Memory leak in a client using EJBs deployed in a Bea Weblogic 10.0.0 cluste

Hi all,
We are having a memory leak in a client using stateless EJBs deployed in cluster. The client is a Tomcat 6.0.18 with java 6 but it is reproduced using Tomcat 5 with java 5. The client is calling a Weblogic Server 10.0 making
calls to an EJB deployed in cluster that has two instances installed in two different machines.
The client works fine if we shutdown one of the server instances and so when the client is using only one instance.
Resuming the environment:
Client Side:
1 HP-Itanium machine with HP-UX.
1 Tomcat 6 with java 6 (reproduced with java 5)
Bea Weblogic client (wlclient.jar) for Weblogic 10.0.0
Server Side:
2 HP-Itanium machines with HP-UX
Bea Weblogic Server 10.0.0 installed in both machines
An unique domain
Two Bea instances (one per machines) associated to a Bea Cluster
EJBs deployed in both instances
We have monitored the memory consumed in Tomcat and we have noticed that the VM memory PS OLD GEN grows up permanently when we make tests having the two server side Bea Instances up. We have extended
the memory VM parameters in Tomcat client till 1G and it's only a way to delay the end: the free memory is empty, the GC is not able to free no more byte and the CPU is 100% consumed by the GC work. At the end Tomcat Client
doesn't accept more http petitions and must be restarted.
Besides, we have studied the VM memory in Tomcat using jmap and importing it using Eclipse Memory Analyzer. We have seen some strange memory blocks of several Mbytes that are always growing up and that are stored
under data structures in the package com.sun.corba:
com.sun.corba.se.impl.legacy.connection.SocketFactoryConnectionImpl (4.5Mb)
|
-> com.sun.corba.se.impl.transport.CorbaResponseWaitingRoomImpl
|
-> java.util.Hashtable
|
-> java.util.Hashtable$Entry
|
-> java.util.Hashtable$Entry
-> java.util.Hashtable$Entry
-> java.util.Hashtable$Entry
Has anybody any idea about this problem?
Thanks in advance.

Similar Messages

Memory leak in OCI while using AQ

There seems to be a serious memory leak in the OCI driver (9.2.0.1) when using a JAVA client to dequeue a database queue (Advanced Queuing).
Continuous dequeuing causes the heap memory to increase, and this memory never gets freed which leads me to suspect a memory leak in the OCI components (as the memory allocated for the JVM is constant). The heap memory increases by 3-4 MB after a dequeue of 1000 RAW messages,
Has anyone come across this problem before and if so are there any solutions? Changing to a thin driver is not a solution for me due to other requirements.
I'm using using Oracle client v9.2.0.1 libraries running on Solaris 8.
The source code for my JAVA test client is as below:
/* JAVA dequeue */
package com.ubsw.risk.pce.eventqueues.test;
import oracle.AQ.*;
import java.sql.*;
import oracle.jdbc.*;
public class testRawDequeue {
public testRawDequeue() {
public static void main(String[] args) {
Connection conn = null;
AQSession aq_sess = null;
try {
Class.forName("oracle.jdbc.driver.OracleDriver");
//Use OCI connection
conn = DriverManager.getConnection("jdbc:oracle:oci:@DB_NAME.world","user","password");
conn.setAutoCommit(false);
Class.forName("oracle.AQ.AQOracleDriver");
while(true) {
aq_sess = AQDriverManager.createAQSession(conn);
runTest(aq_sess);
aq_sess.close();
aq_sess = null;
System.gc();
} catch (Exception e) {
e.printStackTrace();
System.out.println(e.toString());
try {
if (aq_sess != null) {
aq_sess.close();
if (conn != null) {
conn.close();
} catch (SQLException sqle) {
public static void runTest(AQSession aq_sess) throws AQException, SQLException
AQQueueTable q_table;
AQQueue queue;
AQMessage message;
AQRawPayload raw_payload;
AQEnqueueOption enq_option;
String test_data = "new message";
AQDequeueOption deq_option;
byte[] b_array;
/* Get a handle to a queue - in aquser schema: */
queue = aq_sess.getQueue ("user", "raw_msg_queue");
System.out.println("Successful getQueue");
/* Creating a AQDequeueOption object with default options: */
deq_option = new AQDequeueOption();
/* Dequeue a message: */
message = queue.dequeue(deq_option);
System.out.println("Successful dequeue");
/* Retrieve raw data from the message: */
raw_payload = message.getRawPayload();
b_array = raw_payload.getBytes();
System.out.println("bytes:" + b_array.toString());
queue.close();
((AQOracleSession)aq_sess).getDBConnection().commit();

