Platforms

All HPO, version 16.5 and following


Abstract

Space-separated list of list names for which search-related files should be preloaded


Examples

z/VM:

SEARCH_PRELOAD = 'BIGLIST HUGELIST ANOTHERLIST'

Unix:

SEARCH_PRELOAD="BIGLIST HUGELIST ANOTHERLIST"

export SEARCH_PRELOAD

Windows:

SEARCH_PRELOAD=BIGLIST HUGELIST ANOTHERLIST


Details

The SEARCH_PRELOAD feature is a space-separated list of list names for which you want to preload all the files that might be necessary for the fastest possible search experience.  It is set in the site configuration file in the usual way (see Examples) or can be set in the site configuration section of the web interface.

The enhanced search engine in LISTSERV 16.5 and following is very fast as long as the data is already in memory. Without SEARCH_PRELOAD, if your search encompasses 1,000 text files and a 100MB compressed index, there will be a very noticeable startup time, at least on conventional drives, every time the search is run.  Naturally, running LISTSERV on expensive server-class SSD drives would speed the process up, but by enabling SEARCH_PRELOAD, the average customer gets a $0.00 solution to the problem, and can put all their “free” RAM to good use instead of leaving it unused.

Certain customers have one main list for which they may want to provide the best possible search performance, faster than even Google. With SEARCH_PRELOAD, LISTSERV will load all the relevant files at startup (in the background) and keep them loaded for the fastest possible access.


Default Value

Not set.


Wildcards

Not allowed.  Each list to be preloaded must be explicitly specified.