Issue
After a Fusion upgrade, JDBC connector crawls may ingest fewer documents than expected. The crawl completes without failure, but the total document count is lower than previous known baselines.
This behavior often affects connectors using index pipelines with stages that generate multi-valued fields, such as "Split Keywords," where the destination field is defined as single-valued in the Solr schema.
Diagnosis
To confirm this issue:
Review the document count shown in the job history view for the connector.
Compare against historical expected counts or non-production environments.
Navigate to the collection schema in the Solr configuration UI and inspect the definition of the field used by the pipeline stage.
Look for a definition like the following:
<field indexed="true" multiValued="false" name="form_type_code" required="false" stored="true" type="text_en"/>A
multiValued="false"setting will cause indexing errors when multiple values are sent for this field.In the job history or logs, you may see indicators that documents were dropped, or a mismatch between processed and indexed document totals.
Environment
Fusion 5.9.x (applicable to any version where schema changes are not automatically carried over during upgrades)
Cause
The root cause is a mismatch between the output of an index pipeline stage (e.g., a "Split Keywords" processor that outputs an array of values) and the schema definition of the target field, which is marked as single-valued (multiValued="false"). When the pipeline attempts to write multiple values into a single-valued field, documents are silently dropped or indexing errors may occur, depending on the configuration.
This issue typically arises during or after upgrades where schema changes made in lower environments were not promoted to production, especially if schema promotion was manual or handled outside the upgrade automation process.
Resolution
Update the Solr schema definition for the affected field to support multiple values. In the Fusion UI or by direct schema API update, change the field definition from:
<field indexed="true" multiValued="false" name="form_type_code" required="false" stored="true" type="text_en"/>To:
<field indexed="true" multiValued="true" name="form_type_code" required="false" stored="true" type="text_en"/>After making the change:
Re-index the affected data sources.
Confirm that the expected number of documents are now ingested.
If needed, validate by comparing the new ingestion count with historical norms or known good environments.
Additional guidance
Always validate whether custom schema changes made in staging or development have been promoted to production during Fusion upgrades.
Use collection schema diffing tools or config-sync repositories to ensure parity between environments.
For environments using GitOps or Helm-based schema promotion, verify pull request history or deployment logs to confirm schema changes were included.
If ingestion issues persist after correcting the schema, additional root causes may exist and further investigation is required.