This sounds very similar to the memory leak I have in Oracle 9i using Pro*C++. Everytime a connect is made memory appears to leak and it only happens in multithreaded mode not default mode. There is a thread about this under the Oracle C++ call interface. Under 9i it appears to leak about 60K per connect rather than 60 bytes.
Paul

Problem running Java client of EJB deployed on WebSphere

Hi,
I am using websphere studio 5.1.2 with fix pack 3. I have a sample Stateless Session Bean (EJB) deployed and running on websphere.
I wrote a small Java client program as below.
import java.util.Hashtable;
import HelloJavaHome;
import HelloJava;
import javax.naming.InitialContext;
import javax.naming.Context;
import javax.ejb.EJBHome;
import javax.ejb.EJBObject;
public class RemoteConn {
public static void main(String[] ar) throws Exception{
String greeting = "";
try{
Hashtable env = new Hashtable();
//env.put(Context.INITIAL_CONTEXT_FACTORY, "com.ibm.websphere.naming.WsnInitialContextFactory");
//env.put(Context.INITIAL_CONTEXT_FACTORY, "com.ibm.ejs.ns.jndi.CNInitialContextFactory");
env.put(Context.INITIAL_CONTEXT_FACTORY, "com.ibm.ws.naming.util.WsnInitCtxFactory");
//env.put(Context.PROVIDER_URL, "corbaloc:iiop:myhost.mycompany.com:2809/NameServiceCellPersistentRoot");
env.put(Context.PROVIDER_URL, "iiop://localhost:2809");
System.out.println("Before creating context ");
InitialContext initContext = new InitialContext(env);
System.out.println("Before looking up HelloJavaHome.");
//Object obj = initContext.lookup("java:comp/env/ejb/HelloJavaHome");
Object obj = initContext.lookup("ejb/HelloJavaHome");
System.out.println("After lookup javahome.");
HelloJavaHome home = (HelloJavaHome) javax.rmi.PortableRemoteObject.narrow(obj, HelloJavaHome.class);
HelloJava remote = home.create();
greeting = remote.getGreeting("John");
catch(Exception e){
e.printStackTrace();
System.out.println("Greeting::"+ greeting);
I did not have any problem if I write and run this client from Websphere studio environment..
However its giving lot of compilation and runtime errors when I tried to run from command line.
I could eliminated compilation errors by setting the below jar files in the classpath.
testclient.jar; (client jar of my EJB)
j2ee.jar;
naming.jar;ras.jar;
wsexception.jar;
bootstrap.jar;
namingclient.jar;
websphere.jar;server.jar;
ejbcontainer.jar;
ecutils.jar;
Now it is throwing runtime error after the statement, "Before looking up HelloJavaHome.", has printed. Here is the stack trace of exception ..
Exception in thread "main" java.lang.NoClassDefFoundError: com.ibm.CORBA.iiop.GlobalORBFactory
at com.ibm.ejs.oa.EJSORBImpl.class$(EJSORBImpl.java:44)
at com.ibm.ejs.oa.EJSORBImpl.initializeORB(EJSORBImpl.java:195)
at com.ibm.ejs.oa.EJSClientORBImpl.(EJSClientORBImpl.java:93)
at com.ibm.ejs.oa.EJSClientORBImpl.(EJSClientORBImpl.java:65)
at com.ibm.ejs.oa.EJSORB.init(EJSORB.java:385)
at com.ibm.ws.naming.util.Helpers.getOrb(Helpers.java:284)
at com.ibm.ws.naming.util.WsnInitCtxFactory.getInitialContextInternal(WsnInitCtxFactory.java:369)
at com.ibm.ws.naming.util.WsnInitCtx.getContext(WsnInitCtx.java:112)
at com.ibm.ws.naming.util.WsnInitCtx.getContextIfNull(WsnInitCtx.java:422)
at com.ibm.ws.naming.util.WsnInitCtx.lookup(WsnInitCtx.java:143)
at javax.naming.InitialContext.lookup(InitialContext.java:347)
at RemoteEJBConn.main(RemoteEJBConn.java:38)
I appreciate if sombody could help me ASAP fixing this problem.
Thanks in advance.

