on 03-21-2013 1:21 PM
Dear All,
I am into SAP Basis and new to NetWeaver systems. Recently I came across memory issue, due to memory issue my PRD system got crashed after crash we have restarted complete instance (only Java stack system, single instance with 2 server nodes) and now it is running fine. But I have few questions from business. Reason for crash and analyze the memory usage for 2 weeks.
My questions: how to perform the memory analysis for NetWeaver systems (Basics to advance)? Could you please help me to do the memory analysis
Thanks & Regards
Praveen
Hello Praveen,
At the starting point we can
jstartup/recorder = java -classpath ../j2ee/cluster/bootstrap/launcher.jar com.sap.engine.offline.OfflineToolStart com.sap.engine.flightrecorder.core.Collector ../j2ee/cluster/bootstrap -node %nodeID% %startTime% -bz $(DIR_GLOBAL) u2013exitcode %exitcode%
Please refer the following link
http://help.sap.com/saphelp_sm32/helpdata/de/56/a945412c98f323e10000000a155106/content.htm
Once you set up the flight reciorder you will be able to get the information about the crash.2.
2.Download Eclipse analyser
http://www.eclipse.org/mat/
Search for *hprof in the follwoing directory C:\usr\sap\SID\JCXX\j2ee\cluster\
You will get as a result of this hprof files if it is a memory error
You can analyse the dumps using the memory analyser and try to find the root cause analysis.
SAP note 1515139 - will hep you for BI errors.
Have a great day!
Kind Regards,
Jacob
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hello
I have reffered eclipse tool but it is licensed version I can not install it on client system. is there any way to do manul monitoring and find it out every hour memory usage?
The Memory Analyzer Tool can be installed on a workstation or laptop.
You use this tool to analyse memory leaks.
If your requirement is to monitor the system activity then I suggest you to use the Wily Introscope tool.
Cheers
RB
Hello
If you need to identify why the system crashed then you need to analyse the trace files from the work directory and also the default traces.
If there are OutOfMemory errors in the trace files then it is obvious that there is lack of memory for the server processes to process the requests.
In the event of a OOM the system will terminate the activity handled by the server process and will throw a heapdump at the OS level (mostly in the serverX directory).
You use the dump file to analyse where the leak was using the MAT tool.
Have you found the answers to my first ?
Good Luck
RB
Hello
You need to check how much memory is configured on the box.
Check how much memory is assigned to the Java stack. Check in config tool the memory assigned to both server nodes and the dispatcher node.
What database is used here ? Check how much memory is assigned to the DB.
What JRE are you using on the server ?
Find out these details
Cheers
RB
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
User | Count |
---|---|
78 | |
10 | |
7 | |
6 | |
6 | |
6 | |
6 | |
6 | |
5 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.