Introduction

The configuration framework is designed to let the administrator and the developer decide from where to read a configuration. The purpose of the configuration framework is to have the ability to ask one manager for a property, but to read, if necessary, from multiple sources. Due to its extendable framework it can easily be customized.

The heart of the configuration are the configuration.xml files located in the core and configuration cartridge. These files include the configuration of the configuration engine instance used at ConfigurationMgr.

Code Examples

The following examples illustrate how to use the configuration framework. Every access is started at the ConfigurationMgr (which starts a configuration engine and loads the configuration.xml of each cartridge).

We recommend to extract the retrieval of configuration to context-specific configuration classes.

Providing a context-specific configuration interface and implementation has several advantages:

• better documentation capabilities on configuration “getter” methods

• better testability of business logic using different configuration values (mocking or providing different configurations to logic)

• centralized access to configuration framework (less dependencies in business logic to platform)

Access Cluster or Server Configuration

In previous versions it was helpful to have different configurations on different application servers (e.g., for JobServer). In K8s all application servers should behave the same. So @ClusterConfiguration fits best for global or application server configuration.

It is recommended to provide a default configuration. This will lead to consistent behavior of “not-configured” values. Otherwise, handling “null” values can be different at different implementation classes.

import com.intershop.beehive.core.capi.configuration.Configuration;
import com.intershop.beehive.core.capi.configuration.ServerConfiguration;
import com.intershop.beehive.core.capi.configuration.ClusterConfiguration;

public class BusinessConfigurationImpl implements BusinessConfiguration
{
    @Inject
    @ClusterConfiguration
    Configuration clusterConfig;
    
    @Inject
    @ServerConfiguration
    Configuration serverConfig;
    
    public boolean isFeatureEabled()
    {
        return serverConfig.getBoolean("config.key.boolean", true);
    }
    public boolean getGlobalMaxSize()
    {
        return clusterConfig.getInteger("config.key.integer", 42);
    }
}

Access Domain-Specific Configuration

Business configurations are mostly related to an organization, channel, or other context. Domain-specific configuration fits to most of these requirements.

// via domain configuration
public Integer getMaxSize(Domain domain)
{
    return domain.getConfiguration().getInteger("config.key.integer", 24);
}

Access via Static ConfigurationMgr

For the rare cases in which injection is not available, the ConfigurationMgr provides a static method getInstance() to provide full access the the configuration framework.

@Inject
ConfigurationMgr configMgr;
# or via static (avoid static class members for assignment)
ConfigurationMgr configMgr = ConfigurationMgr.getInstance();
// get a configuration for scope=instance
Configuration configInstance = configMgr.getInstanceConfiguration();

// get a configuration for scope=cluster
Configuration configCluster = configMgr.getClusterConfiguration();

// full definition of configuration context
Configuration yourConfig = configMgr.getConfiguration(String scope, ConfigurationContext context);

// configuration manager provides more capabilities to define the configuration context
String configDomainString = configMgr.getConfiguration(eTest).getString("config.key.string", "defaultValue");

Access via Object Path in Templates

The object path is often used in templates and queries. It is possible to access the configuration via the object path notation. The domain (#CurrentSite) has a method getConfiguration(), which is used for that example.

<isprint value="#CurrentSite:Configuration:String("<parameter>")#" />
<isprint value="#CurrentSite:Configuration:Integer(<group>.<subgroup>.<parameter-of-subgroup>)#" />

configuration.xml

Beside the configuration files of the core and configuration cartridge, all cartridges can define their own configuration.xml to extend the lookup and behavior of the configuration framework. The file has a predefined location at the cartridge resource folder /resources/${CartridgeName}/config/configuration.xml. Important declarations take place in the following cartridges:

• configuration: system scopes and finder

• orm: finder declaration to find/declare orm.properties

• core: main declarations of property files, readers and scopes

• bc_service: service-specific configuration

The file contains any:

• reader - can read a configuration file or resource from URI locations

• finder - providing a lookup in the system to find URIs (which can be read by the “readers”)

• writer - can write adapted configuration to a resource (URI) - is the opposite of reader

• scope - define which sets belongs together

• set - declares scope/finder/writer combination and represent the loaded properties

