[TYPO3-Solr] Use replicated Solr servers

Olivier Dobberkau olivier.dobberkau at dkd.de
Wed Sep 26 17:01:01 CEST 2012


Am 26.09.12 16:32, schrieb Peter Beernink:
> Hi,
> 
> I've got a load-balanced/fail-over TYPO3 environment which runs both the
> website as the database in a replicated setup.
> In this environment I want to include the Solr, which also needs to run
> in a multiple-server setup.
> As I'm thinking about the master-slave setup of Solr to accomplish this
> I was wondering whether anybody knows if it is possible to somehow
> configure the TYPO3 Solr extension to use one server for update and
> commits and a different server to perform the actual search?
> This would make it possible to send all updates etc to the master and
> perform the search queries on a 'local' slave.
> 
> Thanks in advance,
> Peter

Hi Peter,

This is not possible at the moment. But as Solr is HTTP you could do
that separation with a proxy setup. ie. with nginx or even varnish.

Our experience is that Solr is very robust and does not need such a high
availaibility setup for normal use cases.

Have a look on the blog from etsy about their way of propagating solr
indexes to get an idea of real workloads.

http://www.slideshare.net/giokincade/solr-etsy-apache-lucene-eurocon-9913208

http://codeascraft.etsy.com/2012/01/23/solr-bittorrent-index-replication/

Have fun!

Olivier



More information about the TYPO3-project-solr mailing list