Deleting duplicate connections from the schema?
Hi all! I have a Java app working with a Cassandra-based JG, that checks the JG schema at startup and adds any missing elements via the
JanusGraphManagement interface. Due to a bug that existed for a long time that app created thousands of duplicate connections between same node and edge labels via management.addConnection().
This has become a problem, because these connections are getting cached in the StandardSchemaCache which has unlimited size and started taking up all the heap. I'm looking for a way to safely delete the duplicated connections from the schema without dropping the schema and without disrupting other instances of the app working with this graph.
Does anyone have experience with anything similar? I'm currently exploring the internals of JanusGraphManagement and ways to use the tx.query() interface to remove the unwanted relations, but I'd really appreciate any tips and ideas of an easier/safer solution.0 Replies