websphere generates heapdump method

Recently, due to the poor performance of the application in the performance test environment (Linux version 2.6.18-92.el5), it needs to be analyzed. Therefore, it is necessary to analyze the heapdump under websphere.

The steps to generate heapdump are as follows:

1. ps -ef|grep java to query the thread ID of the java process.

2, kill -3 thread ID number

3, found that only the javacore file is generated. At

this time, you need to configure and increase the environment variables of the system. Continue as follows:

4, export IBM_HEAPDUMP=true

5, export IBM_HEAP_DUMP=true

6, export IBM_HEAPDUMP_OUTOFMEMORY=true

these It can be entered directly on the command line, but it is only valid for this connection operation.

You can also choose to configure it in ~/.bash_profile, depending on your login shell initial configuration file of course.

7. Restart your application process (this is to let the application reload your configuration variables).

8, kill -3 The thread number started by the new application

9, the file will be generated under your default profiles:

such as: /app/was6/IBM/Webpshere/AppServer/profiles/AppSrv01 In

this way , you can use IBM tools to generate The javacore and heapdump files are analyzed too.