OutOfMemory in EP6 SP2

[Former Member]

Hi,

I am running on EP6 SP2 Patch 3 HF4. I have about 100-200 users accessing the EP in an hour. We have start up the EP Dispatcher and EP Server as windows services. We are having a OutOfMemory error everyday and the whole EP would crash (i.e. SAP J2EE Engine Server service stop running). We have to restart the SAP j2EE Engine Server service everyday.

All the SAP Notes regarding OutOfMemory have been read thoroughy and implemented but this memory problem is still outstanding. We do not have any custom iViews in the EP, mostly the iViews from the Business Packages for MSS and ESS.

Here is our service.ini file :

Service_1_JavaPath=D:\jdk1.3.1_10\

Service_1_Name=cluster\server

Service_0_MainClass=com.inqmy.boot.Start

ServiceCount=2

Service_1_Port=5561

Service_0_RootDir=D:\usr\sap\EPPD\j2ee\j2ee_00\cluster\dispatcher

Service_0_Timeout=10000

Service_1_RootDir=D:\usr\sap\EPPD\j2ee\j2ee_00\cluster\server

Service_0_JavaPath=D:\jdk1.3.1_10\

Service_1_MainClass=com.inqmy.boot.Start

Service_0_Name=cluster\dispatcher

Service_1_Parameters=

Service_1_Timeout=10000

Service_1_JavaParameters=-Dmemory.manager=1280M -Xms1280M -Xmx1280M -XX:MaxPermSize=192M -XX:PermSize=192M -XX:MaxNewSize=256M -XX:NewSize=256M -XX:SoftRefLRUPolicyMSPerMB=1 -classpath ".;.\system-lib\boot.jar;.\system-lib\jaas.jar;" -Djava.security.policy=.\java.policy -Dorg.omg.CORBA.ORBClass=com.inqmy.system.ORBProxy -Dorg.omg.CORBA.ORBSingletonClass=com.inqmy.services.iiop.internal.ORB -Djavax.rmi.CORBA.PortableRemoteObjectClass=com.inqmy.system.PortableRemoteObjectProxy -Djavax.rmi.CORBA.UtilClass=com.inqmy.system.UtilDelegateProxy -Djava.library.path=D:\usr\sap\EPPD\j2ee\j2ee_00\os_libs

Service_0_Port=5501

Service_0_Parameters=

Service_0_JavaParameters=-Dmemory.manager=64M -Xmx64M -Xms64M -classpath ".;.\system-lib\boot.jar;" -Djava.security.policy=.\java.policy -Dorg.omg.CORBA.ORBClass=com.inqmy.services.iiop.internal.ORB -Dorg.omg.CORBA.ORBSingletonClass=com.inqmy.services.iiop.internal.ORB -Djavax.rmi.CORBA.PortableRemoteObjectClass=com.inqmy.system.PortableRemoteObjectProxy -Djavax.rmi.CORBA.UtilClass=com.inqmy.system.UtilDelegateProxy

Any help would be greatly appreciated....

Thanks.

Regards,

Mike Lee

SAP EP Administrator.

[Former Member]

Hi,

Sorry to Poke in between, but we too are facing the similar issue in our Portal.

The Systems crashes after throwing the OutOfMemory exception.

The Situation used to occur twice a week,

For this issue, we have upgraded the JDK version on the UNIX server.

after JDK upgrade now its like weekly once.

In short there is a bit improvement in the performance. But the issue has not resolved completely.

So I just want to know are you still facing the issue. If no then what fix was applied.

Regards,

Girish

[Former Member]

Could you post the stacktrace you get when you have an outofmemory exception (should be in the console logs somewhere)?

Alternatively, try making a thread dump as described in note 599539 when you see the memory usage is increasing dramtically.

This sort of question is very well suited for an OSS.

[Former Member]

Hi Mike,

there are two possible reasons for an OutOfMemoryError:

1.The VM cannot allocate memory for a new object because there is no heap available (garbage collector cannot free enough space anymore).

2.A classloader tries to load a new class but cannot do so because the perm space is full.

The OutOfMemoryError exception does not tell you, if 1. or 2. was the reason.

You can add -verbose:gc to your server parameters, so that the garbage collections can be profiled afterwards. The central note for memory issues is 634689

regards, Karsten

[Former Member]

Hi Dagfinn,

I have made a threaddump using note 599539.

I have just collected the thread dump file when the memory usage is increasing.

Any idea how to read the threaddump file?

Any help would be greatly appreciated.

Thanks.

Regards,

Mike Lee

[Former Member]

Hi Karsten,

yes, we have add -verbose:gc to the server parameters. From the gc.log, it was growing steadily upwards when

we collect the gc info. I have read through all the

SAP Notes regarding OutOfMemory and implement the solutions. But I am still having the memory problem after the EP is running for 1 day. Our EP server crash everyday and we no way to know what is the exact problem.

Do hope to have your assistance, please.

THanks.

Regards,

Mike Lee

[Former Member]

Hi Mike,

please have a look at this how to guide:

https://www.sdn.sap.com/irj/servlet/prt/portal/prtroot/com.sap.km.cm.docs/documents/a1-8-4/how to analyze performance problems.pdf

I think the chapters "How to Analyze Full Thread Dumps" and "JAVA VM MEMORY ANALYSIS" sound like the answers to your questions

Regards, Karsten