Clustering Event mode does not remove cached data

[smithkm]

When the Cluster module is in event mode old data is being retained after notification of an update.

With two nodes sharing a JDBCConfig catalog, update a catalog entry (title of a feature type for instance) then check on the other node, after waiting for the sync delay set in cluster.properties and the title will not have updated. It will update eventually, on the order of minutes.

The ResourcePool which is doing the caching is being notified and is apparently cleaning up its cache in response. There must be additional caching going on somewhere.

[jdeolive]

The jdbcconfig facades cache objects internally to save trips to the database... could be in this case those caches are not properly being invalidated. I would look there.

[smithkm]

Yep, that was it.