Replication replication

Adobe Experience Manager as a Cloud Service uses the Sling Content Distribution capability to move the content to replicate to a pipeline service run on Adobe Developer that is outside of the AEM runtime.

Methods of Publishing Content methods-of-publishing-content

Quick Un/Publish - Planned Un/Publish publish-unpublish

This feature lets you publish the selected pages immediately, without the additional options possible through the Manage Publication approach.

For more information, see Manage Publication.

On and Off Times - Trigger Configuration on-and-off-times-trigger-configuration

The additional possibilities of On Time and Off Time are available from the Basic tab of Page Properties.

To realize the automatic replication for this feature, enable Auto Replicate in the OSGi configuration On Off Trigger Configuration:

Manage Publication manage-publication

Manage Publication offers more options than Quick Publish, allowing for the inclusion of child pages, customization of the references, and starting any applicable workflows and offering the option to publish later.

Including a folder’s children for the “publish later” option invokes the Publish Content Tree workflow, described in this article.

You can find more detailed information on Manage Publication on the Publishing Fundamentals documentation.

Publish Content Tree Workflow publish-content-tree-workflow

You can trigger a tree replication by choosing Tools - Workflow - Models and copying the Publish Content Tree out-of-the-box workflow model, as shown below:

Do not invoke the original model. Instead, make sure to first copy the model and invoke that copy.

Like all workflows, it can also be invoked via API. For more information, see Interacting with Workflows Programmatically.

Alternatively, you can create a Workflow Model that uses the Publish Content Tree process step:

Parameters

Logging

When the tree activation workflow step starts, it logs its configuration parameters on the INFO loglevel. When paths are activated, an INFO statement is also logged.

A final INFO statement is logged after the workflow step has replicated all paths.

Also, you can increase the loglevel of the loggers below com.day.cq.wcm.workflow.process.impl to DEBUG/TRACE to get even more log information.

If there are errors, the workflow step terminates with a WorkflowException, which wraps the underlying Exception.

The following are examples of logs that are generated during a sample publish content tree workflow:

21.04.2021 19:14:55.566 [cm-p123-e456-aem-author-797aaaf-wkkqt] *INFO* [JobHandler: /var/workflow/instances/server60/2021-04-20/brian-tree-replication-test-2_1:/content/wknd/us/en/adventures] com.day.cq.wcm.workflow.process.impl.treeactivation.TreeActivationWorkflowProcess TreeActivation options: replicateAsParticipant=false(userid=workflow-process-service), agentId=publish, chunkSize=100, filter=, enableVersion=false

21.04.2021 19:14:58.541 [cm-p123-e456-aem-author-797aaaf-wkkqt] *INFO* [JobHandler: /var/workflow/instances/server60/2021-04-20/brian-tree-replication-test-2_1:/content/wknd/us/en/adventures] com.day.cq.wcm.workflow.process.impl.ChunkedReplicator closing chunkedReplication-VolatileWorkItem_node1_var_workflow_instances_server60_2021-04-20_brian-tree-replication-test-2_1, 17 paths replicated in 2971 ms

Resume Support

The workflow processes content in chunks, each of which represents a subset of the full content to be published. If the workflow is stopped by the system, it restarts and processes the chunk that was not yet processed. A log statement states that content was resumed from a specific path.

Replication API replication-api

You can publish content using the Replication API featured in AEM as a Cloud Service.

For more information, see the API Documentation.

Basic Usage of the API

@Reference
Replicator replicator;
@Reference
ReplicationStatusProvider replicationStatusProvider;

....
Session session = ...
// Activate a single page to all agents, which are active by default
replicator.replicate(session,ReplicationActionType.ACTIVATE,"/content/we-retail/en");
// Activate multiple pages (but try to limit it to approx 100 at max)
replicator.replicate(session,ReplicationActionType.ACTIVATE, new String[]{"/content/we-retail/en","/content/we-retail/de"});

// ways to get the replication status
Resource enResource = resourceResolver.getResource("/content/we-retail/en");
Resource deResource = resourceResolver.getResource("/content/we-retail/de");
ReplicationStatus enStatus = enResource.adaptTo(ReplicationStatus.class);
// if you need to get the status for more more than 1 resource at once, this approach is more performant
Map<String,ReplicationStatus> allStatus = replicationStatusProvider.getBatchReplicationStatus(enResource,deResource);

Replication with Specific Agents

When replicating resources, as in the example above, only the agents that are active by default are used. In AEM as a Cloud Service, it means only the agent called “publish”, which connects the author to the publish tier.

To support the preview functionality, a new agent called “preview” has been added, which is not active by default. This agent is used to connect the author to the preview tier. If you want to replicate only by way of the preview agent, you must explicitly select this preview agent by way of an AgentFilter.

See the following example:

private static final String PREVIEW_AGENT = "preview";

ReplicationStatus beforeStatus = enResource.adaptTo(ReplicationStatus.class); // beforeStatus.isActivated == false

ReplicationOptions options = new ReplicationOptions();
options.setFilter(new AgentFilter() {
  @Override
  public boolean isIncluded (Agent agent) {
    return agent.getId().equals(PREVIEW_AGENT);
  }
});
// will replicate only to preview
replicator.replicate(session,ReplicationActionType.ACTIVATE,"/content/we-retail/en", options);

ReplicationStatus afterStatus = enResource.adaptTo(ReplicationStatus.class); // afterStatus.isActivated == false
ReplicationStatus previewStatus = afterStatus.getStatusForAgent(PREVIEW_AGENT); // previewStatus.isActivated == true

In case you do not provide such a filter and only use the “publish” agent, the “preview” agent is not used and the replication action does not affect the preview tier.

The overall ReplicationStatus of a resource is only modified if the replication action includes at least one agent which is active by default. In the example above, this flow was not the case. The replication was just using the “preview” agent. Therefore, you must use the new getStatusForAgent() method, which allows querying the status for a specific agent. This method also works for the “publish” agent. It returns a non-null value if there has been any replication action done using the provided agent.

Methods of Invalidating Content invalidating-content

You can directly invalidate content by using either Sling Content Invalidation (SCD) from author (the preferred method) or by using the Replication API to invoke the publish Dispatcher flush replication agent. See Caching page for further details.

Replication API capacity limits

Replicate fewer than 100 paths at a time, with 500 being the limit. Above the limit, a ReplicationException is thrown.
If your application logic does not require atomic replication, this limit can be overcome by setting the ReplicationOptions.setUseAtomicCalls to false, which accepts any number of paths, but internally create buckets to stay below this limit.

The size of the content transmitted per replication call must not exceed 10 MB. This rule includes the nodes and properties, but not any binaries (workflow packages and content packages are considered binaries).

Troubleshooting troubleshooting

To troubleshoot replication, navigate to the Replication Queues in the AEM Author Service Web UI:

If the content couldn’t be published, the whole publication is reverted from the AEM Publish Service.

In that case, the main, editable queue shows a red status and should be reviewed to identify which items caused the cancelation of the publication. By clicking that queue, its pending items show up, from which a single item or all items can be cleared if needed.