Try using the IBM JVM (theres one shipped with WebSphere in websphere_home\AppServer\java
Also include idl.jar and ffdc.jar on the classpath

Jdk 1.2 clients running ejb deployed on oracle 8.1.5

Oracle claims that it is possible to run jdk 1.2 clients with ejb on oracle 8.1.5 in the faq : http://technet.oracle.com/products/oracle8i/htdocs/jserver_faq/815faq0012.html
If you have JDK1.2 on the client side and:
8.1.5 clients connecting to an 8.1.5 server: This does not work, because of conflicts with the ORB supplied in JDK1.2.
8.1.5 clients connecting to an 8.1.6 server: This does not work, because of conflicts with the ORB supplied in JDK1.2.
8.1.6 clients connecting to an 8.1.5 server: This works provided the "server" portion of the application is developed/loaded using the client side tools and jars that are supplied with 8.1.5.
What do they meen with 8.1.6 clients? and how to make 8.1.6 clients?

Oracle 8.1.6's JVM uses JDK 1.2. Oracle's 8.1.6 JDBC drivers support JDBC2 and JDK2. The 8.1.6 drivers are available for download from OTN under the Technologies section.
If you develop a client (application, applet, servlet, JSP, etc.) using the Oracle 8.1.6 JDBC drivers, then this is considered an Oracle 8.1.6 client.
Laura

Will the memory leak for queue when used in producer and consumer mode in DAQ to transfer different sized array.

In the data acquisition, I use one loop to poll data from hardware, another loop to receive the data from polling loop sent by queue.
But everytime the size of the transferred data array may not be the same, so the system may assign different array size and recycle very frequently.
Will it cost memory leak. Or will it slow down the performance, since the array size is not fixed, so every time need to create a new sized array.
Any suggestion or better method.
Solved!
Go to Solution.

As i understand your description, your DAQ-loop acquires data with the setting '-1' for samples to read at the DAQmx read function. This results in the different array sizes.
Passing those arrays directly to a queue is valid and it does not have significant drawback in performance (at least as far as i know) and it definetly does not leak memory.
So the question is more or less:
Is it valid that your consumer receives different array sizes for analysis? How does your consumer handle those arrays?
hope this helps,
Norbert
CEO: What exactly is stopping us from doing this?
Expert: Geometry
Marketing Manager: Just ignore it.

Memory leaks in MFC while using CDatabase::OpenEx()

