Goal
Ensure a smooth transition when upgrading from Fusion 5.9.4 to 5.9.14 by addressing key compatibility changes related to job history status casing, schema file management, and dynamic field name behavior.
Environment
Fusion 5.9.4 to 5.9.14
Self-hosted on Amazon EKS
Solr 9.6.1
Guide
Job scheduler status values are now lowercase
In Fusion 5.9.14 and later, job status values stored in the system_job_history collection are lowercase. In earlier versions such as 5.9.4, values may appear in uppercase, depending on how pipelines or jobs were implemented. This difference can impact logic that relies on case-sensitive status checks (e.g., for collection swaps or downstream automation).
Recommendation:
Normalize all job status values using a consistent case (e.g., convert to lowercase in your logic).
If backward compatibility is required, convert based on version checks or explicitly cast the values to uppercase or lowercase as needed.
This change appears to have been introduced after version 5.9.9.
Handle dual schema files in configsets
After upgrading to Fusion 5.9.14, you may observe both managed-schema and managed-schema.xml files within your Solr configsets. Fusion uses managed-schema as the authoritative schema file if both exist. However, having both can cause confusion or schema API inconsistencies in some environments.
Recommendation:
Retain only one schema file per collection. Prefer keeping
managed-schema, which aligns with current Solr usage.
To remove the unnecessary schema file using Zookeeper CLI:
# Step into your Zookeeper pod
kubectl exec -it <zookeeper-pod-name> -- bash
# Navigate to the ZooKeeper bin directory
cd /apache-zookeeper-*/bin
# Launch the zkCli shell
./zkCli.sh
# List configs
ls /configs/<collection-name>
# If both files exist:
[_fusion_owner.txt, ..., managed-schema, managed-schema.xml, ...]
# Delete the unwanted file
delete /configs/<collection-name>/managed-schema.xml
# Re-verify
ls /configs/<collection-name>Note: Perform this in a lower environment first and validate indexing and query pipeline behavior before replicating in production.
There is no bulk-delete command available; each collection must be reviewed and updated individually.
Additional notes
Fusion does not currently support preventing creation of
managed-schema.xmlby default. However, if both schema files exist, Fusion will prioritizemanaged-schema.When creating collections from configsets, include only the desired schema file to avoid post-deployment cleanup.