Typically you need N/2 + 1 nodes for a cluster to be active, so a 3-node cluster can only handle one simultaneous failure. If you want to handle 2 simultaneous failures, use a 5-node cluster; for 3, use a 7-node cluster.
None of this is specific to ActiveMQ; this applies to many clustered technologies: LevelDB, Zookeeper, Oracle, etc. Tim On Nov 27, 2015 10:13 AM, "vieksmiles" <[email protected]> wrote: > Hi, > > I've installed *activemq(5.13-SNAPSHOT)* in 3 servers(A,B,C) and Zookeeper > in all three. I'm doing replicated LevelDB store approach. > > I produced 10 msgs(I can see all of them in one of the servers on UI). Now > if I shutdown activemq process in server-A, I can receive msgs but if I > shutdown activemq in server-B also, i'm not able receive any msgs. > > NOTE:I'm using failover as > > > failover:(tcp://server-A:61616,tcp://server-B:61616,tcp://server-C:61616)?randomize=false > > My doubt is > > *what is max no of server failover activemq can handle > > *Or is there any fixed no of servers to be up and running for activemq to > work(as 2 in my 3 clusters) > > Please help me I serched for this in web, but went in vain.. > > > > -- > View this message in context: > http://activemq.2283324.n4.nabble.com/max-server-down-an-activemq-can-tolerate-in-replicated-levedb-store-tp4704318.html > Sent from the ActiveMQ - User mailing list archive at Nabble.com. >
