Hi vkulichenko, The reason is actually a two part problem that I am trying to solve.
1. I want to evenly distribute my ignite computations between the nodes of my cluster as I had noticed that when having 3 nodes and 10 ignite calls, they were being distributed as such 6+2+2=10. Note here I am using the default configuration and doing an ignite.compute().call(myCallable) several times, but ignite doesn't seem to evenly distribute the calls. 2. The more difficult task at hand at the moment is that I am doing affinity calls on my cache and I want to find a way to be sure that my affinity calls will be distributed among my cluster nodes. FairAffinityFunction tries to evenly distribute partitions among the cluster nodes, but it does not guarantee if my keys will map to the same partition or not and I need to always be utilizing as much nodes as I can, or else I might have idle nodes doing nothing because the affinity call is not routing to them. I'd appreciate any advice on the matter. Thanks, Rami -- View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Ignite-RoundRobinLoadBalancingSpi-Per-Task-not-distributing-tasks-tp10991p11004.html Sent from the Apache Ignite Users mailing list archive at Nabble.com.
