Edson,
> Am I correct on supposing it is possible to have 2 bakckups for one master?
You can have as many backup nodes as you want or need.
But each master node will accept only one backup node during operation.
The second backup node will remain idle until the first backup node is somehow
gone.
the brokers need to be in sync from the start. there is a waitForSlave
attribute that can force the master to stall till the slave connects. see:
https://issues.apache.org/activemq/browse/AMQ-596
2009/7/29 brendanr brendan.rona...@gmail.com
Hi,
I am having an issue with a simple master/slave
Thank you for your reply.
I know about the 'waitForSlave' attribute, but it is not really what I need.
I would like to be able to add a backup machine (and hence a slave broker)
at some later time, without having to restart the master broker and all the
producer/consumer connections to it. I
We also plan to use master/slave.
The 'waitForSlave' seems a bit strange for me, since our main reason for
adding some slaves is to avoid a single point of failure (i.e. the
master). But what if the slave fails? Does the required 'waitForSlave'
mean that we can't run a master without his
pure master slave is handy for data replication but is lock step rather than
catch up, is limited to a single pair and has no auto recovery strategy.
Master slave with a shared file system or shared database is what you want
for fault tolerance. The first to obtain a lock is a master and the next
You may need to look at share file system or JDBC master slave
configuration.
The group clustering type replication with automatic group additions is
something that will be addressed in the 6.0 time frame. Some initial support
for replicating persistent messages is already part of KahaDB, it may