Tacotron2: RuntimeError: Address already in use

[lsuperman]

[lsuperman]

hello, everyone, I am trying to run the tacotron2 with 2 gpu, my command is
CUDA_VISIBLE_DEVICES=0,1 nohup /opt/anaconda3-5.1.0/bin/python3.6 -m multiproc train.py --output_directory=outdir --log_directory=logdir -c outdir/checkpoint_13000 --warm_start >> output.out &

but,the output as following:

FP16 Run: False
Dynamic Loss Scaling: True
Distributed Run: True
cuDNN Enabled: True
cuDNN Benchmark: False
Initializing Distributed
Traceback (most recent call last):
File "train.py", line 303, in
args.warm_start, args.n_gpus, args.rank, args.group_name, hparams)
File "train.py", line 179, in train
init_distributed(hparams, n_gpus, rank, group_name)
File "train.py", line 49, in init_distributed
world_size=n_gpus, rank=rank, group_name=group_name)
File "/opt/anaconda3-5.1.0/lib/python3.6/site-packages/torch/distributed/distributed_c10d.py", line 354, in init_process_group
store, rank, world_size = next(rendezvous(url))
File "/opt/anaconda3-5.1.0/lib/python3.6/site-packages/torch/distributed/rendezvous.py", line 95, in _tcp_rendezvous_handler
store = TCPStore(result.hostname, result.port, start_daemon)
RuntimeError: Address already in use
/opt/anaconda3-5.1.0/lib/python3.6/site-packages/torch/distributed/distributed_c10d.py:86: UserWarning: torch.distributed.reduce_op is deprecated, please use torch.distributed.ReduceOp instead
warnings.warn("torch.distributed.reduce_op is deprecated, please use "
['train.py', '--output_directory=outdir', '--log_directory=logdir', '-c', 'outdir/checkpoint_13000', '--warm_start', '--n_gpus=2', '--group_name=group_2019_04_10-105411', '--rank=0']
['train.py', '--output_directory=outdir', '--log_directory=logdir', '-c', 'outdir/checkpoint_13000', '--warm_start', '--n_gpus=2', '--group_name=group_2019_04_10-105411', '--rank=1']
~

[lsuperman]

Any ideas would be greatly appreciated

[delgerdalai]

Your previous execution is not completely terminated.
netstat -nltp this command will show process lists which are listen ports. Now, you have to kill process which is listen 54321 port. kill -9 PID. in this case. kill -9 1594. Then try to start training.

[lsuperman]

@delgerdalai thanks