such as a development or production environment) and set various system directories. Once the startup sequence is completed and the repository is available, the majority of configuration is done in the config workspace in the repository.

Properties are not stored in a single property file but come from multiple sources. These sources are processed or parsed in this order:

<properties>
  <property>
    <name>defaultPublicURI</name>
    <value>redirect:/help.html</value>
  </property>
</properties>

Sources processed earlier may be overridden by sources processed later. The processing order and distribution of properties into several sources allows for flexible customization.

Some important properties set in the default magnolia.properties file:

# Repository configuration
magnolia.repositories.config = WEB-INF/config/default/repositories.xml
magnolia.repositories.home = $\{magnolia.app.rootdir\}/repositories
magnolia.repositories.jackrabbit.config = WEB-INF/config/repo-conf/jackrabbit-bundle-h2-search.xml

# Defining the instance as author instance
# Only used for the initial installation.
# Afterwards configuration in the config repository is used.
# The value is saved in /server/admin
magnolia.bootstrap.authorInstance=true

# Switch to false to enhance the performance of Javascript generation
magnolia.develop=false

# Set to true if bootstrapping/update should be performed automatically
magnolia.update.auto=false

 magnolia.app.rootdir is set programmatically during Magnolia start-up. Its value depends on the servlet container and on the environment, for instance on a production Tomcat it would be something like /some-path/some-tomcat/webapps.

Multiple configurations on a single Web application archive

Default properties are common regardless of the server name or webapp name. Webapp specific properties are installed only if the webapp name matches. Correspondingly, server specific properties are only installed if the server name matches. For example, Magnolia ships with two webapps by default: magnoliaAuthor and magnoliaPublic.

It is possible to set a different persistence manager per environment. For example, on the development instance you could use the default embedded Derby database, whilst on production you could use a production-scale persistent storage such as a MySQL database.

See WAR file with multiple configurations for more on this topic.

Customization

The structure and the order of source execution allows you to provide sophisticated configurations and flexible customization. The default Magnolia configuration (delivered in the web application bundle) provides an example of how the mechanism works:

• WEB-INF/config/default/magnolia.properties
• WEB-INF/config/magnoliaAuthor/magnolia.properties
• WEB-INF/config/magnoliaPublic/magnolia.properties

Magnolia provides a single web application which configures itself depending on the servlet context it is installed in. If your environment has multiple staging systems with differing configuration needs, you can apply different configurations by adding a server name to each of the listed paths.

For complex environments, the order of loading files can be defined in WEB-INF/web.xml as a context-param element:

<context-param>
   <param-name>magnolia.initialization.file</param-name>
      <param-value>
         WEB-INF/config/${servername}/${webapp}/magnolia.properties,
         WEB-INF/config/${servername}/magnolia.properties,
         WEB-INF/config/${webapp}/magnolia.properties,
         WEB-INF/config/default/magnolia.properties,
         WEB-INF/config/magnolia.properties
      </param-value>
</context-param>

Defining properties

Properties can be accessed through the info.magnolia.init.MagnoliaConfigurationProperties class.

In addition to these examples you can define arbitrary properties (magnolia.home is one example).

Extending configuration

You can extend a configuration by defining an extends property and setting its value to the source configuration you want it to inherit. The target configuration inherits everything from the source and adds its own exceptions. This can save time and effort as you only need to define exceptions explicitly. The mechanism is only available in the config repository.

In the example below, the sportstation site definition extends the travelsite definition. The definition inherits all configuration from the travel configuration and adds its own domains, internationalization and URI-to-repository mappings, theme and templates. The extends property can point to the source configuration with an absolute or relative path.

Overriding

Extending is additive by default, which means that configuration specified at the extending level is added to the inherited configuration. Setting the extends property to override changes this behavior. An override allows the extending node to completely remove the inherited content and replace it with its own content entries. In the example below, the travel site definition supports two locales, en (English) and de (German), whereas site travel-fr is targeted to French speakers only.

The French site extends travel. It inherits all configuration except the locales. An override under the locales node removes en and de. Only the French locale fr defined at this level is applied. This means that authors can enter only French content on the French site.

Observation

Observation is a feature of the Java Content Repository that enables applications to register interest in events that describe changes to a workspace. The applications can then monitor and respond to those events. The observation mechanism dispatches events when a persistent change is made to the workspace. Magnolia uses observation heavily. For instance, observation is used to reload module configurations and to reload all objects provided by the FactoryUtil. To use observation you must at least specify the workspace, the path to the node which should be observed, and an event listener. The event listener's onEvent() method is called whenever there are changes in the observed node.

