Hi, I have the same experence when using CORBA and the provider is Borland Visibroker. When I new a instance ORB, there will be three threads form by Java. If you never release the ORB, the threads never died. The solution is using singlton to avoid that.
Your problem may not in the Memory. Maybe it's the thread problem. Hope this can help. Jacob FanChiang Assoiciate Researcher Customer Care and Billing System Dept. Telecommunication Laboratories Chunghwa Telecom Co., Ltd. Tel: +886 2 3707 1098 Fax +886 2 2325 5844 Email: [EMAIL PROTECTED] ----- Original Message ----- From: "Matt Fisher" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Thursday, September 27, 2001 9:05 AM Subject: OutOfMemoryError from Tomcat 3.2 > Hello, > > I have an application running on a Windows2000 server, using Tomcat 3.2, Apache > 1.3.20, and a SQL Server 2000 database. This application will be going live next > week. The initial user base will be approximately 20, but ultimately we will > have a user base of 100+. > > During our stress testing today (approximately 20 simultaneous sessions), we > started getting a lot of errors (java.lang.OutOfMemoryError) on the server. The > errors are logged in Tomcat's jvm.stderr logfile. They started occurring during > the stress test, and stopped occurring thereafter. > > I have read that on Solaris, Tomcat 3.2 may get OutOfMemory errors if there are > too many sessions at one time. Our server is running Windows 2000, but I wonder > if our problems also stem from too many sessions. If so, is there a parameter we > can tweak, a fix for Tomcat 3.2, or a new release which addresses this issue? > > During a previous incident where OutOfMemory errors were occurring, I checked > the Windows 2000 performance monitor. Memory usage was approximately 10% of the > 1 gigabyte capacity. It strikes me as odd that we'd get OutOfMemory errors, > unless the JVM only has access to a small portion of the machine's memory. > > We also saw "index out of bounds" errors during the stress test, but I am > guessing that this is somehow related to the memory errors. > > Any help you can provide would be appreciated. > > Yours, > > - Matt Fisher > . . . . . . . . . . . . . . . > Matthew J. Fisher > Applications Programmer > University of Chicago Press > 1427 E. 60th Street > Chicago, IL 60637-2954 > Tel 773-834-5386 >
