[core] ray.init(raylet_socket_name=...) doesn't work with multiple workers

[wuisawesome]

What is the problem?

Ray version and other system information (Python version, TensorFlow version, OS):
Ubuntu 18.04 (quite certain this applies to all *nix systems)

Reproduction (REQUIRED)

Please provide a script that can be run to reproduce the issue. The script should have no external library dependencies (i.e., use fake or mock data / environments):

import ray
from time import sleep
ray.init(num_cpus=4, raylet_socket_name="/tmp/abcdef")

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

ray.get([foo.remote(i) for i in range(4)])

Throws an exception

If we cannot run your script, we cannot fix your issue.

• I have verified my script runs in a clean environment and reproduces the issue.
• I have verified the issue also occurs with the latest wheels.