The stability of Neo4j CE server

[limelyj]

Neo4j Server version: 4.0.0(community)
Neo4j Browser version: 4.0.1
Operating System: macOS Mojave 10.14.6
settings:
- NEO4J_dbms_memory_pagecache_size=2G

Hi Community,

I'm using the docker version of Neo4j server. Everything's fine, except that the server will crash and restart when long time query is executed.

Example of query include: MATCH (n) DETACH DELETE n when nodes are plenty much, or when complicated search is applied.

Did anyone face this kind of problem? or is anything wrong with my settings?

[anthapu · Accepted solution]

Welcome to the community.

When you run a query like that it is run as a single transaction. So, you need enough heap memory to be able to complete that operation.

What is your system memory available?

If the system is crashing when you have a complex query it could be due to same reason, not enough heap memory available.

There are 3 options here.

1. modify the query to run in smaller batches.
2. Increase the heap memory.

For stability purposes we can add a query guard for memory.

By default the heap allocated to a query is unlimited.

Please take a look at these configurations

https://neo4j.com/docs/operations-manual/current/reference/configuration-settings/#config_dbms.memor...

By setting dbms.memory.transaction.max_size you are limiting how much memory a single transaction can use. if it exceeds that size the query will be killed.

[limelyj]

Thanks for the answer!

I'm trying on my own laptop, with many other processes, so the memory left is only about 2G. Setting dbms.memory.transaction.max_size helps a lot.