Hi,
This question has been asked previously but the explanation was not in detail and i could never reach to the bottom of it.
I hope i can elaborate more on the problem statement and i can get a resolution/explanation from the experts here.
Consider the following two sample codes;
1.
int count = 200;
for (int i=0; i <count; ++i)
try
CDatabase *db = new CDatabase;
BOOL bRes = db->OpenEx(_T("DSN=MyData;UID=anon;PWD=pass"));
db->Close();
delete db;
db=NULL;
catch (CDBException* e)
e->Delete();
2.
CDatabase *db = new CDatabase;
int count = 200;
for (int i=0; i <count; ++i)
try
BOOL bRes = db->OpenEx(_T("DSN=MyData;UID=anon;PWD=pass"));
db->Close();
catch (CDBException* e)
e->Delete();
delete db;
db=NULL;
return 0;
delete db;
db=NULL;
The first sample code leaks a lot of memory and it can be easily observed from the task manager, the memory usage keeps on growing.
The second sample code does not leak any memory if i observe the memory usage from the task manager.
To find out the cause of the memory leak i ran the code through rational purify, both the codes are leaking memory according to rational purify. The first code leaks significantly more memory than the second one. The DLL pointed by rational purify are MFC
DLLs (inserting the screen shot below)
Is this a known issue with the MFC DLL or am i doing something wrong?
I have a server application where i have to create CDatabase object multiple times and i end up leaking a lot of memory over a period of time.
I can provide more information about this issue if required. Thanks in advance.

I am trying to reproduce this issue on my side, but it seem there is no memory leak in my simple sample. I use _CrtDumpMemoryLeaks() to test the memory leak.
https://msdn.microsoft.com/en-us/library/d41t22sb.aspx
If I comment this line: db->Close();
I can detect the memory leak at CDatabase
*db =
new CDatabase;
See the screenshot.
If follow your sample code, there is memory leak message in the output view pane.
#define CRTDBG_MAP_ALLOC
#include <stdlib.h>
#include <crtdbg.h>
#ifdef _DEBUG
#define new DEBUG_NEW
#endif.....int CMFCCdbMLTestApp::ExitInstance()
// TODO: Add your specialized code here and/or call the base class
_CrtDumpMemoryLeaks();
return CWinApp::ExitInstance();
May you can use some use tool like WinDbg tool to find more information about this issue.
We are trying to better understand customer views on social support experience, so your participation in this interview project would be greatly appreciated if you have time. Thanks for helping make community forums a great place.
Click
HERE to participate the survey.

Memory leaks in NSURLConnection when using with ssl.

Hi,
I am experiencing a memory leak while using nsurlconnection with https request.
I have a web-service over http and as well as HTTPS. When i call the web-service with HTTP i don't get any memory leak while when I call the web-service over HTTPS i got memory leaks, responsible library for this memory leak is cfnnetwork and responsible caller is nsurlrequest::setsslrequest.
I just wanna make sure that should we have to set any property in nsurl for HTTPS request..
I will be really helpful If someone can help me out.
thanks in advance.

I get the same(Or similar) issue... 48 Bytes for every call is being leaked. Instruments says that it's NSURLConnection that's leaking an NSString somewhere... Blowed if I can find why, build and analyse doesn't complain about that piece of code (Except that I'm releasing the url request late... i.e. by the caller instead of the method that allocates it. Naughty I know, but autorelease'ing the object had the same problem).
This is on the simulator BTW...
H

Client using ejb-ref to lookup

Hi,
     I read the docs, I searched for answers but there are still some things I don't
understand regarding a stand alone client application to a weblogic server 8.1
that could look up a bean using ejb-ref. I don't want to use the JNDI name.
     There is the client-application.xml and the MyClientJar.runtime.xml but i'm not
sure I understand where to put them. I have the client-application.xml in my
client jar, the runtime.xml one in the same directory like specified in the
documentation, I tried to put the xml and the jar in my classpath and running
the client, no success the client cannot lookup using ejb-ref
     I read that MyClientJar.jar should go in my ServerModule.ear so I did that too,
still doesn't work.
     Something I'm not doing right or something I don't understand.
     Any help or comments appreciated
     Thank You
     Max

Hi,
     I read the docs, I searched for answers but there are still some things I don't
understand regarding a stand alone client application to a weblogic server 8.1
that could look up a bean using ejb-ref. I don't want to use the JNDI name.
     There is the client-application.xml and the MyClientJar.runtime.xml but i'm not
sure I understand where to put them. I have the client-application.xml in my
client jar, the runtime.xml one in the same directory like specified in the
documentation, I tried to put the xml and the jar in my classpath and running
the client, no success the client cannot lookup using ejb-ref
     I read that MyClientJar.jar should go in my ServerModule.ear so I did that too,
still doesn't work.
     Something I'm not doing right or something I don't understand.
     Any help or comments appreciated
     Thank You
     Max

MEGA memory leak!! cannot use PS ideas please?! deadline looming!!

