[rqd] Fix issue on rqd when killing a frame that no longer exists (#1533)

When killing a frame on a rqd node that has restarted and lost track of
its running frames, the request should return a status to be handled
accordingly and trigger the lostProc logic. The current behavior leaves
the frame stuck at RUNNING until the job itself is killed or eaten.

Author: DiegoTavares

rqd/rqd/rqdservicers.py

@@ -68,6 +68,10 @@ def KillRunningFrame(self, request, context):
         if frame:
             frame.kill(message=request.message)
         else:
+            context.set_details(
+                "The requested frame to kill was not found. frameId: {}".format(
+                    request.frame_id))
+            context.set_code(grpc.StatusCode.NOT_FOUND)
             log.warning("Wasn't able to find frame(%s) to kill", request.frame_id)
         return rqd.compiled_proto.rqd_pb2.RqdStaticKillRunningFrameResponse()