I am trying to use hazelcast server over multiple nodes. I have implemented the load all in the map store implementation. I am wondering whether this should only be enabled on on server node or all of them? If I deploy the same on all nodes, would this not create database read operations which should not be needed. If I need to deploy the load all only on one node, what is the best strategy (code/API call based or config) that would allow me to cleanly implement the scenario whereby only one server node implements the load all implementation for map store. I can always deploy different code on different servers but I would like to avoid that and wondering about better choices.
Every node needs to have the same configuration/jars etc.
Currently the MapLoader.loadAllKeys is executed on one of the nodes in the cluster. Once the keys are loaded, they are assigned to the owning partitions where the actual data is loaded using the MapLoader.loadAll(keys) method.
Do you think sharing the same configuration/jars is a problem?
来源：https://stackoverflow.com/questions/25310915/load-all-implementation-for-hazelcast