Is your feature request related to a problem? Please describe.
At present, when some rpc errors occur, the client cannot obtain the error information well.
-
If a brpc overcrowded error occurs, the client's behavior is to wait until the query times out, but the correct behavior should be to return the error immediately.
-
When the BE node is added to the blacklist due to an rpc error, the client only displays there is no scannode backend without a clear error message.