How to perform a full data resynchronization for Databricks Delta Share
Use this guide after making configuration changes to a Databricks datashare adapter (e.g., enabling activity logging or changing ovOnly) to ensure all existing records are reflected with the updated settings.
OVERVIEW
Configuration changes to a Databricks datashare adapter (such as toggling ovOnly=false or enabling activity logging) apply only to new events going forward. To backfill existing records under the updated configuration, a full sequential resynchronization must be performed across all data types in a specific order.
RESYNCHRONIZATION ORDER
Each data type must complete fully before the next step is triggered. Running steps out of order can cause cast exceptions or data inconsistencies in Databricks.
| Step | Data Type & Details |
|---|---|
| 1 | Entities Trigger the entity sync first. This is the foundation — all subsequent data types depend on entities being fully synchronized. → No dependency — start here |
| 2 | Relations Sync relationship records between entities. Must not begin until entity sync has completed successfully. → Begin after Entities complete |
| 3 | Interactions Sync Interaction data. Depends on Relations being fully synchronized to avoid cast exceptions or data inconsistencies. → Begin after Relations complete |
| 4 | Matches Sync Match data. Depends on Relations being fully synchronized to avoid cast exceptions or data inconsistencies. → Begins after Relations complete |
| 5 | Merges Sync merge records. These reference match data and must not run before Matches have completed → Begins after Matches complete |
| 6 | Activities The final sync step. Activity records are synced last, after all structural data types are in place. → Final step |
Important notes
— Each data type must complete successfully before the next one is triggered. Running them out of order can cause cast exceptions or data inconsistencies.
— Records may take up to 5 minutes to appear in the Databricks share after each step completes.
— This resync applies to both ovOnly=true and ovOnly=false configurations. If ovOnly was recently changed, the resync is required to reflect all-values data for existing records.
— New events generated after the configuration change will flow automatically — only existing records require resynchronization.
TRIGGERING EACH SYNC STEP VIA API
Use the following API calls to trigger each resynchronization step. Authenticate with a valid Reltio Bearer token. Confirm the previous step's sync has completed before issuing the next call.
Authorization: Bearer <your_reltio_token> (required on all requests)
Step 1: Entities
POST to the entities resync endpoint. Replace {tenant_id}, {env}, with your environment values.
POST https://{env}.reltio.com/reltio/api/{tenantId}/syncToDataPipeline?dataTypes=entitiesStep 2: Relations
After entities sync confirms success, trigger relations resync.
POST https://{env}.reltio.com/reltio/api/{tenantId}/syncToDataPipeline?dataTypes=relationsStep 3: Interactions
After relations sync confirms success, trigger interactions resync.
POST https://{env}.reltio.com/reltio/api/{tenantId}/syncToDataPipeline?dataTypes=interactionsStep 4: Matches
After relations sync confirms success, trigger matches resync.
POST https://{env}.reltio.com/reltio/api/{tenantId}/syncToDataPipeline?dataTypes=matchesStep 5: Merges
After matches sync confirms success, trigger merges resync.
POST https://{env}.reltio.com/reltio/api/{tenantId}/syncToDataPipeline?dataTypes=mergesStep 6: Activities
Final step. After links sync confirms success, trigger activities resync.
POST https://{env}.reltio.com/reltio/api/{tenantId}/syncToDataPipeline?dataTypes=activities
Validation after resync
After all six steps complete, verify the following in Databricks:
✓ Each table should have records that are visible and reflect the updated configuration.
✓ If ovOnly was changed to false, confirm that non-OV (all-values) attributes are now present on records.
✓ Activity events are flowing and visible in the share.
✓ If any records appear missing, allow up to 5 minutes for propagation before re-investigating.
Product area: Analytics Pipelines · Databricks Applies to: Data Pipeline Hub (DPH) Last updated: July 2026
Comments
Please sign in to leave a comment.