if you use the failover: transport on the client, it will attempt to remove the old connection on the broker when it reconnects.
The other thing to look at are the tcp keepAlive settings, set keepAlive=true on the broker transport connector and configure the OS to have a lower keepAlive defaults such that the inactive tcp connection is dropped before 60s. With a lower keepAlive, the abruptly closed connection will be cleaned up earlier. The inactivity monitor will help, but not if the broker is blocked in a write. On 14 October 2011 17:22, Bashar Saleh <bas...@ihjez.com> wrote: > [Setup] > ActiveMQ 5.0 > Broker - Ubuntu Linux 10.04 > Client - Java client running on Windows > > > [Problem] > I have an issue with a durable subscription reconnection. When the client > above is disconnected (due to a bad connect, cable disconnection, loosing > connectivity), the client automatically throws an exception and attempts to > reconnect to broker (every 5 seconds). Upon successfully restoring internet > connectivity, the connection requests to the durable subscription are > rejected with exception "Client with id ...... is already connected". The > already connected is the same client which lost connectivity earlier. > > After about 60 second, the broker finally drops the now long gone connection > and the client is finally able to reconnect. > > > Is there a way to instruct ActiveMQ to accept the new client connection > request from the same IP address with the same unique session ID (part of > client ID) as a continuation of the now gone connection? > > Remember, the connection was closed abruptly and not gracefully. > > -Bashar > -- http://fusesource.com http://blog.garytully.com
