All,
I asked about this problem last week, and Andrea replied, but I was not able
to get back to it until now. Please read the email below for a statement of
the problem. I've attached a file that has a version of the wfs query that
works with 260 polygon features, but breaks with 700 or so. With GS 1.4 it
worked well with 2500 features.
I'm using a PostGis DB. I can send you a shapefile of the layer, but it is
~ 2 MB and I didn't want to send it along unsolicited. Please let me know
if you need it.
And the 'memory leak' I mentioned is no longer it problem....it wasn't
Geoserver.
Thanks,
Steve
-----Original Message-----
From: geoserver-users-bounces@lists.sourceforge.net
[mailto:geoserver-users-bounces@lists.sourceforge.net] On Behalf Of Andrea
Aime
Sent: Friday, January 04, 2008 11:57 AM
To: Stephen Crawford
Cc: geoserver-users@lists.sourceforge.net
Subject: Re: [Geoserver-users] 1.6 RC2 problems
Stephen Crawford ha scritto:
All,
I have recently upgraded from GS 1.4 to 1.6 RC2 (on linux with JDK
1.6.0_03). One of my applications does some calculations on the
client side and then makes a wfs request with a filter to get some
geometries. The request may at times request up to 2500 features.
This has always worked smoothly with GS 1.4, but now with 1.6 it
crashes at 700 or so features. By crash I mean I get an error (in
geosererver logs) that looks like this:
2008-01-04 10:49:06,816 ERROR [org.geotools.rendering] -
java.lang.StackOverflowError
at
org.geotools.filter.FilterCapabilities.supports(FilterCapabilities.java:431)
at
org.geotools.filter.visitor.PostPreProcessFilterSplittingVisitor.visit
LogicO
perator(PostPreProcessFilterSplittingVisitor.java:617)
at
org.geotools.filter.visitor.PostPreProcessFilterSplittingVisitor.visit
(PostP
reProcessFilterSplittingVisitor.java:600)
at org.geotools.filter.AndImpl.accept(AndImpl.java:60)
at
org.geotools.filter.visitor.PostPreProcessFilterSplittingVisitor.visit
LogicO
perator(PostPreProcessFilterSplittingVisitor.java:672)
at
org.geotools.filter.visitor.PostPreProcessFilterSplittingVisitor.visit
(PostP
reProcessFilterSplittingVisitor.java:600)
at org.geotools.filter.AndImpl.accept(AndImpl.java:60)...with the final three lines repeating for the number of features
requested. The same thing happens when I use a similar filter to make
a wms request. Again, these both worked well with 1.4.The filter in both cases is something like this:
<Filter><Or>
<PropertyIsEqualTo>
<PropertyName>areasymbol</PropertyName>
<Literal>AL003</Literal>
</PropertyIsEqualTo>
<PropertyIsEqualTo>
<PropertyName>areasymbol</PropertyName>
<Literal>AL009</Literal>
</PropertyIsEqualTo>
..........
</Or></Filter>
And I must say that we have generally had memory leak problems since
going to 1.6. I know have the min and max heap size set to -Xms64
-Xmx1024, which is higher than I had previously for 1.4
We were not aware of either problem, so I don't have a workaround for you.
This seems serious, but I need more information in order to reproduce it.
Which datastore are you using?
Is there any chance you can give us a sample of your data and a request that
makes GeoServer break?
Cheers
Andrea
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft Defy all challenges.
Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Geoserver-users mailing list
Geoserver-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-users
fragi_filter_wfs.txt (129 KB)