I am running hazelcast members and clients with below context:
- I want the clients to control the (de)serialization, and onboard new clients without hazelcast members being involved in serializtion business.
- Want to make Hazelcast Map persistent based on naming conventions, e.g. All maps with "*Store" matching pattern, needs to be persisted to cassandra db.
- Since i chose to persist in binary format, I would like same byte[] stored in hazelcast to be supplied to the mapstore interface.
My problem is - hazelcast seems to apply deserializer before calling the mapstore .store() method and looks for the serializer class with originally stored typeId on the server side. The logical solution is either hazelcast does not apply deserializer for mapstore if my mapstore type is byte[], or have let me define deserializer which would work regardless of typeId. Looks like both does not seem to be possible.
Any idea to tackle the problem is greatly appreciated.