Ok. I managed to fix the "pure ssl transport" problem. And, apparently, the abnormal memory consumption usage with nio+ssl was related to this ssl misconfiguration! I'll try to explain. Until today, I used to configure the SSL context AFTER creating the transports, namely:
This apparently works well with NIO+SSL EXCEPT when I try to connect a CMS client: in this case, the abnormal memory consumption is triggered. With pure SSL, the code above simply does not work and the broker cannot find the correct keystores (I saw this enabling the javax.net.debug=all). The point is that once I fixed the code above (postponing the transport connector creation AFTER the ssl context configuration) I got both the pure SSL and the NIO+SSL scenarios up and running and the memory issue with NIO+SSL disappeared! I'm a little bit puzzled about this behavior, indeed. Does this make sense to you? Thank you! matteo -- View this message in context: http://activemq.2283324.n4.nabble.com/Abnormal-NIOTransport-memory-usage-tp4665860p4665885.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.