Generally speaking, each repository instance must have its own repository home. Even if you're fully backed by a database, jackrabbit will still use the home filesystem to host its workspace configuration and lucene indexes. These indexes are instance specific and jackrabbit holds a lock on the directory to prevent access by other instances.
So yes, if you have 2 nodes on the same machine, each one must have its own repository home. Stéphane Le Jeu 20 novembre 2008 4:38, seemamani a écrit : > > Stephane, > > > Would the repository home need to differ as well for the 2 instances on > the same WAS node (physical machine)? I'm asking because I saw this post > http://www.nabble.com/Clustering-and-.lock-file-td14179957.html#a14188068 > Here Dominique has mentioned that "This sounds like having two > repositories using the same home directory. The only thing that should be > shared between two clustered instances are: the location of the persistent > storage and the journal. " The repository home is configured to be a > physical folder on the same machine on which both server instances are > running, hence the two instances would typically share the same repository > home and repository.xml. Do I need to try and make the repository home > also configurable based on server name? > > Thanks, > Seema > -- > View this message in context: > http://www.nabble.com/Jackrabbit-clustering-Questions-tp20586637p20594501 > .html > Sent from the Jackrabbit - Users mailing list archive at Nabble.com. > > >
