-
Bug
-
Resolution: Done
-
Highest
-
None
-
None
-
None
If a receiver is incapable of keeping up with a sender's stream of messages, and the RMR receive thread does not get enough cycles to dequeue and drop messages, the RMR code on the sender will incorrectly disconnect and start a new session. If the processes are long lived this causes an excessive number of TCP sessions which leads to a sender crash.
In some cases there seems to be a memory leak associated with this behaviour.