Skip to main content
Benjamin C. (Entrust) a posé une question dans * Data 360 *

I am using the Salesforce content data kit to ingest ContentDocument, ContentDocumentLink, and ContentDocumentVersion data for a Data 360 Search Index. 

 

Salesforce Engineering confirmed that an initial or full index build processes all ContentDocumentVersion records included through the configured DMO relationship, not just each document’s latest published version. This means historical versions that remain in the DMO can be processed again during a new index build or rebuild, increasing Unstructured Data Processing credit consumption. 

 

I have not found a supported declarative way to limit the ContentDocumentVersion DMO or Search Index to records matching the document’s LatestPublishedVersionId. Salesforce Support is investigating whether a Data Space Filter could help, but with the way I understand Data Space Filter that doesn't seem like a viable option (crossing my fingers).  

 

Has anyone implemented a supported solution that ensures only the latest published ContentVersion for each ContentDocument is available to a Search Index? For example, have you successfully used a Data Space Filter, Data Transform, custom DMO, or a different data-stream design without breaking the standard file relationships required for indexing AND only indexes the latest ContentVersion data?

3 réponses
  1. 23 oct., 10:30

    Hi @Benjamin C.

    , 

     

    I agree with Santosh that Data Space Filters are a weak fit, but one angle is worth testing before you rule them out.  

     

    ContentVersion has a standard IsLatest field. If your content kit stream brings it in, try a dedicated data space used only for the Search Index, with a filter on IsLatest = true on the ContentVersion DMO. Because the filter evaluates current DMO values, superseded versions should drop out when they flip to false on the next sync. I’d avoid an ingestion-time stream filter for this. An update that flips a record to IsLatest = false may never land, so the old version could stay in the DMO as stale data.  

     

    I’d stay away from a Data Transform into a custom DMO. The index depends on the standard file relationships, and rebuilding those is where things usually break. 

0/9000