Magnolia provides a helper class to assist you in using observation for your project. Use

Node2Bean

Most configuration is stored in the config workspace. To transfer the repository stored configuration into a Java object, a mechanism called Node2Bean is used. Node2Bean populates a Java Bean from the content of a repository node including sub nodes. Note that configuration details are not restricted to the config node. The following table shows where the Node2Bean mechanism is currently used.

For developers, module configuration (ModuleManagerlmpl) using the module class is the most important current usage of Content2Bean.

Node2Bean in module instances

Module configuration data is transferred into a Bean from /modules/<module name>/config. The Bean class to build is defined in the module descriptor XML file.

<!DOCTYPE module SYSTEM "module.dtd">
  <module>
    <name>samples</name>
    <displayName>Magnolia Samples Module </displayName>
    <class>
      info.magnolia.module.samples.SamplesModule
    </class>

Components. If a path in the config workspace is given rather than a concrete class name, then Node2Bean is used to build the component instance.

# Map to a path in the config workspace
info.magnolia.cms.i18n.I18nContentSupport = /server/i18n/content

Additional items such as components, templates and virtual URI mappings are configured at module level.

Data types

Node2Bean analyses the bean's "setter" and "adder" methods using introspection and uses them if a suitable configuration value is available. With "adder" methods (using the singular form of the node names) you can populate collections and maps. With this mechanism, Node2Bean can support all possible data types:

• Simple data types like String, int, long, float, double, boolean (to specify "true" you can use "true", TRUE", or "1") with the suitable "setter" method
• Other data types matching the "setter" method's signature
• Collections with String values or other data types by specifying a class property
• Maps with keys and values as Strings or other data types by specifying a class property

All sub elements are also built using Node2Bean.

Classes

The class used to instantiate an object through the Node2Bean mechanism is determined through reflection or by explicitly referencing a class in the class node data. By referencing a specific class you can override Magnolia default configuration and implement your own caching behavior, security mechanism and so on.

Numbered items:

1. config: Entry point of the transformation. In the module descriptor SampleConfig class is used. Set text and number properties.
2. sub: Subbean. The class is determined using reflection if it is not explicitly defined.
3. items: Collection. The corresponding add method is used to determine the class and populate the collection if existing.
4. item2: Special item with its own class and additional properties.
5. parameters: Collection of key-value pairs.

Simple data types

Values with simple data types must be defined as properties. Each property name must match its respective setter method.

Collections

To configure a collection you have to create a sub node and a suitable "setter" method:

• All properties of the collection node will create simple String entries in the collection. The properties' names are not used by Node2Bean, only the values.
• All sub nodes of the collection node will be treated as objects of the type specified in the class property in the sub node. If no class attribute is specified, a Map will be created instead. The sub nodes' names are not used by Node2Bean.
• All sub nodes of the collection node will be treated as maps unless they have a ".". The properties' names and the sub nodes' names are not used by Node2Bean.

Maps

The rules to populate a map are the same as with collections, except that the properties' names and the sub nodes' names are used as key values.

Example: cache configuration

The configuration creates an object of type 

 public void setCachePolicy(CachePolicy cachePolicy);
    public void setFlushPolicy(FlushPolicy flushPolicy);
    public void setBrowserCachePolicy(BrowserCachePolicy browserCachePolicy);
    public void setExecutors(Map executors);
    public void addExecutor(String name, CachePolicyExecutor executor);

Note that all necessary setters are available. For the executors node there are setter and adder methods. As the adder is more specific the setExecutors method will not be used.

This is what happens with setCachePolicy(CachePolicy cachePolicy).

Since there is a class property defined, let's look at the

public void setVoters(VoterSet voters);

Now let's have a look at the addExecutor (String name, CachePolicyExecutor executor) method in CacheConfiguration and the configuration:

As this method has two arguments, the node name bypass is passed as the first argument and a Bypass object as the second argument. Because CachePolicyExecutor is an interface, the implementing class is specified.

Voters

Voters are used in Magnolia whenever configuration values are not assigned at startup but instead depend on rules. For example the cache module has to determine if a requested resource may be cached or not. The rules to determine values should be configurable. The rules are user-defined using voters which evaluate established criteria by determining true or false of each rule. Voters are currently used for:

• Filter configuration: uses voters to determine whether a filter should be executed or bypassed.
• Cache configuration: uses voters to determine whether a file should be cached or not.

The basic concept of voters uses Voter classes which calculate an int vote value, where positive (1, 2, 3, ...) results are treated as "yes" or "true" and (0, -1, -2, ...) results are treated as "no" or "false". If you have a set of voters, then the result of a voting is the largest absolute result. If there are two voters with the same absolute result, then the one with the higher positive value will be taken.

Examples:

For most of the "real world" voters only boolean results make sense. These boolean voters return "1" for a "true" and "0" for a "false" result.

Voter examples

For further information, please see the