Setting up your Project

Modifying Project Setup Details

In order to be built and deployed successfully with Cloud Manager, existing AEM projects need to adhere to some basic rules:

  • Projects must be built using Apache Maven.

  • There must be a pom.xml file in the root of the Git repository. This pom.xml file can refer to as many sub-modules (which in turn may have other sub-modules, etc.) as necessary.

  • You can add references to additional Maven artifact repositories in your pom.xml files. Access to password-protected artifact repositories is supported when configured. However, access to network-protected artifact repositories is not supported.

  • Deployable content packages are discovered by scanning for content package zip files which are contained in a directory named target. Any number of sub-modules may produce content packages.

  • Deployable Dispatcher artifacts are discovered by scanning for zip files (again, contained in a directory named target) which have directories named conf and conf.d.

  • If there is more than one content package, the ordering of package deployments is not guaranteed. Should a specific order be needed, content package dependencies can be used to define the order. Packages may be skipped from deployment.

Activating Maven Profiles in Cloud Manager

In some limited cases, you may need to vary your build process slightly when running inside Cloud Manager as opposed to when it runs on developer workstations. For these cases, Maven Profiles can be used to define how the build should be different in different environments, including Cloud Manager.

Activation of a Maven Profile inside the Cloud Manager build environment should be done by looking for CM_BUILD environment variable described above. Conversely, a profile intended to be used only outside of the Cloud Manager build environment should be done by looking for the absence of this variable.

For example, if you wanted to output a simple message only when the build is run inside Cloud Manager, you would do this:

        <profile>
            <id>cmBuild</id>
            <activation>
                  <property>
                        <name>env.CM_BUILD</name>
                  </property>
            </activation>
            <build>
                <plugins>
                    <plugin>
                        <artifactId>maven-antrun-plugin</artifactId>
                        <version>1.8</version>
                        <executions>
                            <execution>
                                <phase>initialize</phase>
                                <configuration>
                                    <target>
                                        <echo>I'm running inside Cloud Manager!</echo>
                                    </target>
                                </configuration>
                                <goals>
                                    <goal>run</goal>
                                </goals>
                            </execution>
                        </executions>
                    </plugin>
                </plugins>
            </build>
        </profile>
NOTE

To test this profile on a developer workstation, you can either enable it on the command line (with -PcmBuild) or in your Integrated Development Environment (IDE).

And if you wanted to output a simple message only when the build is run outside of Cloud Manager, you would do this:

        <profile>
            <id>notCMBuild</id>
            <activation>
                  <property>
                        <name>!env.CM_BUILD</name>
                  </property>
            </activation>
            <build>
                <plugins>
                    <plugin>
                        <artifactId>maven-antrun-plugin</artifactId>
                        <version>1.8</version>
                        <executions>
                            <execution>
                                <phase>initialize</phase>
                                <configuration>
                                    <target>
                                        <echo>I'm running outside Cloud Manager!</echo>
                                    </target>
                                </configuration>
                                <goals>
                                    <goal>run</goal>
                                </goals>
                            </execution>
                        </executions>
                    </plugin>
                </plugins>
            </build>
        </profile>

Password-Protected Maven Repository Support

NOTE

Artifacts from a password-protected Maven repository should only be used very cautiously as code deployed through this mechanism is currently not run through all of the quality rules implemented in Cloud Manager’s Quality Gates. Therefore it should only be used in rare cases and for code not tied to AEM. It is advised to also deploy the Java sources as well as the whole project source code alongside with the binary.

In order to use a password-protected Maven repository from Cloud Manager, specify the password (and optionally, the username) as a secret Pipeline Variable and then reference that secret inside a file named .cloudmanager/maven/settings.xml in the git repository. This file follows the Maven Settings File schema. When the Cloud Manager build process starts, the <servers> element in this file will be merged into the default settings.xml file provided by Cloud Manager. Server IDs starting with adobe and cloud-manager are considered reserved and should not be used by custom servers. Server IDs not matching one of these prefixes or the default ID central will never be mirrored by Cloud Manager. With this file in place, the server id would be referenced from inside a <repository> and/or <pluginRepository> element inside the pom.xml file. Generally, these <repository> and/or <pluginRepository> elements would be contained inside a Cloud Manager-specific profile, although that is not strictly necessary.

