I'm using Solr Java replication with multiple master cores ("at_bat" &
"on_deck"), and a single slave core ("at_bat")
The "at_bat" cores of the master and slave are used for processing search
requests, and the "on_deck" core is used for complete index rebuilds. Once a
rebuild is complete, the "at_bat" core is SWAPped with the "on_deck" core.
The single slave core is configured to poll the master "at_bat" core. When the
swap occurs, the slave detects the version change, and then the entire set of
index files is replicated to a new index.YYYYMMDDhhmmss directory.
Does anyone have advice on this approach; findings, issues encountered, possibly
a way to work around the ever-growing copies of core data directories without
having to use custom cleanup scripts?