Re: [Neo4j] What do propertystore.db and propertystore.db.strings store exactly?

2015-07-18 Thread Zongheng Yang
If those are serialized String objects then I'm seeing the following mismatch between measurement and calculation: The graph I'm using has 4.9 million nodes, each of which has 40 string properties (each of which has 16 characters). It has 70 million directed edges, each of which has 1 string prop

Re: [Neo4j] modifying more than 63 indexes in a single transaction not supported

2015-07-18 Thread Axel Morgner
See also https://github.com/neo4j/neo4j/issues/4993 Am 16.07.2015 um 17:16 schrieb Axel Morgner: > We see the same exception when bulk-updating nodes with more than 63 > properties (which is a common use-case) in one transaction with Structr. > > There are only three manual indexes, so there's som

Re: [Neo4j] Indexes for inequality comparisons

2015-07-18 Thread Alireza Rezaei Mahdiraji
Great, thanks :) On Saturday, July 18, 2015 at 11:23:26 AM UTC+2, Michael Hunger wrote: > > this summer :) > > Am 17.07.2015 um 12:10 schrieb Alireza Rezaei Mahdiraji < > alire...@gmail.com >: > > > That is promising, and when would that be ? > > Cheers, > Alireza > > On Thursday, July 16, 2015

Re: [Neo4j] What do propertystore.db and propertystore.db.strings store exactly?

2015-07-18 Thread Chris Vest
> Does this sound right? Also, node properties & relationship properties are > interleaved and stored together in these files, right? Yes and yes. > Lastly -- is everything in (1) and (2) deserialized JVM objects in raw bytes > *or* just UTF-8 characters? It could make a difference, since if

Re: [Neo4j] Indexes for inequality comparisons

2015-07-18 Thread Michael Hunger
this summer :) > Am 17.07.2015 um 12:10 schrieb Alireza Rezaei Mahdiraji > : > > > That is promising, and when would that be ? > > Cheers, > Alireza > > On Thursday, July 16, 2015 at 4:39:30 PM UTC+2, Michael Hunger wrote: > This is being worked on for 2.3 and will be hopefully part of the 2

Re: [Neo4j] NEO4J Carshes often

2015-07-18 Thread Michael Hunger
What Chris said > Try adding dbms.pagecache.memory=2g (or a smaller value) to neo4j.properties. also disable the node and relationship autoindexes you most likely don't need them. You should also probably upgrade to M02 or use 2.2.3 (stable). And you might set in neo4j.properties keep_logical_

Re: [Neo4j] Help with edge query

2015-07-18 Thread Michael Hunger
not sure what it expects. for recent versions of Neo4j you'd change that to: (Don't you also need id's for the source and target?) MATCH (x:Person)-[:DIRECTED]->(m) return x.name, n.title Not sure what it expects as edge query, usually it is the source and target id's plus edge properties or t

Re: [Neo4j] Feasibility of an in-memory graph queried with Cypher, preferably in scala

2015-07-18 Thread Michael Hunger
Pat, for console.neo4j.org I create in-memory graph databases (and put them into each users web-session) using the ImpermanentGraphDatabase for org.neo4j:neo4j-kernel:test-jar package. That works pretty well. It is still acid but uses an virtual, in-memory FS. Michael > Am 17.07.2015 um 17:21

[Neo4j] Feasibility of an in-memory graph queried with Cypher, preferably in scala

2015-07-18 Thread Patrick O'Reilly
I have a small graph with approximately 1000 vertices and 5000 edges, and perhaps 100 properties at each vertex. The graph only needs to be readable - I have no write requirements and I don't care about integrity or ACID etc. I'd like to embed the graph in my program (preferably scala but java