LLamaIndex and Neo4J

Neo4j Graph Platform Cypher
1

We uploaded chunks created from 25 (about 100 page) documents to perform RAG over a property graph. A total of about 9K nodes and 16K relationships. The below query is hung for over an hour in llamindex code when you instantiate a property graph store as well as in the Neo4J browser.

Any recommendations around what the issue might me or how do I speed it up?

neo4j_graph_store = Neo4jPropertyGraphStore(
    username="neo4j",
    password="xx",
    database="xx",
    url=URI)

> DEBUG:llama_index.graph_stores.neo4j.neo4j_property_graph:In coming query: CALL apoc.meta.graphSample() YIELD nodes, relationships RETURN nodes, [rel in relationships | {name:apoc.any.property(rel, 'type'), count: apoc.any.property(rel, 'count'
> )}] AS relationships
2

try set refresh_schema = false .
after debugging it seems to be the step that takes so much time ...

 graph_store = Neo4jPropertyGraphStore(
            username="neo4j",
            password=os.environ["NEO4J_PASSWORD"],
            url=os.environ["NEO4J_URL"],
            refresh_schema=False
        )