[Koha-devel] db from 3.8.5 to 3.8.10

[Paul]

At 04:46 PM 3/7/2013 -0300, Tomas Cohen Arazi wrote:
>Paul, we've been through 3.8.0 to 3.8.10 without any issues. Doing a full 
>rebuild should make your registers appear. rebuild_zebra.pl is just not a 
>fullrebuild, but an incremental one.

Thanks for the reply. Yes, as I said (below), a full c/l re-index -a -r -v 
and -b -r -v -x solves the problem:
$ su koha
$ cd usr/share/koha
$ ./bin/migration_tools/rebuild_zebra.pl -a -r -v
$ ./bin/migration_tools/rebuild_zebra.pl -b -r -v -x
but I'm trying to avoid an hourly (or more frequent) cronjob on our 
production server (3.8.5.) I've got it set up at 30 min intervals on the 
sandbox (3.8.10) and it's not too painful in low usage, but am truly 
looking for a programmatic solution for "real life."

Although -r is given as "reset", I thought it was equivalent to "rebuild" 
as compared to -z "process_zebraqueue" which is truly incremental. 
Certainly given the CPU time and the "verbose" output, it looks as though 
-r is not just incremental but a complete rewrite. But I'll also freely 
admit that <http://www.indexdata.com/zebra/doc/> is a challenging "read" in 
Koha context.

Best - Paul


>Regards
>To+
>
>
>
>On Thu, Mar 7, 2013 at 3:53 PM, Paul <paul.a at aandc.org> wrote:
>At 12:04 PM 3/7/2013 -0500, Jared Camins-Esakov wrote:
>Paul,
>
>[snip]
>The standard CRON job runs successfully.
>
>[snip]
>A full c/l re-index -a -r -v and -b -r -v -x solves the problem. As I can 
>reproduce this behaviour (by restoring the same dump) is there anything 
>helpful that I could look into? Â Could this be related to Bug 9511? The 
>fix does not appear to work for me, I've just updated bugzilla.
>
>
>This doesn't sound like a bug. When restoring a dump you *must* completely 
>reindex from the command line. Always. There are no exceptions.
>
>
>Sorry, I wasn't clear. After restoring the dump and running 'installer', 
>the -z -x CRON functions with 99% [???] accuracy. What I'm looking at is 
>any possible "similarity" of Zebra behaviour to try and get the CRON 
>working in 3.8.10 which is reported on bugzilla 9511 to be functional for 
>two libraries that have updated from 3.8.5, but does not (for me) resolve 
>the problem of deleted biblios, nor of overwriting an existing biblio with 
>a Z39.50 import.
>
>I've been chasing around after a solution for a couple of days (part time) 
>with no success whatsoever. Maybe I'm grasping at straws?
>
>Our volunteers are adding 150-200 biblios per day, 7/7, many of which are 
>Z39.50 "overwriting" of embryonic data from previously importing 
>spreadsheets into 3.4.6, and regrouping items (particularly serials) under 
>a single biblio. Â I find that I'm asked about once an hour on average to 
>"completely re-index" and am somewhat desperate to find a programmatic 
>solution.
>
>I appreciate any/all suggestions.
>
>Best - Paul