[
https://issues.apache.org/jira/browse/JENA-985?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14624558#comment-14624558
]
Andy Seaborne commented on JENA-985:
------------------------------------
TDB has internally has a very large cache for NodeIds and Nodes. The cache has
an LRU policy. This is configurable with {{StoreParams}} but the default
values should work well. This cache is quite important for performance.
As {{graph.find}} runs, it is filling up that cache when run first time, or
replacing cache entries with Node_URIs, if there was code already run.
What you are seeing is expected behaviour. RAM usage will eventually stablise
- it may take a while because the cache is large.
> Iterate using Apache Jena ExtendedIterator on Graph with big amount of triples
> ------------------------------------------------------------------------------
>
> Key: JENA-985
> URL: https://issues.apache.org/jira/browse/JENA-985
> Project: Apache Jena
> Issue Type: Bug
> Components: Core
> Affects Versions: Jena 2.13.0
> Environment: *Hardware*
> Windows 7 64-bit
> Intel Core i7 4785T @ 2.20GHz
> RAM 16,0GB DDR3
> 465GB Samsung SSD 850 EVO 500G SCSI Disk Device (SSD)
> *Software environment*
> java version "1.7.0_75"
> Java(TM) SE Runtime Environment (build 1.7.0_75-b13)
> Java HotSpot(TM) 64-Bit Server VM (build 24.75-b04, mixed mode)
> *Running options*
> VM options: -Xmx14g
> Reporter: Eugene Tenkaev
> Priority: Minor
>
> I'm generating Apache Jena Graph from DBpedia dumps and now I want iterate
> through all "dbpedia-owl:abstract".
> So I do something like this:
> {code:java}
> ExtendedIterator<Triple> iterator = Graph.find(Node.ANY,
> NodeFactory.createURI("dbpedia-owl:abstract"), Node.ANY);
> {code}
> But then I try to iterate, memory consumption is increased, so looks like
> "ExtendedIterator" store found nodes.
> I use VisualVM profiler and found that while I iterate, count of
> "com.hp.hpl.jena.graph.Node_URI" is increasing.
> I try to do "iterator.reset()" but this takes no effect.
> Is this bug or feature?:D
> Can I iterate through all DBpedia abstracts without storing nodes and without
> increasing consumption of memory that gc can't freed?
> Sorry for my bad english.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
