Javacore log file splitter

The Javacore Splitter uses a filter to reduce the size of the data that is grouped into each record.

Javacore files contain a single timestamp and so are processed as one log record. The Javacore splitter ensures that the contents of the javacore file are grouped into a single log record. Javacore files contain a lot of information on the state of the JVM at the time of the javacore dump. As a result javacore files can be large in size. However, not all of this data needs to be indexed and annotated. The Javacore Splitter uses a filter to reduce the size of this data. The following entries in a javacore file are filtered by the splitter:

  • 1TISIGINFO
  • 1TIDATETIME
  • 1TIFILENAME
  • 2XHOSLEVEL
  • 3XHCPUARCH
  • 3XHNUMCPUS
  • 1CIJAVAVERSION
  • 1CIVMVERSION
  • 1CIJITVERSION
  • 1CIGCVERSION
  • 1CIRUNNINGAS
  • 1CICMDLINE
  • 1CIJAVAHOMEDIR

The entire thread stack trace is also filtered. All other data is blocked by the filter.

Turning off the filter

The filter can be configured to be on or off. It is on by default. To turn off the filter, follow these steps:

  1. Create a file called javacore_insightpack.config
  2. Add the following key/value to the file: splitter.filter.on=false
  3. Save the file and copy it to your home directory on the IBM® Operations Analytics Log Analysis system
Note: Turning off the filter will pass the entire javacore file contents into IBM Operations Analytics Log Analysis. This will affect the performance of searches on the IBM Operations Analytics Log Analysis Search workspace as the entire javacore file contents will be contained in the logRecord annotation.