I wrote a small script to replace zebraqueue_daemon, which instead calls rebuild_zebra -a -b -z ever few minutes as a stopgap. A better approach would be a perl script that check the queue every n seconds (e.g. 15 seconds) and then call rebuild_zebra -a -b -z if the queue is non-zero in length. I'll try to pull that together and post it here later. this approach should be both timely, cpu efficient (the check will cost almost nothing) and not have any memory problems. Is there any reason that the script should not also clean done entries from the zebra queue? Keeping the queue short makes the test for unprocessed entries cheap. -Doug- 2009/12/1 Magnus Enger <magnus@enger.priv.no>
2009/11/29 Jesse <pianohacker@gmail.com>:
2009/11/29 Doug Kingston <dpk@randomnotes.org>
Also, can someone comment on the time taken by rebuild_zebra.pl vs the older zebraqueue_daemon? We have around 13K records.
zebraqueue_daemon works well (in my limited testing) and has little lag, but due to some problems with code organization in Koha as a whole, it leaks memory pretty steadily. rebuild_zebra.pl in a cronjob is the approach of most Koha users.
Does that mean
http://git.koha.org/cgi-bin/gitweb.cgi?p=Koha;a=blob_plain;f=INSTALL.debian-... should be updated? It seems to recommend both setting up the zebraqueue_daemon and running rebuild_zebra.pl from cron?
Regards, Magnus Enger libriotech.no _______________________________________________ Koha mailing list Koha@lists.katipo.co.nz http://lists.katipo.co.nz/mailman/listinfo/koha