蜜豆视频

Configuring OSGi for 蜜豆视频 Experience Manager as a Cloud Service configuring-osgi-for-aem-as-a-cloud-service

is a fundamental element in the technology stack of 蜜豆视频 Experience Manager (AEM). It is used to control the composite bundles of AEM and its configurations.

OSGi provides the standardized primitives that allow applications to be constructed from small, reusable, and collaborative components. These components can be composed into an application and deployed. This allows easy management of OSGi bundles as they can be stopped, installed, started individually. The interdependencies are handled automatically. Each OSGi Component is contained in one of the various bundles. For more information, see the .

You can manage the configuration settings for OSGi components through configuration files that are part of an AEM code project.

OSGi Configuration Files osgi-configuration-files

Configuration changes are defined in the AEM Project鈥檚 code packages (ui.config) as configuration files (.cfg.json) under runmode specific config folders:

/apps/example/config.<runmode>

The format of OSGi configuration files is JSON-based using the .cfg.json format defined by the Apache Sling project.

OSGi configurations target OSGi components via their Persistent Identity (PID), which defaults to the OSGi component鈥檚 Java鈩� class name. For example, to provide OSGi configuration for an OSGi service implemented by:

com.example.workflow.impl.ApprovalWorkflow.java

an OSGi configuration file is defined at:

/apps/example/config/com.example.workflow.impl.ApprovalWorkflow.cfg.json

following the cfg.json OSGi configuration format.

Runmode Resolution runmode-resolution

Specific OSGi configurations can be targeted to specific AEM instances by using runmodes. To use runmode, create config folders under /apps/example (where example is your project name), in the format:

/apps/example/config.<author|publish>.<dev|stage|prod>/

Any OSGi configurations in such folders are used if the runmodes defined in the config folder name match the runmodes used by AEM.

For example, if AEM is using the runmodes author and dev, configuration nodes in /apps/example/config.author/ and /apps/example/config.author.dev/ are applied, while configuration nodes in /apps/example/config.publish/ and /apps/example/config.author.stage/ are not applied.

If multiple configurations for the same PID are applicable, the configuration with the highest number of matching run modes is applied.

This rule鈥檚 granularity is at a PID level. This means you cannot define some properties for the same PID in /apps/example/config.author/ and more specific ones in /apps/example/config.author.dev/ for the same PID. The configuration with the highest number of matching runmodes is effective for the entire PID.

When developing locally, a runmode startup parameter, -r, is used to specify the runmode OSGI configuration.

$ java -jar aem-sdk-quickstart-xxxx.x.xxx.xxxx-xxxx.jar -r publish,dev

Verifying runmodes

AEM as a Cloud Service runmodes are well defined based on the environment type and service. Review the complete list of available AEM as a Cloud Service runmodes.

OSGi configuration values specified by runmode can be verified by:

The resulting view displays all OSGi component configurations for the selected tier(s) with their applicable OSGi configuration values. These values can be cross-referenced with the OSGi configuration values in the AEM project鈥檚 source code under /apps/example/osgiconfig/config.<runmode(s)>.

To verify the appropriate OSGi configuration values are applied:

Types of OSGi Configuration Values types-of-osgi-configuration-values

There are three varieties of OSGi configuration values that can be used with 蜜豆视频 Experience Manager as a Cloud Service.

How to Choose the Appropriate OSGi Configuration Value Type how-to-choose-the-appropriate-osgi-configuration-value-type

The common case for OSGi uses inline OSGi configuration values. Environment-specific configurations are used only for specific use cases where a value differs between dev environments.

Environment-specific configurations extend the traditional, statically defined OSGi configurations that contain inline values, providing the ability to manage the OSGi configuration values externally via the Cloud Manager API. It is important to understand when the common and traditional approach of defining inline values and storing them in Git, should be used, versus abstracting the values into environment-specific configurations.

The following guidance addresses when to use non-secret and secret environment-specific configurations:

When to Use Inline Configuration Values when-to-use-inline-configuration-values

Inline configurations values are considered the standard approach, and should be used when possible. Inline configurations provide the benefits of:

Whenever defining an OSGi configuration value, start with inline values, and only select secret or environment-specific configurations if necessary for the use case.

When to Use Non-secret Environment-specific Configuration Values when-to-use-non-secret-environment-specific-configuration-values

Only use environment-specific configurations ($[env:ENV_VAR_NAME]) for non-secret configuration values when the values vary for the preview tier or vary across development environments. This includes local development instances and any 蜜豆视频 Experience Manager as a Cloud Service development environments. Other than for setting unique values for the preview tier, avoid using non-secret environment-specific configurations for 蜜豆视频 Experience Manager as a Cloud Service Stage or Production environments.