As an example, let’s say that the repository is at https://repository.myco.com/maven2, the username Cloud Manager should use is cloudmanager and the password is secretword.

First, set the password as a secret on the pipeline:

$ aio cloudmanager:set-pipeline-variables PIPELINEID --secret CUSTOM_MYCO_REPOSITORY_PASSWORD secretword

Then reference this from the .cloudmanager/maven/settings.xml file:

<?xml version="1.0" encoding="UTF-8"?>
<settings xmlns="http://maven.apache.org/SETTINGS/1.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
        xsi:schemaLocation="http://maven.apache.org/SETTINGS/1.0.0 http://maven.apache.org/xsd/settings-1.0.0.xsd">
    <servers>
        <server>
            <id>myco-repository</id>
            <username>cloudmanager</username>
            <password>${env.CUSTOM_MYCO_REPOSITORY_PASSWORD}</password>
        </server>
    </servers>
</settings>

And finally reference the server id inside the pom.xml file:

<profiles>
    <profile>
        <id>cmBuild</id>
        <activation>
                <property>
                    <name>env.CM_BUILD</name>
                </property>
        </activation>
        <build>
            <repositories>
                <repository>
                    <id>myco-repository</id>
                    <name>MyCo Releases</name>
                    <url>https://repository.myco.com/maven2</url>
                    <snapshots>
                        <enabled>false</enabled>
                    </snapshots>
                    <releases>
                        <enabled>true</enabled>
                    </releases>
                </repository>
            </repositories>
            <pluginRepositories>
                <pluginRepository>
                    <id>myco-repository</id>
                    <name>MyCo Releases</name>
                    <url>https://repository.myco.com/maven2</url>
                    <snapshots>
                        <enabled>false</enabled>
                    </snapshots>
                    <releases>
                        <enabled>true</enabled>
                    </releases>
                </pluginRepository>
            </pluginRepositories>
        </build>
    </profile>
</profiles>

Deploying Sources

It is a good practice to deploy the Java sources alongside with the binary to a Maven repository.

Configure the maven-source-plugin in your project:

        <plugin>
            <groupId>org.apache.maven.plugins</groupId>
            <artifactId>maven-source-plugin</artifactId>
            <executions>
                <execution>
                    <id>attach-sources</id>
                    <goals>
                        <goal>jar-no-fork</goal>
                    </goals>
                </execution>
            </executions>
        </plugin>

Deploying Project Sources

It is a good practice to deploy the whole project source alongside with the binary to a Maven repository - this allows as to rebuild the exact artifact.

Configure the maven-assembly-plugin in your project:

        <plugin>
            <groupId>org.apache.maven.plugins</groupId>
            <artifactId>maven-assembly-plugin</artifactId>
            <executions>
                <execution>
                    <id>project-assembly</id>
                    <phase>package</phase>
                    <goals>
                        <goal>single</goal>
                    </goals>
                    <configuration>
                        <descriptorRefs>
                            <descriptorRef>project</descriptorRef>
                        </descriptorRefs>
                    </configuration>
                </execution>
            </executions>
        </plugin>

Skipping Content Packages

In Cloud Manager, builds may produce any number of content packages.
For a variety of reasons, it may be desirable to product a content package but not deploy it. This may be useful, for example, when building content packages used only for testing or which will be repackaged by another step in the build process, that is, as a sub-package of another package.