Hello PS community!!
I'm running PS 5.5 on a Macbook pro Retina version 10.7.5
Despite having more than 30gb free on my disk, as soon as I open my file (Doc 199.1M/655.6M) within 1 minute, all 30gb is consumed leaving just 10MB and I cannot work anymore after the 'scratch disks are full message appears' then as soon as I close PS, all of the memory is restored!
The scratch disks are set to the default macHD. like they always have been with no problems and I'm working in other programs such as Premier pro with huge files and there's no problem.
It's like there's a black hole only for PS, please help if anybody has any ideas I have a deadline loomig and I can't do anythung in PS at the moment!
Many thanks
ALLAN.

Thanks c.pfaffenbichler
Yes it's a top spec retina 15' with a solid state drive.
My setup has always been sweet and worked brilliantly this is a new problem, i was handling huge 1.5gb ps.files with no issue and haven't changed any settings since then.
my HD is just much fuller now.
Around the same time the message began popping up when I open Premier pro that "The scratch disks are write protected or unavailable. To open this project, thescratch disks will be set to your Documents folder. Would you like continue?"
(Then I just click yes and it all works no problem)
I suppose these issues are related.
I don't generally use smart objects but that's never been an issue with regards to extra weight of files
Cheers

Inject EJB using @EJB in Servlet Filter on Weblogic 11g

Hi All,
I want to inject the EJB (Local interface) into the Servlet Filter and the EAR is deployed on Weblogic 11g.
My question is:
Shall the @EJB Annotation work on Weblogic 11g or it will be ignored in case of Servlet or Servlet Filter?
OR
I have to do look up as below and mention the references in web.xml and weblogic xml file:
I know below code should be used when you have remote interface.
Hashtable env = new Hashtable();
env.put( Context.INITIAL_CONTEXT_FACTORY, "weblogic.jndi.WLInitialContextFactory" );
Context ctx = new InitialContext( env );
ctx.lookup( "myejb" );
Thanks

Hi,
It should work in 11g.
Regards,
Kal

Writing a Standalone Client for EJB 3.0 Bean For Weblogic 10

Steps for Writing a Standalone Client
Client Code
package com.client;
import java.util.Properties;
import javax.naming.Context;
import javax.naming.InitialContext;
import javax.naming.NamingException;
import javax.rmi.PortableRemoteObject;
import com.StatelessService;
public class StatelessClient {
     public static void main(String [] args){
          Context ctx = null;
          try {
               ctx = getInitialContext("t3://localhost:7001/protossbservice","weblogic","weblogic");
               Object ref = ctx.lookup("StatelessService#"+StatelessService.class.getName());
               StatelessService service = (StatelessService)PortableRemoteObject.narrow(ref,StatelessService.class);
               System.out.println("StatelessClient.getMessage() ++");
               String message = service.getMessage();
               System.out.println("StatelessClient.getMessage --");
                    System.out.println("Message from Bean "+message);
     } catch (NamingException e) {
               e.printStackTrace();
     private static Context getInitialContext(String url, String user, String password) throws NamingException {
        Properties h = new Properties();
        h.put(Context.INITIAL_CONTEXT_FACTORY,
                "weblogic.jndi.WLInitialContextFactory");
        h.put(Context.PROVIDER_URL, url);
        h.put(Context.SECURITY_PRINCIPAL, user);
        h.put(Context.SECURITY_CREDENTIALS, password);
        return new InitialContext(h);
Business Interface
package com;
public interface StatelessService {
     public String getMessage();
Stateless Session Bean
package com;
import javax.annotation.PostConstruct;
import javax.ejb.Remote;
import javax.ejb.Stateless;
* Test Stateless Session Bean
@Stateless(mappedName="StatelessService")
@Remote(StatelessService.class)
public class StatelessServiceBean implements StatelessService {
     public String getMessage(){
          return "I am invoked";

