Neo4j: GC Monitor: Application threads blocked for

Question

I have a Vertex node entity which has a List of relationships.

@NodeEntity (label="User")
public class Vertex  {
    private Long id;
    private String name;
    @Index(unique=true)
    private String email;
    @Relationship(type="WORKS_WITH", direction = Relationship.OUTGOING)
    private  List<Edge> teammates;
..
}

@RelationshipEntity(type = "WORKS_WITH")
public class Edge {
    @GraphId
    private Long relationshipId;
    @Property
    private double weight;
    @StartNode
    private Vertex src;
    @EndNode
    private Vertex dest;
}

I have a Spring Boot application and I am using spring-data-neo4j (4.0.0) with a local neo4j server. When I save a Vertex object with 3-10 edges in the edges list it works well. If the edges list becomes larger (up to 20-30 edge objects), the application fails to save the vertex. I save the vertex using the GraphRepository from org.springframework.data.neo4j:

vertexRepository.save(vertex); 

I have left all the tunning parameters as default since I am not saving large amounts of data yet. I have set the size of the memory pool and permanent generation space to:

 export JAVA_OPTS="-Xmx256M -XX:MaxPermSize=512M"

In the messages log I get:

2016-05-31 12:12:06.592+0000 WARN  [o.n.k.i.c.MonitorGc] GC Monitor: Application threads blocked for 327ms.
2016-05-31 12:12:08.256+0000 WARN  [o.n.k.i.c.MonitorGc] GC Monitor: Application threads blocked for 260ms.
2016-05-31 12:12:09.915+0000 WARN  [o.n.k.i.c.MonitorGc] GC Monitor: Application threads blocked for 258ms.
2016-05-31 12:12:12.690+0000 WARN  [o.n.k.i.c.MonitorGc] GC Monitor: Application threads blocked for 269ms.

The Neo4j Documentation suggests that for 2 Million Nodes you need 512Mb of Heap so that would suffice for my case.

My machine is a Google Compute Engine Instance: custom (2 vCPUs, 8 GB memory). What could be the problem here?

Answer 1

Upgrading to SDN 4.1.1 solved my problem.