'Relationship[nnn] in use error' when issuing concurrent Cypher DELETE queries when cache is enabled

[lassewesth]

@blevine: 'Environment:
Neo4j 1.8GA (also repro on 1.9 M01)
Ubuntu 12.04
Java:
java version "1.6.0_32"
Java(TM) SE Runtime Environment (build 1.6.0_32-b05)
Java HotSpot(TM) 64-Bit Server VM (build 20.7-b02, mixed mode)
Hardware:
Quad core home-brew, 8 G memory

I have a test that is designed to test out a retry strategy for dealing with deadlock exceptions. The test creates 50 nodes and multiple relationships between one node and the next node. It then issues 50 DELETE queries against each of these nodes. The specific DELETE query is:

START n = node({neoId}) MATCH n-[r?]-() DELETE r, n

Some of these queries produce the error: "Relationship[nnn] not in use"

Once this error is seen, any Cypher query that would encounter that relationship returns the error:

"Relationship[nnn] not found. This can be because someone else deleted this entity while we were trying to read properties from it, or because of concurrent modification of other properties on this entity. The problem should be temporary."

For example, the query:

START r = rel(*) return r 

will return this error. Also note that the query

START r = rel(*) return count(*) 

will return a non-zero value even though all of the relationships should have been deleted.

This condition persists until the database is restarted.

This appears to be a caching issue. When the node cache is disabled by setting cache_type=none, this problem does not occur.

I have a test (requires Node.js) that reproduces this problem at https://github.com/blevine/neo4j-concurrent-delete-test. The conc directory contains a full Node.js project. You should be able to cd into conc and run npm test. Or if you don't have npm installed, cd into conc and run node index.js'

[lassewesth]

---

strings:

• "This issue was migrated from "
• ""
  values:
• https://github.com/neo4j/community/issues/962
  valueCount: 1

[lassewesth]

@wfreeman: +1. I've seen the same thing. Requires server restart to clear sometimes.

[lassewesth]

@peterneubauer: THanks for reporting, @jakewins is wokring on it.

[jakewins]

This is fixed in Neo4j 1.8.2 and in upcoming 1.9 :)