<configuration-setup>
  <readers>
    <reader class="..." />
  </readers>
  <finders>
    <finder name="..." class="..." />
  </finders>
  <writers>
    <writer name="..." class="..." />
  </writers>
  <scopes>
    <scope name="..." depends="..." />
  </scopes>
  <sets>
    <set finder="..." scope="..." writer="..." defaultWriter="..." writableKeys="..." />
  </sets>
</configuration-setup>

• The readers section defines the readers to use:

  • class: the class name of the reader; if no class is found, an exception will be thrown

• The finders section defines the finders to use:

  • name: the name by which this finder will be registered

  • class: the class name of the finder to use; if no class is found, an exception will be thrown

• The writers section defines the writers to use:

  • name: the name by which this writer will be registered

  • class: the class name of the writer to use; if no class is found, an exception will be thrown

• The scopes section defines the different scopes for which configurations will be created:

  • name: the name of the scope

  • depends: if a scope depends on the existence of other scopes

• The set section defines the configuration sets that will be found and read:

  • finder: the finder to use to find an URI a proper reader can read and provide a configuration set

  • scope: the comma-separated list of scopes for which this configuration set is intended

  • writer: the name of a writer to use if this set should also be written

  • defaultWriter: used if the writer of this set is the default writer

  • writableKeys: the comma-separated list of keys the writer can write

  • any other attribute of this tag will be treated as a bean property if the defined finder and a setter method with the given attribute name will be triggered; e.g., <set fileName="..."/> will trigger the setFileName(...) method

Scopes

The configuration engine can provide different configurations for different purposes. Therefore, the concept of scopes was introduced. A scope is a logical structuring of configurations that are needed for this purpose.

Stating that a scope depends on (an)other scope(s), means that in order to load configurations for this scope (e.g., instance), you initially have to load all configurations on which this scope depends (startup). Since startup uses the system configuration that provides you with all system properties (intershop.HomeDirectory), the instance scope cannot be loaded properly as long as it depends on values from the startup scope.

The fact that one scope depends on another scope does not automatically mean that if you request a configuration for this scope, you will also have configuration sets of the dependent scopes in it. That is, if you request a configuration for the scope instance which depends on startup, any startup configuration set is not necessarily in the instance scope.

The following picture outlines the currently used scopes:

System Properties

The system properties include all properties defined for the scope system. This includes:

• VM properties (command line argument -Dproperty=value)

• System environment variables

Those properties will be available inside the configuration framework using the following keys:

• Original key of the property (e.g., PATH)

• 'system.' plus the original key of the property (e.g., system.PATH)

As operation systems like Linux and Windows have restrictions with environment variables, the final property lookup via the configuration framework executes the following steps for the system scope:

1. Transform the key as follows:

   1. Make the key uppercase.

   2. Replace each dot ('.') by an underscore ('_').

   3. Replace each dash ('-') by an underscore ('_').
      Transformation example:
      intershop.jdbc.dataSourceName → INTERSHOP_JDBC_DATASOURCENAME

2. Try a lookup using the transformed key.

3. If this is a success, the value is returned.

4. If this is a failure, try another lookup using the original key.

This allows to override properties defined in a scope below the system scope using system environment properties that are restricted by the OS.

In rare cases the property key is required to follow a strict naming (e.g., camel case). So there is another option to add properties to the system scope: there are properties supported that follow the naming scheme ISH_ENV_* (* can be anything supported by the OS). The values of all these properties have to follow a special syntax:

key0=value0[,key1=value1[,...keyN=valueN]]

For instance there may be two ISH_ENV_* properties that may be defined like:

ISH_ENV_DB=intershop.databaseType=mssql,intershop.jdbc.dataSourceName=defaultDB,intershop.jdbc.driverType=thin
ISH_ENV_STATING=staging.process.NrOfParallelProcessors=2

If these properties are defined in this way, the following properties will be defined inside the configuration framework:

Finders

There is an implicit list of finders which provide URIs that can be read by the readers below.

Placeholders in Set Declaration

The configuration set will be configured with a finder and specific attribute values for this finder. Each attribute value can use the placeholder "${key}" to reference an existing property of an upper scope. For example, the sets for domain scope can use "cluster" scoped configuration values.

