Spurious warning message when submitting many tasks.

import ray

ray.init()

@ray.remote 
def double(x): 
    return x * 2

results = [double.remote(i) for i in range(100000)]

ray.get(results)

Running the above in IPython succeeds but prints the following warning.

(pid=78589) [libprotobuf ERROR external/com_google_protobuf/src/google/protobuf/wire_format_lite.cc:603] String field ‘ray.rpc.CoreWorkerStats.current_task_func_desc’ contains invalid UTF-8 data when serializing a protocol buffer. Use the ‘bytes’ type if you intend to send raw bytes. (pid=raylet) [libprotobuf ERROR external/com_google_protobuf/src/google/protobuf/wire_format_lite.cc:603] String field ‘ray.rpc.CoreWorkerStats.current_task_func_desc’ contains invalid UTF-8 data when parsing a protocol buffer. Use the ‘bytes’ type if you intend to send raw bytes.

Note that the ray.get is not necessary to trigger the warning.

[ericl]

I think the only relation to many tasks is that it delays exit until the first stats heartbeat.

[raulchen]

I just thought of a possible cause. When --load-code-from-local is not set, Python’s function descriptor contains a fourth element, which is the hash value of the source code. This element is raw binary and not UTF-8-decodable. But I don’t see how it’s related with many tasks.

To fix this issue, I think we can just make CoreWorkerStats.current_task_func_desc a single concatenated string, instead of repeated string.

Also, @fyrestone is refactoring FunctionDescriptor as part of the cross-lang feature. He’ll add a c++ class to represent function descriptor. So dealing with such problems will be easier after that.