def get_cluster_from_args(selected_gpus):
cluster_node_ips = '127.0.0.1'
node_ip = '127.0.0.1'
node_ips = [x.strip() for x in cluster_node_ips.split(',')]
node_ips.index(node_ip)
free_ports = None
free_ports = find_free_ports(len(selected_gpus))
if free_ports is not None:
free_ports = list(free_ports)
return get_cluster(node_ips, node_ip, free_ports, selected_gpus)
def get_cluster_from_args(selected_gpus):
cluster_node_ips = '127.0.0.1'
node_ip = '127.0.0.1'
node_ips = [x.strip() for x in cluster_node_ips.split(',')]
node_ips.index(node_ip)
free_ports = None
free_ports = find_free_ports(len(selected_gpus))
if free_ports is not None:
free_ports = list(free_ports)
trainer_endpoints = []
for ip in node_ips:
trainer_endpoints.append(["%s:%d" % (ip, port) for port in free_ports])
return get_cluster(node_ips, node_ip, trainer_endpoints, selected_gpus)
def test_communicator(self):
run_server_cmd = """
from __future__ import print_function
import sys
import os
import time
import threading
import subprocess
import unittest
import numpy
import paddle
import paddle.fluid as fluid
from paddle.fluid.communicator import Communicator
import paddle.fluid.incubate.fleet.base.role_maker as role_maker
from paddle.fluid.incubate.fleet.parameter_server.mode import DistributedMode
import paddle.distributed.fleet as fleet
from test_communicator_geo import TestCommunicatorGeoEnd2End
paddle.enable_static()
class RunServer(TestCommunicatorGeoEnd2End):
def runTest(self):
pass
os.environ["TRAINING_ROLE"] = "PSERVER"
half_run_server = RunServer()
half_run_server.run_ut()
"""
server_file = "run_server_for_communicator_geo.py"
with open(server_file, "w") as wb:
wb.write(run_server_cmd)
port = find_free_ports(1).pop()
os.environ["TRAINING_ROLE"] = "PSERVER"
os.environ["PADDLE_PORT"] = str(port)
os.environ["PADDLE_PSERVERS_IP_PORT_LIST"] = "127.0.0.1:{}".format(port)
_python = sys.executable
ps_cmd = "{} {}".format(_python, server_file)
ps_proc = subprocess.Popen(
ps_cmd.strip().split(" "),
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
time.sleep(5)
os.environ["TRAINING_ROLE"] = "TRAINER"
self.run_ut()
ps_proc.kill()
ps_proc.wait()
outs, errs = ps_proc.communicate()
if os.path.exists(server_file):
os.remove(server_file)