I have a RedHat DataGrid cluster with two nodes on different servers and I use it from Camel route. So, when I define endpoint to cache I set one of the node host (i.e.):
<to uri="infinispan://node1.some.com:11222" />
DataGrid Cluster works fine in terms of caches. they are replicated, distributed etc. But if node1 is down then I have no connection to cache. So question: Do I need to have httpd with mod_cluster upfront as load balancer or there is a way to setup cache cluster level endpoint to do not care about what node is up and how many nodes are there?
BTW: I tried to find an answer, but did not get clear answer so far.
Thanks.
The Hot Rod protocol automatically receives server topology information (i.e. joiners / leavers) as they happen. The connection string specifies the initial hosts, i.e. those that the client will attempt to connect to initially. As long as one of those is up and running, the clients will be able to talk to the whole cluster. To specify multiple initial hosts separate them with semicolons: host1:port1;host2:port2;...