TrackerInitialJob - parallel processing

A parallel processing feature has been introduced in TeamForge 8.1 Patch 1 to improve the performance of the tracker initial load job.

A performance observation was conducted on the tracker initial load job when fetching data from TeamForge and loading them to datamart. For more information on the performance observation, see TrackerInitialJob performance observation.

Based on the result, a parallel processing feature has been introduced to reduce the data processing time. For the observation results on parallel processing, see here.

The parallel processing includes the following steps:

In the ETL_JAVA_OPTS site option token available in the installer configuration directory (/opt/collabnet/teamforge-installer/<Product Version>/conf), change the JVM heap size by setting the Xmx value to -Xmx2048m.
Important: To make this a permanent change, you must modify the above setting before installing the product.

If you do not want to make this a permanent change due to hardware resource constraint or for any other reason, then change the value in the set-env.sh file located in runtime/conf directory as below:

Open set-env.sh in <teamforge-installed-base-dir>/runtime/conf directory.
In ETL_JVM_OPTS, change the Xmx setting to -Xmx2048m.
Restart the CollabNet ETL service.
- sudo /etc/init.d/collabnet restart etl
Warning: If the heap size (Xmx) is not set to 2048 MB for the ETL service, there is every possibility of the ETL service throwing an exception pertaining to insufficient memory while running the tracker initial load job.