• Article Type: General
• Product: Aleph
• Product Version: 20

Description:
The "How To Run Index Jobs" document says, in regard to p_manage_01 and p_manage_02: "having too many (extraction) processes running at the same time will cause the extraction process to 'run ahead' -- at the expense of the subsequent single-threaded steps."

How can we tell if these processes are running ahead?

Resolution:
When the job is running, you will see that column 1 of the p_manage_01_a.cycles file is way ahead of the other columns.

If you are looking at the log after the job is run, you can calculate how long it was before the p_manage_01_d step finished after the p_manage_01_a steps (the extraction steps) finished.

If this is less than 10% of the total run time, then the extraction steps are not running ahead and you may want to consider increasing the number of processes specified in the job submission.

• Article last edited: 10/8/2013