Created by: danleifeng

PR types

Function optimization

PR changes

APIs

Describe

fleetrun: add endpoints log for details

after fixed:

-----------  Configuration Arguments -----------
gpus: None
ips: 127.0.0.1
log_dir: log
server_num: None
servers:
training_script: fleetrun_demo.py
training_script_args: []
worker_num: None
workers:
------------------------------------------------
WARNING 2020-09-21 11:42:33,790 launch.py:476] Not found distinct arguments. Default use gpu collective mode
INFO 2020-09-21 11:42:33,792 launch_utils.py:437] Local start 4 processes. First process distributed environment info (Only For Debug):
    +=======================================================================================+
    |                        Distributed Envs                      Value                    |
    +---------------------------------------------------------------------------------------+
    |                 PADDLE_CURRENT_ENDPOINT                 127.0.0.1:23008               |
    |                     PADDLE_TRAINERS_NUM                        4                      |
    |                     FLAGS_selected_gpus                        0                      |
    |                PADDLE_TRAINER_ENDPOINTS  ... 0.1:26420,127.0.0.1:27405,127.0.0.1:10590|
    |                       PADDLE_TRAINER_ID                        0                      |
    +=======================================================================================+

INFO 2020-09-21 11:42:33,793 launch_utils.py:438] details abouts PADDLE_TRAINER_ENDPOINTS can be found in log/endpoints.log.

log/endpoints.log:

PADDLE_TRAINER_ENDPOINTS:
127.0.0.1:23008
127.0.0.1:26420
127.0.0.1:27405
127.0.0.1:10590