activemq-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From atrieger <atrie...@trieger.org>
Subject How do I gracefully fail back from Slave to Master ?
Date Thu, 31 Jul 2008 01:07:57 GMT

Activemq 5.1
Suppose a master with a slave and persistence enabled, all queues.
Messages in transit exist in the disk-based stores of each broker.
Master crashes, fine, slave has all state mirrored, clients pause, reconnect
to slave due to failover:() transport configuration on their end.
Now I know you cant fail back without first shutting down the slave, copying
the state back (rsync) and starting the master...
but what I'd like to do is not perform this operation with messages trapped
midstream in the message store of the slave... because this could be part of
a network of brokers.

I'd like to stop 'something' in the slave (after its the only broker up now,
so its the master really), like its acceptor? via mbeans? so that it cannot
take any new messages... but not prevent it continuing to deliver its
messages until all its queues are empty.

Now it has no clients and is idle, except possibly still has consumers that
are messaged from the other network of brokers, but it should have no
directly connected producers and therefore should own no messages.

Now I can shutdown this slave, and its consumers will failover to other
brokers in the network of brokers.  I have time to rsync state to the
original master, start the original master and start this slave.

Does Activemq support this using persistence in local kaha?  (not jdbc)
-- 
View this message in context: http://www.nabble.com/How-do-I-gracefully-fail-back-from-Slave-to-Master---tp18745898p18745898.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.


Mime
View raw message