The standard configuration.xml contains placeholders for system-specific folders and environments. The way it works, e.g., for environments is shown below:

• The environment is defined in environment.properties (e.g., "environment=PRD"), which is defined in scope "instance" first.

• All "PRD"-specific configurations can be placed in PRD.properties if the configuration.xml contains:

  <set finder="property" scope="cluster,server,domain" fileName="${IS_CLUSTER_CONFIG}/system/config/cluster/${environment}.properties"/>

• The domain-specific configuration is more complex since a regular expression is used to check the matched files (e.g., PRD_uri.properties).

  <set finder="domain-folder" scope="domain" matches="^${environment}_[\w&&[^_]]*$" fileExtensions="properties,xml"/>

• In case domain properties should be environment and staging type dependent (staging.system.type mostly defined at staging.properties), an additional line in the configuration.xml will load such property files (e.g., PRD_editing_uri.properties).

  <set finder="domain-folder" scope="domain" matches="^${environment}_${staging.system.type}_[\w&&[^_]]*$" fileExtensions="properties,xml"/>

• Sets support an attribute priority that is used for sorting (default: 50). This example declares a resource based property file inside cartridge “demo” with a high priority.

  <set finder="resource" scope="cluster,server,domain" cartridge="demo" resourceName="config/cluster/demo.properties" priority="100"/>

Readers

This is the list of currently available readers (package names are abbreviated):

Writers

The following writers are currently used:

There should be at least one default writer in your system that will get any values set to the configuration framework - and does not lose them. Therefore, the transient writer has been created which updates the transient configuration set.

A writer will only be used to write keys it is allowed to write. To this end, the getWritableKeys() function is used to check whether a writer can write these keys. If no writer is found to write a key, the default writer will be used (if one is given).

environment.properties

The environment.properties file contains the default environment name “production”.

• The environment name should only contain word- and number characters, normally one of these: production, int, dev, development, uat, uat1, etc.

The deployment configuration will declare a specific value via the ENVIRONMENT environment variable.

Reloading

For development purposes, a reloading mechanism of property files has been implemented. It can be enabled by setting intershop.configuration.CheckSource in icm.properties to true. The delay for checking (and necessarily reloading) the property files can be set using intershop.configuration.CheckSourceInterval (which specifies the reloading interval in milliseconds).

Configuration Locations

Who should store configuration values and where?

Configuration of Developer

A developer wants to open up an implementation for configuration values. So the developer has to include a line to get the configuration from the framework.

String value = ConfigurationMgr.getString(key, defaultValue);

Providing a default value in here is the direct and best way to do it, because the code does not depend on a further configuration step.

A developer of an application type can override this defaultValue via a cartridge property file, located in the cartridge source directory resources/resources/<cartridge>/config. These configurations are included in the configuration finder "cartridge".

Configuration of Administrator

An administrator can declare configuration values via environment variables.

Configuration of Business User

Configuration values of business users should be stored in the database. For this purpose, two possibilities are available at the moment:

• Application preferences (for functionality and behavior) and

• Domain preferences (for data-related configuration)

The developer of the business management application is responsible for deciding on the correct location of the property.

The developers often define configuration values for different environments and domain. This can be reached by own configuration.xml declaration, which can provide domain- and environment-specific configuration values.

Service Configurations

There is another scope named service with finder, reader, and writer. This scope is used for the Managed Service Framework.

The finder creates URIs for the service scope which only the reader can read and for which only the writer is allowed to write.

The reader creates configuration sets which look at the ServiceConfigurationPOAttributeValuePO table using the ServiceConfigurationPO which has to be provided in the context used to get the configuration values from the set.

The writer, on the other hand, writes in the same table, also using the ServiceConfigurationPO provided in the context.

The service scope is somehow a standalone scope. If someone requests configurations from this scope, only one reader is triggered returning only one type of configuration set. Currently this scope does not contain other properties than properties from the ServiceConfigurationPOAttributeValuePO table. Values from the appserver.properties or PreferencePO table are not included, for instance. But of course, this can be changed any time by creating an own configuration.xml file.