emekaco wrote:
Hi,
I am a trainee although with a good background in java.
I'm urgently in need of a free ebook for EJB 3.0 with the latest version of Netbean. Pls, help me with one if you have. my email is [email protected]
http://www.theserverside.com/tt/books/wiley/masteringEJB3/index.tss
Here's a good free e-book but without Netbean. It covers everything you would need to know about EJBs.
>
Again, is container managed bean good for a big project?
The question is too vague to be answered. It's a design decision that needs more details.

Deploying MDB on BEA WebLogic 7.0 to listen to Oracle AQ

Hello All:
I have been tasked to create a MDB that will listen to a queue in AQ. What I am curious about, however, is if anyone has ever had any experience in configuring the WebLogic server so it can interface with, and listen to, AQ?
I am going to create a concrete sender class to handle the transmission of messages inside the network, but the design goal, for scalability purposes, is to use a MDB to listen to the queue. We cannot use any Oracle Application Server or Web Server per my client's directives.
Please help. Thanks.

I am very interested in viewing your information on Oracle AQ and BEA 8.1 but the link you provided,
http://marvel.oracle.com/pls/otn/f?p=17000:8:25148205767784132
displays an error page with the following message:
ORA-01400: cannot insert NULL into ("DEVTRENDS"."WWC_ASK_QUESTION_ACCESS_LOG$"."DISPLAYID")
I would really appreciate some help I have been hunting for documentation on how to access AQ from a Java. I am trying to write a JCA adapter for BEA WLS 8.1 to access the AQ for integration purposes.
Thanks

Errors encountered when trying to use a Custom Realm on BEA Weblogic Platform Domain .....

Hi,
We have created a WebLogic Platform Domain. A WebLogic Portal application(Portal
7.0) and some Web Service apps are running on this domain.
We have created a Custom Security Realm b'cos of our application requirements
and now when I startup the Platform Domain, I see lot of errors.
Some of the errors typically are
"<Jan 16, 2003 4:07:02 PM EST> <Error> <HTTP> <101256> <The run-as user: wlisystem,
for the servlet: ApplicationView for the webapp: /WLI_AI_Workshop_Control_Web,
could not be resolved to a valid user in the system. Please check if the user
exists.
javax.security.auth.login.LoginException: Authentication Failed: User wlisystem
denied in Realm Adapter realm weblogic"
or
Unable to deploy EJB: wlai-eventprocessor-ejb.jar from wlai-eventprocessor-ejb.jar:weblogic.ejb20.WLDeploymentException:
weblogic.ejb20.interfaces.PrincipalNotFoundException: Authentication Failed: User
wlisystem denied in Realm Adapter realm weblogic
Do we have to create any predefined user accounts in the Security Store to get
rid of these errors. I would appreciate if anyone can suggest some tips or workarounds
for configuring or creating a Custom Security Realm for Web Logic Platform Domain.
Thanks
Vikram

