2010/8/2 Brice Sanchez <brice.sanchez@sys-tech.net>:
  

Thank you for your reply,

I've already done several installations of Koha in DEV and Single mode. But
when we have multiple installations on one server we have problems with the
Zebra indexing engine.

This is why we try to virtualize (with Xen), or to install a single Koha
running with several databases.

We are therefore looking for feedback and help to build the right
architecture : that is the most easily maintainable for 20 School and  can
easily accommodate 500 simultaneous users.
    

We've successfully installed multiple instances within the same server
using the following schema: each instance is called koha_<instance
name>. Then in an standard install when you're asked for the install
dir you put koha_<instance name> and the koha installer creates:

/etc/koha_<instance name>
/usr/share/koha_<instance name>
/var/lib/koha_<instance name>
etc...

You can then set a shell variable:

# export KOHAINSTANCE=koha_<instance name>

and run the following for finishing the setup:

# ln -s $KOHAPATH/bin/koha-zebra-ctl.sh /etc/init.d/$KOHAINSTANCE-zebra-daemon
# update-rc.d $KOHAINSTANCE-zebra-daemon defaults
# ln -s /etc/$KOHAINSTANCE/koha-httpd.conf
/etc/apache2/sites-available/$KOHAINSTANCE
# a2ensite $KOHAINSTANCE
# apache2ctl reload

Assuming you're running on Ubuntu/Debian.

You then have to fix your cronjobs to match this schema. Probably
having to prepend KOHA_CONF=/etc/koha_<instance name>/koha-conf.xml to
your commands, for example:

If you want to rebuild the zebra database of an instance:
su - koha -c "KOHA_CONF=/etc/$KOHAINSTANCE/koha-conf.xml
/usr/share/koha_saludpublica/bin/migration_tools/rebuild_zebra.pl -a
-b -w -r -v"

Or put an incremental reindex in the cronjob:

0/2 * * * *    root         KOHA_CONF=/etc/$KOHAINSTANCE/koha-conf.xml
/usr/share/koha_saludpublica/bin/migration_tools/rebuild_zebra.pl -a
-b -z


If you have trouble with this setup, feel free to contact.

To