To accommodate these scenarios, Cloud Manager will look for a property named cloudManagerTarget in the properties of built content packages. If this property is set to none, the package will be skipped and not deployed. The mechanism to set this property depends upon the way the build is producing the content package. For example, with the filevault-maven-plugin you would configure the plugin like this:

        <plugin>
            <groupId>org.apache.jackrabbit</groupId>
            <artifactId>filevault-package-maven-plugin</artifactId>
            <extensions>true</extensions>
            <configuration>
                <properties>
                    <cloudManagerTarget>none</cloudManagerTarget>
                </properties>
        <!-- other configuration -->
            </configuration>
        </plugin>

With the content-package-maven-plugin it is similar:

        <plugin>
            <groupId>com.day.jcr.vault</groupId>
            <artifactId>content-package-maven-plugin</artifactId>
            <extensions>true</extensions>
            <configuration>
                <properties>
                    <cloudManagerTarget>none</cloudManagerTarget>
                </properties>
        <!-- other configuration -->
            </configuration>
        </plugin>

Build Artifact Reuse

In many cases, the same code is deployed to multiple AEM environments. Where possible, Cloud Manager will avoid rebuilding the code base when it detects that the same git commit is used in multiple full-stack pipeline executions.

When an execution is started, the current HEAD commit for the branch pipeline is extracted. The commit hash is visible in the UI and via the API. When the build step completes successfully, the resulting artifacts are stored based on that commit hash and may be reused in subsequent pipeline executions.

Packages are reused across pipelines if they are in the same program. When looking for packages that can be reused, AEM disregards branches and reuses artifacts across branches.

When a reuse occurs, the build and code quality steps are effectively replaced with the results from the original execution. The log file for the build step will list the artifacts and the execution information which was used to build them originally.

The following is an example of such log output.

The following build artifacts were reused from the prior execution 4 of pipeline 1 which used commit f6ac5e6943ba8bce8804086241ba28bd94909aef:
build/aem-guides-wknd.all-2021.1216.1101633.0000884042.zip (content-package)
build/aem-guides-wknd.dispatcher.cloud-2021.1216.1101633.0000884042.zip (dispatcher-configuration)

The log of the code quality step will contain similar information.

Examples

Example 1

Consider that your program has two development pipelines:

  • Pipeline 1 on branch foo
  • Pipeline 2 on branch bar

Both branches are on the same commit ID.

  1. Running Pipeline 1 first will build the packages normally.
  2. Then running Pipeline 2 will reuse packages created by Pipeline 1.

Example 2

Consider that your program has two branches:

  • Branch foo
  • Branch bar

Both branches have the same commit ID.

  1. A development pipeline builds and executes foo.
  2. Subsequently a production pipeline builds and executes bar.

In this case, the artifact from foo will be reused for the production pipeline since the same commit hash was identified.

Opting Out

If desired, the reuse behavior can be disabled for specific pipelines by setting the pipeline variable CM_DISABLE_BUILD_REUSE to true. If this variable is set, the commit hash is still extracted and the resulting artifacts will be stored for later use, but any previously stored artifacts will not be reused. To understand this behavior, consider the following scenario.

  1. A new pipeline is created.
  2. The pipeline is executed (execution #1) and the current HEAD commit is becdddb. The execution is successful and the resulting artifacts are stored.
  3. The CM_DISABLE_BUILD_REUSE variable is set.
  4. The pipeline is re-executed without changing code. Although there are stored artifacts associated with becdddb, they are not reused due to the CM_DISABLE_BUILD_REUSE variable.
  5. The code is changed and the pipeline is executed. The HEAD commit is now f6ac5e6. The execution is successful and the resulting artifacts are stored.
  6. The CM_DISABLE_BUILD_REUSE variable is deleted.
  7. The pipeline is re-executed without changing the code. Since there are stored artifacts associated with f6ac5e6, those artifacts are reused.

Caveats

  • Build artifacts are not reused across different programs, regardless if the commit hash is identical.
  • Build artifacts are reused within the same program even if the branch and/or pipeline is different.
  • Maven version handling replace the project version only in production pipelines. Therefore if the same commit is used on both a development deploy execution and a production pipeline execution and the development deploy pipeline is executed first, the versions will be deployed to stage and production without being changed. However, a tag will still be created in this case.
  • If the retrieval of the stored artifacts is not successful, the build step will be executed as if no artifacts had been stored.
  • Pipeline variables other than CM_DISABLE_BUILD_REUSE are not considered when Cloud Manager decides to reuse previously created build artifacts.

Develop your Code Based on Best Practices

Adobe Engineering and Consulting teams have developed a comprehensive set of best practices for AEM developers.

On this page