Per your requests I removed the Tomcat Native DLL and observed the following:
1) Most important, I could not get it to crash using the same repeatable
process as before
2) It did not appear significantly or functionally slower.
Thank you for your help. Though I'm a bit disappointed that Tomcat has this
issue, I am nonetheless happy to have a work around that meets my needs.
________________________________
Brad A. Bode
Principal
Software Systems
Foundry Engineering
________________________________
From: Simone Giannecchini <[email protected]>
To: Brad Bode <[email protected]>
Cc: "[email protected]"
<[email protected]>
Sent: Wednesday, April 10, 2013 10:47 AM
Subject: Re: [Geoserver-users] JVM goes up in smoke
Ciao Brad,
please, read below...
Regards,
Simone Giannecchini
==
GeoServer training in Milan, 6th & 7th June 2013! Visit
http://geoserver.geo-solutions.it for more information.
==
Ing. Simone Giannecchini
@simogeo
Founder/Director
GeoSolutions S.A.S.
Via Poggio alle Viti 1187
55054 Massarosa (LU)
Italy
phone: +39 0584 962313
fax: +39 0584 1660272
mob: +39 333 8128928
http://www.geo-solutions.it
http://twitter.com/geosolutions_it
-------------------------------------------------------
On Wed, Apr 10, 2013 at 7:00 PM, Brad Bode <[email protected]> wrote:
> Thank you for the quick response. I did some cursory searching and could
> not find answers to the following:
>
> 1) What is the purpose of Tomcat Native? It seems it is intended to make the
> server faster and more compatible with other server tech. If that's so,
> removing native support will make my server slower, which isn't desirable.
> Since GeoServer is a must for our application I suppose I will have to.
> Luckily we don't have many concurrent users to worry about.
It should speed up tomcat handling of output stream. In my experience
it does not make a noticeable difference.
Others may have different opinions, as such I would like to hear them.
>
> 2) How do you remove tomcat native? Is it as simple as removing the
> "tcnative-1.dll" file?
>
It should be.
Check the doc (here as an instance for more info
http://tomcat.apache.org/tomcat-7.0-doc/apr.html)
> I appreciate the help. I realize this isn't your problem. And as Andrea
> pointed out, I am amazed that this issue has not received more notice. If
> something as simple as a connection closing prematurely can crash the entire
> VM, then it should receive major attention.
Well, this has been around for a while. Believe there are very
disappointing bugs around, I guess one does what one can :)
>
> Is there anything I can do to help?
As Andrea pointed out we could try to spend even more time to check
if/how we can work this around. He already tried in the past,
and I tried myself too. I guess the best thing to do would be fixing
the apr code directly (I did not check if it is available or not).
>
> ________________________________
> Brad A. Bode
> Principal
> Software Systems
> Foundry Engineering
>
> ________________________________
> From: Simone Giannecchini <[email protected]>
> To: Brad Bode <[email protected]>
> Cc: "[email protected]"
> <[email protected]>
> Sent: Wednesday, April 10, 2013 1:16 AM
> Subject: Re: [Geoserver-users] JVM goes up in smoke
>
> Ciao Brad,
> thanks for getting back to us.
> I believe this problem is generated by the tomcat native connector.
> Can you confirm that by uninstalling it the problem goes away?
>
> Looking at this link http://bit.ly/Zm91zN it looks like we are
> flushing a stream that's been closed.
> This happens frequently when talking to clients as they drop the
> connection while we are sending an image. I will talk to andrea to see
> if we can improve this but I kind of remember it would be almost
> impossible to know whether or not a stream has been dropped by a
> client.
>
> Long story short, short term fix should be disabling tomcat native.
>
> Let me know if this helps.
>
> Regards,
> Simone Giannecchini
> ==
> GeoServer training in Milan, 6th & 7th June 2013! Visit
> http://geoserver.geo-solutions.it for more information.
> ==
>
> Ing. Simone Giannecchini
> @simogeo
> Founder/Director
>
> GeoSolutions S.A.S.
> Via Poggio alle Viti 1187
> 55054 Massarosa (LU)
> Italy
> phone: +39 0584 962313
> fax: +39 0584 1660272
> mob: +39 333 8128928
>
> http://www.geo-solutions.it
> http://twitter.com/geosolutions_it
>
> -------------------------------------------------------
>
>
> On Wed, Apr 10, 2013 at 1:04 AM, Brad Bode <[email protected]> wrote:
>> I mentioned that I was able to crash the JVM as well. I've been away for a
>> bit, but after returning I was able to duplicate the issue consistently.
