1. You need to provide more info
2. No. Locking issues are usually independent of the transport used
3. UDP per se is not reliable, JGroups add protocols on top to make it reliable, e.g. lossless delivery & retransmission, fragmentation, flow control and so on
Here are the details of our current problems and configuration:
1. JBoss Cache Loader throws deadlock error message
2. JBoss Cache doesn't response to any request (i.e. read, write)
1. Deployed in 3 virtual servers
2. Using pessimistic locking REPEATABLE_READ
3. TCP protocol
We are planning to move to UDP protocol if it can solve the issue and of course if UDP is reliable.
What is your cache loader configuration like?