[Koha] How frequently to run zebra rebuild and what switches

Paul paul.a at aandc.org
Wed Aug 21 10:10:11 NZST 2013


At 11:49 AM 8/20/2013 -0700, Olugbenga Adara wrote:
>[snip] how often should
> > the zebra_rebuild.pl script be executed and with exactly which switches?
> > Right now I run it at 7:30 AM each morning with "-run-as-root -b -r
> > -v"
>I am quite surprised you need to do that since you say you installed from 
>Debian packages. All the installations I have done through the Debian 
>packages always have the zebra incremental indexing properly set in cron 
>and it is usually set to run every 5 minutes.

Quite apart from running it as root (and if it's automated, no need for the 
-v), the normal cron job is *incremental* (-z) and this user is running a 
*rebuild* (-r) every morning, which is absolutely necessary if any 
cataloguers have deleted a biblio, overwritten a "fast add" with a Z39.50, 
changed the 001/003 combination, etc etc.

We only run ./bin/migration_tools/rebuild_zebra.pl -b -r -v -x and (less 
frequently) ./bin/migration_tools/rebuild_zebra.pl -a -r -v manually and at 
the request of our cataloguers. If we've got a couple of conscientious ones 
playing catch-up on older biblios imported from spreadsheets or fast-adds, 
this can be as often as every 20 minutes (but it's trivial to do.)  One of 
my "to do" jobs is to incorporate a script into the staff-admin page so 
that certain (senior) cataloguers can run this without calling me.

The cataloguers cannot work without it -- we insist that every "entry" is 
verified in the OPAC at the time of cataloguing to ensure that what our 
users see is what the cataloguers intended. Hence the need to run -r fairly 
promptly. The search function down-time, while zebra is "cleaning", is only 
a couple of seconds, particularly if you clean the logs, sessions, imports 
and zebraqueue every 24 hours.

Best - Paul



More information about the Koha mailing list