>>
>> Duplicating the issue
>> I have two users issue http requests to our web page simultaneously. One
>> is
>> zooming in and out using the mousewheel within our map (which is calling
>> WMS). The other is simple jumping to our map page, then navigating away,
>> then returning. We do this repeatedly until the JVM crashes within 5
>> minutes.
>>
>>
>> Notes
>> - Our dataset is very a very small shapefile of 1.5 mb.
>> - We are using Spring MVC as our front end and embedding a map into our
>> page
>> using Open Layers to call WMS in Geoserver.
>> - Attached you will find a zip of the logs. Stderr, stout, and a the JVM
>> error mentioned in the top of the stdout log.
>>
>> Log File Comparison
>> I think it may be relevant to look at the last actions in the log file.
>> For
>> starters, the stdout file has the following at the end:
>>
>> 09 Apr 15:24:22 DEBUG [wms.map] - Writing png image ...
>> 09 Apr 15:24:22 DEBUG [geotools.image] - Encoded input image for png
>> writer
>> 09 Apr 15:24:22 DEBUG [geotools.image] - Getting a writer
>> 09 Apr 15:24:22 DEBUG [geotools.image] - Setting write parameters for this
>> writer
>> 09 Apr 15:24:22 DEBUG [geotools.image] - Writer is native
>> 09 Apr 15:24:22 DEBUG [geotools.image] - About to write png image
>>
>> Note that the write.png.image does not say "done" as it does in other
>> logs.
>>
>>
>> Comparing that to the closest timeframe in the stderr:
>> SEVERE: Servlet.service() for servlet [DataForge2] in context with path
>> [/DataForge] threw exception [Request processing failed; nested exception
>> is
>> org.apache.tiles.impl.CannotRenderException: ServletException including
>> path
>> '/WEB-INF/layouts/default.jspx'.] with root cause
>> java.lang.IllegalStateException: getOutputStream() has already been called
>> for this response
>> at org.apache.catalina.connector.Response.getWriter(Response.java:639)
>> at
>>
>> org.apache.catalina.connector.ResponseFacade.getWriter(ResponseFacade.java:214)
>> at
>>
>> javax.servlet.ServletResponseWrapper.getWriter(ServletResponseWrapper.java:105)
>>
>>
>> I'm looking into the problem of how this is happening. The response was
>> likely closed prematurely.
>>
>> And finally, in the top of the stdout:
>>
>> 2013-04-09 15:21:33 Commons Daemon procrun stdout initialized
>> #
>> # A fatal error has been detected by the Java Runtime Environment:
>> #
>> # EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x5acc4bdf, pid=2864,
>> tid=1180
>> #
>> # JRE version: 6.0_38-b05
>> # Java VM: Java HotSpot(TM) Server VM (20.13-b02 mixed mode windows-x86 )
>> # Problematic frame:
>> # C [tcnative-1.dll+0x4bdf]
>> #
>> # An error report file with more information is saved as:
>> # C:\Program Files (x86)\DataForge\webserver\bin\hs_err_pid2864.log
>> #
>> # If you would like to submit a bug report, please visit:
>> # http://java.sun.com/webapps/bugreport/crash.jsp
>> # The crash happened outside the Java Virtual Machine in native code.
>> # See problematic frame for where to report the bug.
>> #
>>
>>
>> I assume this is from the previous failure since it occurs at the earliest
>> point in the file.
>>
>> What I don't know
>> This seems to happen only when I use GeoServer WMS. However, I am doing
>> more
>> testing.
>>
>> Do you have any suggestions as to how to proceed?
>>
>> Do you see anything in the logs that looks suspicious to you?
>>
>> Thank you
>>
>> ________________________________
>> Brad A. Bode
>> Principal
>> Software Systems
>> Foundry Engineering
>>
>> ________________________________
>> From: Simone Giannecchini <[email protected]>
>> To: cmaul <[email protected]>
>> Cc: [email protected]
>> Sent: Friday, March 29, 2013 10:27 AM
>> Subject: Re: [Geoserver-users] JVM goes up in smoke
>>
>> Dear Christian,
>> going back to this, I have run some (relatively basic) tests against
>> ecw using a 2gb dataset (which is by accident a raster covergin
>> australia :)).
