usage of coordination id in RPC header

[Xiao Meng <xi...@simba.com>]

Hi,

I have some question about the usage of coordination id.

https://github.com/apache/incubator-drill/blob/master/protocol/src/main/protobuf/GeneralRPC.proto

In the RPC protocol,  the RPC header has a coordination id (commented as "reusable coordination identifier. Sender defines. Server returns on return. Irrelevant for purely single direction rpc.")

?
-  Does that mean each client can reuse the same coordination id?

For example, client A and client B can send the message with coordination id zero.

- For a single client, could the coordination  be reused?

For example,  a client submits a query to a drillbit and get the query handle back from the server. Then, the server will push the results to the client (send message as exec::rpc::Response) and the client sends Ack message for each chunk of results.

Now if we submit the second query, is the coordination id must be different from the previous ones?

Thanks,

Xiao