Looking at this more carefully, I have a couple of questions.

First, I am not seeing any way to distinguish brokers from regular clients. 
We have a specific need to authenticate brokers in a different manner than
clients.  We want our brokers to accept connections from end clients but
prevent those clients from connecting their own brokers to the network - in
part to prevent bypass of our custom security.

Second, looking at broker-to-broker connections, is there any potential
security hole with the timing of security placed in a BrokerFilter?  Looking
at TransportConnection's processBrokerInfo(), several steps are taken to
establish additional connections to slave brokers and duplex connections
before BrokerFilter's addBroker() is called.  Even if these are disposed
promptly on denying the broker access, I'm concerned there is time for such
a broker to gain unauthorized access to data.


-- 
View this message in context: 
http://activemq.2283324.n4.nabble.com/BrokerFilter-securing-addBroker-via-SSL-certs-tp3086239p3230476.html
Sent from the ActiveMQ - Dev mailing list archive at Nabble.com.

Reply via email to