Repository initialization errors in clustered environment

When running a typical setup of CMS and HST site, each in their own Tomcat instance, configured as cluster (embedded repository), and using the same (empty) MySQL database schema:

After first starting the "CMS Tomcat" and letting it start up, then starting the "Site Tomcat", the latter logs many "Unable to deliver node type operation" errors, e.g.:

16.12.2010 14:42:23 ERROR [org.apache.jackrabbit.core.cluster.ClusterNode.process():937] Unable to deliver node type operation:

resource already exists

(all relevant errors in log attached as hippo-site.log)

It seems that the CMS when it first start, creates the database tables in MySQL, and initialized the repository with all the required namespace and node types. Then when the site starts, its embedded repository, acting as cluster node, will try to catch up with whatever is logged in the repository journal , including the namespace and node type initializations. These will then fail since they already have been initialized.

The errors are in this case "harmless" from a technical point of view, however they cause much confusion with developers and sys admins not aware of what is going on.