First step in such case is to check whether the GC is allowed to unload classes from PermGen. The standard JVM is rather conservative in this regard – classes are born to live forever. So once loaded, classes stay in memory even if no code is using them anymore. This can become a problem when the application creates lots of classes dynamically and the generated classes are not needed for longer periods. In such a case, allowing the JVM to unload class definitions can be helpful. This can be achieved by adding just one configuration parameter to your startup scripts:

-XX:+CMSClassUnloadingEnabled

默认情况下，它被设置为false，因此要启用它，您需要显式地在Java选项中设置以下选项。如果你启用了CMSClassUnloadingEnabled, GC也会扫描PermGen并删除不再使用的类。请记住，此选项仅在UseConcMarkSweepGC也使用下面的选项启用时才有效。因此，当运行ParallelGC或串行GC时，请确保您已经通过指定将GC设置为CMS:

-XX:+UseConcMarkSweepGC

对于Sun JVM，使用命令行参数-XX:MaxPermSize=128m(显然是将128替换为您需要的任何大小)。

Jrockit也帮我解决了这个问题;但是，我注意到servlet的重启时间要短得多，所以虽然在生产环境中效果更好，但在开发过程中却有点拖后腿。

“他们”是错误的，因为我运行的是6.0.29，即使设置了所有选项也有同样的问题。正如蒂姆·豪兰所说，这些选择只会推迟不可避免的事情。它们允许我在遇到错误之前重新部署3次，而不是每次重新部署。

我有我们在这里谈论的问题，我的场景是eclipse-helios + tomcat + jsf，而您所做的是将一个简单的应用程序部署到tomcat。我在这里展示了同样的问题，解决方法如下。

在eclipse转到服务器选项卡双击注册服务器在我的情况下tomcat 7.0，它打开我的文件服务器一般注册信息。在“一般信息”部分单击“打开启动配置”链接，这将打开在虚拟机参数参数选项卡中添加的服务器选项的执行，最后添加这两个条目

-XX: MaxPermSize = 512m
-XX: PermSize = 512m

和准备好了。

App server PermGen errors that happen after multiple deployments are most likely caused by references held by the container into your old apps' classloaders. For example, using a custom log level class will cause references to be held by the app server's classloader. You can detect these inter-classloader leaks by using modern (JDK6+) JVM analysis tools such as jmap and jhat to look at which classes continue to be held in your app, and redesigning or eliminating their use. Usual suspects are databases, loggers, and other base-framework-level libraries.

参见类加载器泄漏:可怕的“java.lang。OutOfMemoryError: PermGen space”异常，特别是它的后续帖子。