When to use secret environment-specific configuration values when-to-use-secret-environment-specific-configuration-values

蜜豆视频 Experience Manager as a Cloud Service requires the use of environment-specific configurations ($[secret:SECRET_VAR_NAME]) for any secret OSGi configuration values, such as passwords, private API keys, or any other values that cannot be stored in Git for security reasons.

Use secret environment-specific configurations to store the value for secrets on all 蜜豆视频 Experience Manager as a Cloud Service environments, including Stage and Production.

Creating OSGi Configurations creating-osgi-configurations

There are two ways create OSGi configurations, as described below. The former approach is typically used for configuring custom OSGi components which have well-known OSGi properties and values by the developer, and the latter for AEM-provided OSGi components.

Writing OSGi Configurations writing-osgi-configurations

JSON formatted OSGi configuration files can be written by hand directly in the AEM project. This is often the quickest way to create OSGi configurations for well-known OSGi components, and especially custom OSGi components that have been designed and developed by the same developer defining the configurations. This approach can also be used to copy/paste and update configurations for the same OSGi component across various runmode folders.

Generating OSGi Configurations using the AEM SDK Quickstart generating-osgi-configurations-using-the-aem-sdk-quickstart

The AEM SDK Quickstart Jar鈥檚 AEM Web Console can be used configure OSGi components, and export OSGi configurations as JSON. This is useful for configuring AEM-provided OSGi components whose OSGi properties and their value formats may not be well understood by the developer defining the OSGi configurations in the AEM project.

OSGi Configuration Property Formats osgi-configuration-property-formats

Inline Values inline-values

Inline values are formatted as standard name-value pairs, following standard JSON syntax. For example:

{
   "my_var1": "val",
   "my_var2": [ "abc", "def" ],
   "my_var3": 500
}

Environment-Specific Configuration Values environment-specific-configuration-values

OSGi configuration should assign a placeholder for the variable that is intended to be defined per environment:

use $[env:ENV_VAR_NAME]

Customers should only use this technique for OSGi configuration properties related to their custom code; it must not be used to override 蜜豆视频-defined OSGi configuration.

Secret Configuration Values secret-configuration-values

OSGi configuration should assign a placeholder for the secret that is intended to be defined per environment:

use $[secret:SECRET_VAR_NAME]

Variable Naming variable-naming

The following applies to both environment specific and secret configuration values.

Variables names must follow the following rules:

Values for the variables must not exceed 2048 characters.

Default Values default-values

The following applies to both environment specific and secret configuration values.

If no per-environment value is set, at runtime the placeholder is not replaced and is left in place since no interpolation happened. To avoid this, a default value can be provided as part of the placeholder with the following syntax:

$[env:ENV_VAR_NAME;default=<value>]

With a default value provided, the placeholder is replaced either with the per-environment value if provided or the provided default value.

Local Development local-development

The following applies to both environment specific and secret configuration values.

Variables can be defined in the local environment so they are picked up by the local AEM at runtime. For example, on Linux庐:

export ENV_VAR_NAME=my_value

It is recommended that a simple bash script is written which sets the environment variables used in the configurations and to execute it before starting AEM. Tools like help with simplifying this approach. Depending on the type of the values, they might be checked into source code management, if they can be shared between everyone.

The values for secrets are read from files. Therefore for each placeholder using a secret a text file containing the secret value must be created.

For example, if $[secret:server_password] is used, a text file named server_password must be created. All these secret files must be stored in the same directory and the framework property org.apache.felix.configadmin.plugin.interpolation.secretsdir must be configured with that local directory.

The org.apache.felix.configadmin.plugin.interpolation.secretsdir is an Sling framework property; so this property is not set in the felix console (/system/console?lang=en), but it is set in the sling.properties file that is used when the system boots. This file can be found in the /conf subdir of the extracted Jar/install folder (crx-quickstart/conf).

example: add this line to the end of the 鈥榗rx-quickstart/conf/sling.properties鈥�-file to configure 鈥榗rx-quickstart/secretsdir鈥� as secret folder:

org.apache.felix.configadmin.plugin.interpolation.secretsdir=${sling.home}/secretsdir

Author versus Publish Configuration author-vs-publish-configuration

If an OSGi property requires different values for author versus publish:

Configuration Examples configuration-examples

In the examples below, assume that there are three dev environments, in addition to the stage and prod environments.

Example 1

The intent is for the value of the OSGi property my_var1 to be the same for stage and prod, but differ for each of the three dev environments.