with 7.0 try to create system in your custom realm.. that may help.
-kiran
"Vikram" <[email protected]> wrote in message
news:[email protected]...
>
Kiran,
First time around, the Custom realm was not authenticating the user. I gotthe
code to authenticate the user successfully. Now the WebLogic serverwouldn't even
start. It would give me an error message which says "User System is notauthorized
to boot Weblogic Server". For your reference, I am attaching the Log file.My
custom realm classes output some debugging statements in the log file.From the
log file u will see that the users are getting authenticated successfully.
Please let me know if you have a custom realm working for you. I might bemissing
something.
Appreciate your help.
Thanks
Vikram
"kirann" <[email protected]> wrote:
does your realm able to authenticate user "wlisystem".
thanks
kiran
"Vikram Datla" <[email protected]> wrote in message
news:[email protected]...
Hi,
We have created a WebLogic Platform Domain. A WebLogic Portalapplication(Portal
7.0) and some Web Service apps are running on this domain.
We have created a Custom Security Realm b'cos of our applicationrequirements
and now when I startup the Platform Domain, I see lot of errors.
Some of the errors typically are
"<Jan 16, 2003 4:07:02 PM EST> <Error> <HTTP> <101256> <The run-asuser:
wlisystem,
for the servlet: ApplicationView for the webapp:/WLI_AI_Workshop_Control_Web,
could not be resolved to a valid user in the system. Please check ifthe
user
exists.
javax.security.auth.login.LoginException: Authentication Failed: Userwlisystem
denied in Realm Adapter realm weblogic"
or
Unable to deploy EJB: wlai-eventprocessor-ejb.jar fromwlai-eventprocessor-ejb.jar:weblogic.ejb20.WLDeploymentException:
weblogic.ejb20.interfaces.PrincipalNotFoundException: AuthenticationFailed: User
wlisystem denied in Realm Adapter realm weblogic
Do we have to create any predefined user accounts in the Security Storeto
get
rid of these errors. I would appreciate if anyone can suggest sometips or
workarounds
for configuring or creating a Custom Security Realm for Web Logic
Platform
Domain.
Thanks
Vikram

HUGE memory leak when using MP3

I downloaded a new FME 2.0 and noticed they added an option
to stream in MP3 format. DO NOT USE that mode (!). If you start
streaming in MP3 format you will get a huge memory leak on the
client side. Your browser starts eating memory like crazy and it
will bury your machine withing an hour or two by taking all the
memory available. I have proved it on several machines, notifed
Adobe. They confirmed it as a bug and promised to fix in the future
releases. Nice, huh? and what people are supposed to do with the
current build?

A new build 2.0.1.1114 has been posted . Please try and let
us know if you face this issue with this build.

Best Practices for EJB--memory leak avoidance

I an researching a memory leak. From what I can tell, we do not want to put any instance variables at the class level. This would incur a memory leak when the ejb containers creates the ejb.
I am wondering if there are any more 'best practices' with EJBs that prevent memory leaks. A good site would help me too
Russ

Thank for your reply.
You are right. I was referring to stateless session
beans.
What I was thinking was the following:
1. Making parameters final unless the parameter
r state is changed inside a method.Won't help with memory leaks. There are benefits for an object being immutable, but I don't think that it eliminates the possibility of the object being leaked.
2. All instance variables in stateless session beans
s should be set to null upon ejbPassivate and
ejbRemove operations. The variables should be
populated on ejbCreate and ejbActivate.Optimizing compilers don't need the hint of setting the reference to null. I don't think that helps, but I'm not 100% certain.
3. Before throwing an exception at the session bean
level, clear up all the instance variables by setting
them to null.Shouldn't scope make it clear to the GC that the objects aren't needed? If they're all primitives, how does this help memory leaks?
I am using a profiler to identify the objects still
being held onto once GC, but I was hoping others with
more experience than I would share some of their tips
on how they avoid memory leaks.I don't know where you're getting these tips, but I don't think they're helpful.
Collections and Singletons would be the places where I'd look. If you add a reference to an object that holds onto it in a collection and never lets go, the GC won't reclaim it.
Singletons aren't cleaned up, because their instance is static. Anything the Singleton refers to won't be cleaned up unless the Singleton relinquishes the reference.
Look for things like that. I think they matter more.
%

Memory leak in a client using EJBs deployed in a Bea Weblogic 10.0.0 cluste

Similar Messages

Maybe you are looking for