The ETL service in CAST Imaging V3 does not start successfully, and no error message is displayed in the user interface.
When attempting to manually Start, Restart, or Resume the service, it transitions to a Paused state within a few seconds.
Because the ETL service is not running, data processing and updates within CAST Imaging gets halted. As a result, ongoing analysis activities are affected, and workflows that rely on ETL execution are delayed.
Action Plan
- Check if there is any ETL connection failure. This can happen if the neo4j database is quarantined,
-
To check if the neo4j database is quarantined, use the neo4j browser (http://localhost:7484) select the database as imaging and run the cypher query SHOW DATABASES to check if neo4j database is in quarantine.
In CAST Imaging V3, the ETL service must connect to the embedded Neo4j graph database. If Neo4j places the database in a quarantined stage then Neo4j does not fully mount the database, port may still appear open, ETL cannot access the graph, ETL then pauses automatically.
Check if any corrupted graph DB is the reason for Neo4j to quarantine the data base.
If the neo4j is in quarantined state then run the below in the system database
CALL dbms.quarantineDatabase('neo4j',false) -
Check if there is any schema mismatch or corrupted graph DB
Check if the Neo4j graph store version doesn’t match the ETL version often after upgrade, or the Neo4j database files are corrupted due to unclean shutdown or disk issue. This can also happen if transaction logs are inconsistent.
In the logs check for messages such as ,Store version mismatch, Database is quarantined, Recovery failed, Failed to start database
When ETL starts, it connects to Neo4j, it validates the expected graph schema/version, if mismatch or corruption is detected then connection fails. ETL cannot proceed and service transitions to Paused. The Windows service wrapper keeps running, but the Java ETL process fails initialization, so the service enters a Paused state rather than stopping entirely.
-
-
Check for any Port conflict. If another process uses the same port then Neo4j may fail to start properly Or ETL cannot connect.
This happens when ETL tries to open a Bolt connection, connection attempt fails that is connection gets refused or there is a timeout. ETL startup validation fails and service goes into the paused state.
This can happen if another Neo4j instance is installed in the same port or security software binding the port etc
-
Check if there is any permission issue. Check if the Windows service account running ETL or Neo4j does not have read/write access to Neo4j data folder or access to ETL logs, or access to temporary directories or access to CAST schema database.
When ETL tries to write temporary graph data or create indexes or access configuration, if access denied exception occurs, then initialization will fail and ETL service will then be moved to the paused status.
This may happen after a service account change, antivirus lockdown etc.
-
Check if there is a Java / memory problem. If there are log messages such as java.lang.OutOfMemoryError, then check if insufficient RAM is allocated to ETL JVM or Neo4j JVM.
During startup when ETL loads application metadata and begins graph preparation if heap memory fails then JVM throws OutOfMemoryError. ETL process then aborts and Windows service wrapper sets status to Paused.
NB: In CAST Imaging ETL is a Java service wrapped inside a Windows service controller. In CAST Imaging V3, the ETL Service works as a background Windows service responsible for, Extracting analysis data from the CAST Analysis schema, transforming it into the graph model used by Imaging, loading it into the Imaging Neo4j graph database . ETL stands for Extract, Transform and then Load
Resources
CAST Imaging v3 Product Documentation
CAST Imaging Troubleshooting Guides
Additional Resources
CAST Highlight Product Documentation
CAST Highlight Troubleshooting Guides
Ticket
56720
Comments