{
 "my_var1": "val",
 "my_var2": "abc",
 "my_var3": 500
}

{
 "my_var1" : "$[env:my_var1]"
 "my_var2": "abc",
 "my_var3": 500
}

Example 2

The intent is for the value of the OSGi property my_var1 to differ for stage, prod, and for each of the three dev environments. Thus the Cloud Manager API must be called to set the value for my_var1 for each dev env.

{
 "my_var1": "val1",
 "my_var2": "abc",
 "my_var3": 500
}

{
 "my_var1": "val2",
 "my_var2": "abc",
 "my_var3": 500
}

{
 "my_var1" : "$[env:my_var1]"
 "my_var2": "abc",
 "my_var3": 500
}

Example 3

The intent is for the value of the OSGi property my_var1 to be the same for stage, production, and just one of the dev environments, but for it to differ for the other two dev environments. In this case, the Cloud Manager API must be called to set the value of my_var1 for each of the dev environments, including for the dev environment which should have the same value as stage and production. It will not inherit the value set in the folder config.

{
 "my_var1": "val1",
 "my_var2": "abc",
 "my_var3": 500
}

{
 "my_var1" : "$[env:my_var1]"
 "my_var2": "abc",
 "my_var3": 500
}

Another way to accomplish this would be to set a default value for the replacement token in the config.dev folder such that it鈥檚 the same value as in the config folder.

{
 "my_var1": "val1",
 "my_var2": "abc",
 "my_var3": 500
}

{
 "my_var1": "$[env:my_var1;default=val1]"
 "my_var2": "abc",
 "my_var3": 500
}

Cloud Manager API Format for Setting Properties cloud-manager-api-format-for-setting-properties

See for information about the Cloud Manager API, and how it should be configured.

Setting Values via API setting-values-via-api

Calling the API deploys the new variables and values to a Cloud environment, similar to a typical customer code deployment pipeline. The author and publish services are restarted and reference the new values, typically taking a few minutes.

PATCH /program/{programId}/environment/{environmentId}/variables

[
        {
                "name" : "MY_VAR1",
                "value" : "plaintext value",
                "type" : "string"  <---default
        },
        {
                "name" : "MY_VAR2",
                "value" : "<secret value>",
                "type" : "secretString"
        }
]

Getting values via API getting-values-via-api

GET /program/{programId}/environment/{environmentId}/variables

See for more information.

Deleting values via API deleting-values-via-api

PATCH /program/{programId}/environment/{environmentId}/variables

To delete a variable, include it with an empty value.

See for more information.

Getting Values via the Command Line getting-values-via-cli

$ aio cloudmanager:list-environment-variables ENVIRONMENT_ID
Name     Type         Value
MY_VAR1  string       plaintext value
MY_VAR2  secretString ****

Setting values via the Command Line setting-values-via-cli

$ aio cloudmanager:set-environment-variables ENVIRONMENT_ID --variable MY_VAR1 "plaintext value" --secret MY_VAR2 "some secret value"

Deleting Values via the Command Line deleting-values-via-cli

$ aio cloudmanager:set-environment-variables ENVIRONMENT_ID --delete MY_VAR1 MY_VAR2

Number of Variables number-of-variables

Up to 200 variables per environment can be declared.

Deployment Considerations for Secret and Environment-Specific Configuration Values deployment-considerations-for-secret-and-environment-specific-configuration-values

Because the secret and environment-specific configuration values live outside of Git, and therefore, are not part of the formal 蜜豆视频 Experience Manager as a Cloud Service deployment mechanisms, the customer should manage, govern, and integrate into the 蜜豆视频 Experience Manager as a Cloud Service deployment process.

As mentioned above, calling the API deploys the new variables and values to Cloud environments, similar to a typical customer code deployment pipeline. The author and publish services are restarted and reference the new values, typically taking a few minutes. The quality gates and tests that are run by Cloud Manager during a regular code deployment are not performed during this process.

Typically, customers would call the API to set environment variables before deploying code that relies on them in Cloud Manager. In some situations, one might want to modify an existing variable after code has already been deployed.

There may be scenarios where a scheduled customer code deployment relies on existing variables to have new values, which would not be appropriate with the current code. If this is a concern, it is recommended to make variable modifications in an additive way. To do that, create new variable names instead of just changing the value of old variables so old code never references the new value. Then when the new customer release looks stable, one can choose to remove the older values.

Similarly, since a variable鈥檚 values are not versioned, a rollback of code could cause it to reference newer values that cause issues. The previously mentioned additive variable strategy would help here as well.

This additive variable strategy is also useful for disaster recovery scenarios where if code from several days prior needed to be redeployed, the variable names and values it references will still be intact. This relies on a strategy where a customer waits a few days before removing those older variables, otherwise the older code would not have appropriate variables to reference.