if this is 5.4.2, add ?updateURIsSupported=false to the failover uri, that should ensure that the list remains static. Is this a 5.3.x client to a 5.4.x broker by any chance?
On 7 February 2011 12:21, ActiveMQ2011User <nils.thi...@unister.de> wrote: > > Hallo, > > we have the following scenario: > * two broker that uses the same MySQL database, the configuration files are > the same > * a client that sends messages to one the active broker with the failover > protocol > > We now use the following failover procotol uri for the sender: > > failover://(tcp://localhost:8009,tcp://localhost:8010) > > We start the first broker and it acquires the lock of the database. The > second broker is also started and waits for the lock. Now we send messages. > We shut down the first broker. The second broker gets the lock and sending > still done. If we restart the first broker and shut downs the second broker, > no reconnection takes place. > > For the MySQL tables we use the InnoDB table type. Also we noticed in the > log messages that the list of available brokers to the client becomes > shorter (so finally only the second broker is in the list). > -- > View this message in context: > http://activemq.2283324.n4.nabble.com/Failover-doesn-t-work-with-two-or-more-brokers-tp3264009p3264009.html > Sent from the ActiveMQ - User mailing list archive at Nabble.com. > -- http://blog.garytully.com http://fusesource.com
