We are having strange issues with UDP based cache clustering, suddently at some point of time replication totally gets stopped between the nodes. Currently we deployed jboss-cache for HTTP-Session for non-sticky application so we are getting trouble at many of times in the day.
Also we tried to migrate with TCP based clustering, but with TCP based clustering we are facing two problems.
1. During node-2 start-up its not picking up the data available from node-1
2. When bringing down the node-2 server, all the cache entries created by node-2 were getting removed from node-1.
Above two cases were not happening in UDP based clustering. Please advise urgently.