Re: CleanerThread never terminates, causes leaks in webapp
I am running a webapp (under both Tomcat 6 7) that exhibits the same behaviour: 14-Apr-2011 10:32:54 org.apache.catalina.loader.WebappClassLoader clearReferencesThreads SEVERE: The web application [/mapmosaic] appears to have started a thread named [Batik CleanerThread] but has failed to stop it. This is very likely to create a memory leak. Under normal development conditions, the app is re-deployed many times per day, and causes an OutOfMemoryError eventually. Even if in production, the memory leak is so slow as to be negligeable, it would be far better to provide a hook so I can terminate it when the webapp closes. Martin On 12 April 2011 21:30, tiwaris tiwa...@gmail.com wrote: I also have encountered this issue. This happens rarely. When I quit my java app, it sill has not released the shell (the process is still running). I then used jvisualvm to monitor all the running threads in my application and noticed that Batik CleanerThread was still running. I incorporated the patch. Will have to see if the issue reappears. -- View this message in context: http://batik.2283329.n4.nabble.com/CleanerThread-never-terminates-causes-leaks-in-webapp-tp2978944p3445621.html Sent from the Batik - Users mailing list archive at Nabble.com. - To unsubscribe, e-mail: batik-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: batik-users-h...@xmlgraphics.apache.org -- From my MacBook Pro - To unsubscribe, e-mail: batik-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: batik-users-h...@xmlgraphics.apache.org
Re: CleanerThread never terminates, causes leaks in webapp
I also have encountered this issue. This happens rarely. When I quit my java app, it sill has not released the shell (the process is still running). I then used jvisualvm to monitor all the running threads in my application and noticed that Batik CleanerThread was still running. I incorporated the patch. Will have to see if the issue reappears. -- View this message in context: http://batik.2283329.n4.nabble.com/CleanerThread-never-terminates-causes-leaks-in-webapp-tp2978944p3445621.html Sent from the Batik - Users mailing list archive at Nabble.com. - To unsubscribe, e-mail: batik-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: batik-users-h...@xmlgraphics.apache.org
Re: CleanerThread never terminates, causes leaks in webapp
Hi Joel, Joel Carranza j...@gatekeeper.com wrote on 02/17/2010 05:57:50 PM: We are using Batik in a webapp context and the thread org.apache.batik.util.CleanerThread never terminates. Correct, it is a daemon thread. My understanding is that this signals the JVM that it can terminate it if there is nothing else of interest in it's context. This is a serious problem as it prevents the webapp's classloader from being garbage collected. Why is that a particularly serious problem? It seems to me that generally speaking the classloader isn't GCed unless the JVM goes away anyway. Am I doing something wrong? Is there a way to clean up and stop this thread when my webapp exits? Batik doesn't provide one currently. It's a little tricky since you shouldn't kill that thread until all objects that have been registered with the associated ReferenceQueue have been dequeued. You might be able to do something with TheadGroups and 'stop'. You could also register a PhantomWeakReference and have it's 'cleared()' method throw a ThreadDeath exception. You also have access to the source and contributions are always welcome ;)
Re: CleanerThread never terminates, causes leaks in webapp
This is a real problem in cases where you want to perform a hot redeploy of a webapp within Tomcat or other application servers. Single JVM, multiple classloaders. A classloader (and all its classes) can only get GC'ed if everything is cleaned up (including running threads). If you don't, after a couple of deploys you get a PermGen: Out of Memory error. GeoTools basically has the identical problem [http://jira.codehaus.org/browse/GEOT-2742]. I will investigate how to apply a similar approach and submit a patch. cheers, Joel Carranza j...@gatekeeper.com Gatekeeper Systems thomas.deweese wrote: Hi Joel, Joel Carranza j...@gatekeeper.com wrote on 02/17/2010 05:57:50 PM: We are using Batik in a webapp context and the thread org.apache.batik.util.CleanerThread never terminates. Correct, it is a daemon thread. My understanding is that this signals the JVM that it can terminate it if there is nothing else of interest in it's context. This is a serious problem as it prevents the webapp's classloader from being garbage collected. Why is that a particularly serious problem? It seems to me that generally speaking the classloader isn't GCed unless the JVM goes away anyway. Am I doing something wrong? Is there a way to clean up and stop this thread when my webapp exits? Batik doesn't provide one currently. It's a little tricky since you shouldn't kill that thread until all objects that have been registered with the associated ReferenceQueue have been dequeued. You might be able to do something with TheadGroups and 'stop'. You could also register a PhantomWeakReference and have it's 'cleared()' method throw a ThreadDeath exception. You also have access to the source and contributions are always welcome ;) -- View this message in context: http://old.nabble.com/CleanerThread-never-terminates%2C-causes-leaks-in-webapp-tp27631790p27642863.html Sent from the Batik - Users mailing list archive at Nabble.com. - To unsubscribe, e-mail: batik-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: batik-users-h...@xmlgraphics.apache.org
Re: CleanerThread never terminates, causes leaks in webapp
Hi everyone, GeoTools basically has the identical problem [http://jira.codehaus.org/browse/GEOT-2742]. I will investigate how to apply a similar approach and submit a patch. I've just noticed bug 48771 [1] was created. (In general, it's always good to have a link between mailing list threads and bug reports, whenever they are tightly related such as in this case.) Thanks, Helder [1] https://issues.apache.org/bugzilla/show_bug.cgi?id=48771 - To unsubscribe, e-mail: batik-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: batik-users-h...@xmlgraphics.apache.org
CleanerThread never terminates, causes leaks in webapp
We are using Batik in a webapp context and the thread org.apache.batik.util.CleanerThread never terminates. This is a serious problem as it prevents the webapp's classloader from being garbage collected. Am I doing something wrong? Is there a way to clean up and stop this thread when my webapp exits? Many thanks Joel Carranza j...@gatekeeper.com Gatekeeper System -- View this message in context: http://old.nabble.com/CleanerThread-never-terminates%2C-causes-leaks-in-webapp-tp27631790p27631790.html Sent from the Batik - Users mailing list archive at Nabble.com. - To unsubscribe, e-mail: batik-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: batik-users-h...@xmlgraphics.apache.org
Re: CleanerThread never terminates, causes leaks in webapp
Hi Joel, We are using Batik in a webapp context and the thread org.apache.batik.util.CleanerThread never terminates. This is a serious problem as it prevents the webapp's classloader from being garbage collected. Am I doing something wrong? Is there a way to clean up and stop this thread when my webapp exits? Well, it's hard to guess is something is being done wrong having provided such little information... ;-) I'm assuming you are embedding an JSVGCanvas such as demonstrated in the demo [1] page...? (Source code quickly available there.) If you can steadily reproduce this, I'd ask you to provide a code snippet of your application and/or the SVG file(s) which are making this happen. :-) Finally, if you are using this within a webapp (I'm assuming a client-side application, but this somehow applies to server-side as well), I'm not sure if you'd even want to terminate the environment many times: that can become a heavy burden. I'd suggest looking into the possibility of (for example) hiding the canvas whenever not needed and display it again only with the needed tweaks/newly created document (somehow a.k.a. object pooling). Many thanks Hope this helps, Helder [1] http://xmlgraphics.apache.org/batik/demo.html - To unsubscribe, e-mail: batik-users-unsubscr...@xmlgraphics.apache.org For additional commands, e-mail: batik-users-h...@xmlgraphics.apache.org