>> I tested on windows with both 2.2.x and master and I was not able to
>> get any strange behavior.
>>
>> Thinking about a way to get some hints on what's going on on your
>> installation, I guess one thing you could do would be reproduce the
>> problematic situation and then
>> troubleshoot GeoServer following this instructions
>> http://docs.geoserver.org/latest/en/user/production/troubleshooting.html
>> to get a feeling about where the problem is.
>>
>> Let us know if you make any progress.
>>
>> Regards,
>> Simone Giannecchini
>> ==
>> Our support, Your Success! Visit http://opensdi.geo-solutions.it for
>> more information.
>> ==
>>
>> Ing. Simone Giannecchini
>> @simogeo
>> Founder/Director
>>
>> GeoSolutions S.A.S.
>> Via Poggio alle Viti 1187
>> 55054 Massarosa (LU)
>> Italy
>> phone: +39 0584 962313
>> fax: +39 0584 1660272
>> mob: +39 333 8128928
>>
>> http://www.geo-solutions.it
>> http://twitter.com/geosolutions_it
>>
>> -------------------------------------------------------
>>
>>
>> On Tue, Mar 19, 2013 at 11:59 PM, cmaul <[email protected]>
>> wrote:
>>> Simone,
>>>
>>> I am a little bit further now. The ecws that didn't work were actually a
>>> layergroup of 11 images.
>>> These images were in EPSG:28355 and reprojected to EPSG:990913.
>>>
>>> I have done a test (4 threads/ 170000 images) to tile one image which
>>> went
>>> fine. Speed was 163 minutes or 17.2 tiles per second, which is o.k. as
>>> well. The next thing I will try is to have a layergroup and NOT to
>>> reproject, i.e. to tile in EPSG:28355. I'll tell you how that works (or
>>> not).
>>>
>>> Simone, if you want any ECW files for testing, I am happy to put stuff on
>>> our FTP server. The 11 files in question are between 8 and 14 GB and have
>>> a
>>> total of 128GB. So, a subset of 4 or 5 neighbouring ecws perhaps.
>>>
>>> Cheers
>>>
>>> Christian
>>>
>>>
>>>
>>> -----
>>> ____________________________
>>>
>>> Dr Christian Maul
>>> Project Manager
>>>
>>> Information Services Branch
>>> Department of Sustainability and Environment
>>> Level13, Marland House, 570 Bourke Street
>>> Melbourne 3000
>>>
>>> PO Box 500, East Melbourne Vic 3002
>>>
>>>
>>> Telephone: +61-3-8636 2325
>>> Telefax: +61-3-8636 2813
>>> --
>>> View this message in context:
>>>
>>> http://osgeo-org.1560.n6.nabble.com/JVM-goes-up-in-smoke-tp5041006p5041499.html
>>> Sent from the GeoServer - User mailing list archive at Nabble.com.
>>>
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> Everyone hates slow websites. So do we.
>>> Make your web apps faster with AppDynamics
>>> Download AppDynamics Lite for free today:
>>> http://p.sf.net/sfu/appdyn_d2d_mar
>>> _______________________________________________
>>> Geoserver-users mailing list
>>> [email protected]
>>> https://lists.sourceforge.net/lists/listinfo/geoserver-users
>>
>>
>> ------------------------------------------------------------------------------
>> Own the Future-Intel(R) Level Up Game Demo Contest 2013
>> Rise to greatness in Intel's independent game demo contest. Compete
>> for recognition, cash, and the chance to get your game on Steam.
>> $5K grand prize plus 10 genre and skill prizes. Submit your demo
>> by 6/6/13. http://altfarm.mediaplex.com/ad/ck/12124-176961-30367-2
>> _______________________________________________
>> Geoserver-users mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/geoserver-users
>>
>>
>
>
------------------------------------------------------------------------------
Precog is a next-generation analytics platform capable of advanced
analytics on semi-structured data. The platform includes APIs for building
apps and a phenomenal toolset for data science. Developers can use
our toolset for easy data analysis & visualization. Get a free account!
http://www2.precog.com/precogplatform/slashdotnewsletter
_______________________________________________
Geoserver-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geoserver-users
