Doesn't limiting the supported locales to just the ones you support help?
http://tapestry.apache.org/tapestry5.0/guide/conf.html Uli Am 29.10.2009 19:39 schrieb Christiansen Merel:
Hello out there, We're running a quite busy website with tapestry 5.0.18 and we're hitting a performance problem with the ClasspathResource class. We're using classpath asset and none of them is localised. For each hit on a page using a classpath asset, we end having 2 missed queries on the ClassLoader for potential localized version of the asset. It does not really slow down the application throughput but we end with an __huge__ CPU consumption. The javacore stack shows numerous threads scanning jar files: 3XMTHREADINFO "[ACTIVE] ExecuteThread: '2' for queue: 'weblogic.kernel.Default (self-tuning)'" (TID:0x3700E400, sys_thread_t:0x3393D030, state:B, native ID:0x009AB0B3) prio=5 4XESTACKTRACE at java/util/zip/ZipFile.getEntry(ZipFile.java:287(Compiled Code)) 4XESTACKTRACE at java/util/jar/JarFile.getEntry(JarFile.java:283(Compiled Code)) 4XESTACKTRACE at java/util/jar/JarFile.getJarEntry(JarFile.java:266(Compiled Code)) 4XESTACKTRACE at sun/misc/URLClassPath$JarLoader.getResource(URLClassPath.java:949(Compiled Code)) 4XESTACKTRACE at sun/misc/URLClassPath$JarLoader.findResource(URLClassPath.java:938(Compiled Code)) 4XESTACKTRACE at sun/misc/URLClassPath.findResource(URLClassPath.java:299(Compiled Code)) 4XESTACKTRACE at java/net/URLClassLoader$3.run(URLClassLoader.java:784(Compiled Code)) 4XESTACKTRACE at java/security/AccessController.doPrivileged(AccessController.java:219) 4XESTACKTRACE at java/net/URLClassLoader.findResource(URLClassLoader.java:781(Compiled Code)) 4XESTACKTRACE at java/lang/ClassLoader.getResource(ClassLoader.java:420(Compiled Code)) 4XESTACKTRACE at java/lang/ClassLoader.getResource(ClassLoader.java:415(Compiled Code)) 4XESTACKTRACE at java/lang/ClassLoader.getResource(ClassLoader.java:415(Compiled Code)) 4XESTACKTRACE at weblogic/utils/classloaders/GenericClassLoader.getResourceInternal(GenericClassLoader.java:170(Compiled Code)) 4XESTACKTRACE at weblogic/utils/classloaders/GenericClassLoader.getResource(GenericClassLoader.java:187(Compiled Code)) 4XESTACKTRACE at weblogic/utils/classloaders/FilteringClassLoader.getResourceInternal(FilteringClassLoader.java:89(Compiled Code)) 4XESTACKTRACE at weblogic/utils/classloaders/GenericClassLoader.getResourceInternal(GenericClassLoader.java:155(Compiled Code)) 4XESTACKTRACE at weblogic/utils/classloaders/GenericClassLoader.getResource(GenericClassLoader.java:187(Compiled Code)) 4XESTACKTRACE at java/lang/ClassLoader.getResource(ClassLoader.java:415(Compiled Code)) 4XESTACKTRACE at weblogic/utils/classloaders/GenericClassLoader.getResource(GenericClassLoader.java:190(Compiled Code)) 4XESTACKTRACE at weblogic/utils/classloaders/ChangeAwareClassLoader.getResource(ChangeAwareClassLoader.java:118(Compiled Code)) 4XESTACKTRACE at org/apache/tapestry5/ioc/internal/util/ClasspathResource.toURL(ClasspathResource.java:59(Compiled Code)) 4XESTACKTRACE at org/apache/tapestry5/ioc/internal/util/AbstractResource.exists(AbstractResource.java:137(Compiled Code)) 4XESTACKTRACE at org/apache/tapestry5/ioc/internal/util/AbstractResource.forLocale(AbstractResource.java:100(Compiled Code)) 4XESTACKTRACE at org/apache/tapestry5/internal/services/AssetSourceImpl.findRelativeAsset(AssetSourceImpl.java:99(Compiled Code)) 4XESTACKTRACE at org/apache/tapestry5/internal/services/AssetSourceImpl.getAsset(AssetSourceImpl.java:93(Compiled Code)) 4XESTACKTRACE at $AssetSource_124955c0e4c.getAsset($AssetSource_124955c0e4c.java(Compiled Code)) 4XESTACKTRACE at org/apache/tapestry5/internal/transform/AbstractIncludeAssetWorker$1.perform(AbstractIncludeAssetWorker.java:79(Compiled Code)) 4XESTACKTRACE at fr/poleemploi/d2i/tapestry/navigation/unifiee/components/Bandeau.setupRender(Bandeau.java(Compiled Code)) The stack comes from weblogic/aix run but the problem is the same on other platforms as well. Does anyone has ever experienced such problems ? The performance penalty seems to be located in the AbstractResource.forLocale() method, especially the potential.exists() call. This part of code haven't evolved with the 5.1.x version. Does anybody have any idea on how to get around this problem ? Thanks in advance -- ChristianZen --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tapestry.apache.org For additional commands, e-mail: users-h...@tapestry.apache.org
--------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tapestry.apache.org For additional commands, e-mail: users-h...@tapestry.apache.org
