Skip to main content

Understanding Tag Propagation Automation Configuration

Angelo Bresci
Updated

Area: Product Issues

Sub-Area: Tag Propagation

Issue

Tag propagation automation is enabled but tags are not flowing to expected downstream datasets or views. This commonly occurs when the automation configuration has incorrect settings for downstream propagation or when users expect column-level tag propagation from the standard tag propagation automation.

You Might Be Asking

  • Why aren't my tags propagating to downstream datasets even though automation is turned on?
  • Does tag propagation work at the column level?
  • How do I trigger tag propagation for existing tags?
  • What's the difference between tag propagation and other propagation automations?

Solution

  1. Verify Tag Propagation Automation Configuration:
    • Navigate to Govern → Automations → Tag Propagation
    • Check that include_downstreams is set to true if you want tags to flow to downstream assets
    • Verify any tag prefix filters are configured correctly for your use case
  2. Understand Tag Propagation Scope:
    • Tag propagation only works at the dataset level, not column level
    • For column-level tag propagation, use Glossary Term Propagation automation instead
    • Propagation is strictly downstream only - tags flow from source to dependent assets
    • Maximum propagation depth is 5 hops downstream
  3. Verify Lineage Requirements:
    • Confirm upstream lineage exists between tagged assets and target datasets
    • Check the lineage graph in DataHub UI: Asset → Lineage tab
    • Ensure lineage was captured during ingestion (SQL parser, view DDL, etc.)
  4. Trigger Propagation for Existing Tags:
    • Tag propagation is event-driven - only fires on new tag changes
    • To propagate existing tags: remove and re-add the tag on the upstream asset
    • Alternatively, use the automation's backfill/bootstrap option if available
  5. Check Common Configuration Issues: 
    // Verify automation settings via GraphQL
query {
  automation(urn: "urn:li:automation:your-tag-propagation-automation") {
    definition {
      ... on TagPropagationDefinition {
        includeDownstreams
        includeUpstreams
        tagFilters {
          prefixes
        }
      }
    }
  }
}

Additional Notes

Tag propagation configuration changes require the automation to be event-triggered again. Simply having the automation "turned on" doesn't guarantee proper configuration - the specific settings like include_downstreams determine propagation behavior. Column-level classification should use Glossary Term Propagation rather than Tag Propagation for proper inheritance.

Related Documentation

Tags: tag-propagation, automation, configuration, lineage, downstream, column-level, governance, dataset-tags, event-driven, troubleshooting