p_manage_02 takes 24 hours (for 1.3 million records)

• Article Type: General
• Product: Aleph
• Product Version: 19.01

Description:
p_manage_02 takes 24 hours (for 1.3 million records). That seems too long.

It is being run with 6 processes on a server with 16 processors.

Resolution:
This *should* be more like 10-15 hours.

As described in the "How To Run Index Jobs" document, a number of factors affect index run time: number of processes, cycle size, archive logging, expand routines, number of index codes, the sort routine (not so much after 2007), etc.

In this case, the problem was the cycle size: the abc01 prof_library had:

setenv p_manage_02_loop_length 1000

This is too low. As described in HTRIJ, if you have more than 500,000 records, you should specify 50000 as the cycle size / loop length.

Note: The usm01 prof_library has values appropriate for the small number of usm01 records; the values in your own prof_library file, for your own records needs to be increased.

• Article last edited: 10/8/2013