[jira] [Comment Edited] (JENA-289) Respect query timeouts in TDB implementation

["Simon Helsen (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/JENA-289?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13464029#comment-13464029 ] 

Simon Helsen edited comment on JENA-289 at 9/27/12 5:48 AM:
------------------------------------------------------------

The main problem we face is that it may cause a production system to go on its knees. We recently had such an instance because of a mistake in how a query was written. Without a reliable abort we put ourselves at great risk. And I should add that during our testing efforst, we are encountering countless queries which run into this situation and refuse to abort. These queries can usually be fixed or rewritten, but it seems remarkably easy for our clients to produce such queries which require the server to be killed. The latter action then often causes corruptions because Jena 2.7.1 is not good at handling abrupt process kills. And since there is no 2.7.4 release with all these fixes...
                
      was (Author: shelsen):
    The main problem we face is that may cause a production system to go on its knees. We recently had such an instance because of a mistake in how a query was written. Without a reliable abort we put ourselves at great risk. 
                  
> Respect query timeouts in TDB implementation
> --------------------------------------------
>
>                 Key: JENA-289
>                 URL: https://issues.apache.org/jira/browse/JENA-289
>             Project: Apache Jena
>          Issue Type: Improvement
>          Components: TDB
>    Affects Versions: TDB 0.9.1
>            Reporter: Mark Buquor
>            Priority: Critical
>         Attachments: TestTimeout.java, TestTimeout.java
>
>
> In general use, we sometimes see queries throw QueryCancelledException several seconds/minutes after the expected timeout. This is acceptable to a degree, but it appears that there are cases where a rogue query could execute unmitigated. The attached testcase is an example of a query that will execute and consume CPU/heap until an OutOfMemoryError is thrown.
> Example: The following query with 10s timeouts executed for ~7 minutes before throwing an OOME.
> Aug 3, 2012 10:18:19 AM Executing query [limit=1,000 timeout1=10s timeout2=10s]: SELECT * WHERE { ?a ?b ?c . ?c ?d ?e }
> java.lang.OutOfMemoryError: GC overhead limit exceeded
>     at java.util.HashSet.<init>(HashSet.java:86)
>     at org.openjena.atlas.iterator.FilterUnique.<init>(FilterUnique.java:26)
>     at org.openjena.atlas.iterator.Iter.distinct(Iter.java:438)
>     at com.hp.hpl.jena.tdb.solver.StageMatchTuple.makeNextStage(StageMatchTuple.java:116)
>     at com.hp.hpl.jena.tdb.solver.StageMatchTuple.makeNextStage(StageMatchTuple.java:44)
>     at org.openjena.atlas.iterator.RepeatApplyIterator.hasNext(RepeatApplyIterator.java:49)
>     at org.openjena.atlas.iterator.Iter$4.hasNext(Iter.java:295)
>     at com.hp.hpl.jena.sparql.engine.iterator.QueryIterPlainWrapper.hasNextBinding(QueryIterPlainWrapper.java:54)
>     at com.hp.hpl.jena.sparql.engine.iterator.QueryIteratorBase.hasNext(QueryIteratorBase.java:108)
>     at com.hp.hpl.jena.sparql.engine.iterator.QueryIterSlice.hasNextBinding(QueryIterSlice.java:76)
>     at com.hp.hpl.jena.sparql.engine.iterator.QueryIteratorBase.hasNext(QueryIteratorBase.java:108)
>     at com.hp.hpl.jena.sparql.engine.iterator.QueryIteratorWrapper.hasNextBinding(QueryIteratorWrapper.java:40)
>     at com.hp.hpl.jena.sparql.engine.iterator.QueryIteratorBase.hasNext(QueryIteratorBase.java:108)
>     at com.hp.hpl.jena.sparql.engine.iterator.QueryIteratorWrapper.hasNextBinding(QueryIteratorWrapper.java:40)
>     at com.hp.hpl.jena.sparql.engine.iterator.QueryIteratorBase.hasNext(QueryIteratorBase.java:108)
>     at com.hp.hpl.jena.sparql.engine.iterator.QueryIteratorWrapper.hasNextBinding(QueryIteratorWrapper.java:40)
>     at com.hp.hpl.jena.sparql.engine.iterator.QueryIteratorBase.hasNext(QueryIteratorBase.java:108)
>     at com.hp.hpl.jena.sparql.engine.ResultSetStream.hasNext(ResultSetStream.java:72)
>     at com.hp.hpl.jena.sparql.resultset.ResultSetApply.apply(ResultSetApply.java:41)
>     at com.hp.hpl.jena.sparql.resultset.XMLOutput.format(XMLOutput.java:52)
>     at com.hp.hpl.jena.query.ResultSetFormatter.outputAsXML(ResultSetFormatter.java:482)
>     at com.hp.hpl.jena.query.ResultSetFormatter.outputAsXML(ResultSetFormatter.java:460)
>     at TestTimout.main(TestTimout.java:84)
> Aug 3, 2012 10:25:41 AM Finished

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira