How does error handling and status codes work in gRPC?

Expected Interview Answer

Step-by-Step Explanation

1. Step 1

   Return a status, not a response

   On failure the server aborts the RPC and returns a status code and message instead of a normal reply message.

2. Step 2

   Pick the right canonical code

   Map the failure to one of gRPC's fixed codes — INVALID_ARGUMENT for bad input, NOT_FOUND for missing entities, UNAVAILABLE for transient outages.

3. Step 3

   Attach structured details

   Use google.rpc.Status with detail protos (e.g. BadRequest, QuotaFailure) serialized into the grpc-status-details-bin trailer for machine-readable context.

4. Step 4

   Propagate to the client

   The client library surfaces the status as an exception or error object exposing the code, message, and any details.

5. Step 5

   Handle by code

   The caller branches on the code — retry UNAVAILABLE, surface INVALID_ARGUMENT to the user, treat DEADLINE_EXCEEDED as a timeout.

Code Example

import grpc

def GetUser(self, request, context):
    user = db.find(request.id)
    if user is None:
        context.set_code(grpc.StatusCode.NOT_FOUND)
        context.set_details(f"user {request.id} does not exist")
        return user_pb2.User()
    return user

try:
    resp = stub.GetUser(user_pb2.GetUserRequest(id=42))
except grpc.RpcError as e:
    if e.code() == grpc.StatusCode.NOT_FOUND:
        print("no such user:", e.details())
    elif e.code() == grpc.StatusCode.UNAVAILABLE:
        retry()

Follow-up Questions

• Which gRPC status codes are safe to retry automatically?
• How do you attach field-level validation errors to a gRPC error?
• How do gRPC status codes map to HTTP/2 and gRPC-Web?
• What is the difference between UNKNOWN and INTERNAL?
• How would you convert application exceptions into gRPC statuses centrally?