The Bulk Ingester communicates with the Live Ingester through ActiveMQ by publishing events to a queue.
For each document in the repository, the Live Ingester consumes the corresponding event and uploads the document content directly to the Amazon S3 bucket used by Knowledge Discovery. Most common file formats, including .doc and .docx, are uploaded directly without requiring a PDF rendition.
Where transformation is required, the Live Ingester requests a PDF rendition from the Transform Service. The Transform Service generates the rendition, stores it in the Shared File Store (SFS), and notifies the Live Ingester when the rendition is available. Transformation is disabled by default and can be enabled for specific file types using the alfresco.transform.mime-type.mapping property.
The Live Ingester then performs two updates in Knowledge Discovery:
- Metadata update: Updates document metadata, such as the creator, description, and other repository properties.
- Content update: After the PDF rendition has been generated, uploads the rendition to the Amazon S3 bucket used by Knowledge Discovery.
As a result, both the document metadata and its searchable content remain synchronized between Alfresco Content Services and Knowledge Discovery.