Reconcile semantics differences between Avro RPC and Thrift RPC exceptions.

Avro Based MasterRPC's and Node to Node Event RPC's have different semantics than thrift in the face of physical network failures.

Question: (I think avro tranceiver == thrift transport)
1) Where/How can I configure Avro Http/Socket transceiver timeouts?
2) Need to make sure thrown exceptions from avro transceiver are handled equivalently to Thrift's transport exceptions
3) Is there an equivalent to SO_TIMEOUT to the http transceiver?
4) One-way Avro rpcs seem to not fail when there are physical network failures (powerdown server/disconnect the wire, NOT kill -9 server).