I am planning to cluster GeoServer. I can find several results in the web saying that often clustering is set up by configuring multiple GeoServer nodes which all share the same data directory.
At the same time there's some results ( http://osgeo-org.1560.n6.nabble.com/Using-GeoServer-with-shared-data-directory-td4468628.html ) that say that all GeoServer features may not work with shared data directory.
I tried to look at the actual GeoServer documentation but I could not find specifics about how a shared data directory can be used, is it supported, which features work with it and which dont. Best I could find was "When setting up a cluster of GeoServer machines it is common to share a single data directory among all the cluster nodes" at http://docs.geoserver.org/latest/en/user/advanced/logging.html.
It is not a given that a shared data directory can be used. The nodes could be writing all sorts of files in the data dir, blocking and interfering each other. Without clear GeoServer documentation about it I am hesitant to use this approach in production…
Is there some authoritative documentation on
- How to use shared data directory with clustered Geoserver
- Other aspects of clustering Geoserver (ie. can I simply add a load balancer in front and everything is ok? 3.2 at http://opengeo.org/publications/geoserver-production/ says that, is it correct?)
Best Answer
Juan gave the following link in twitter to more to read about clustering
http://blog.opengeo.org/2013/04/18/geoserver-in-a-clustered-configuration-part-1/
https://twitter.com/jmarinotero/status/356521026298970113