How to avoid this error in Windows? ImportError: cannot import name 'RRef' from 'torch.distributed.rpc'

[yiluzhou1]

I'm using Windows 10 laptop with RTX A1000 GPU. I installed MonaLabel (with CUDA). However, the following error popped out:
ImportError: cannot import name 'RRef' from 'torch.distributed.rpc'
It seems it is related to unavailability of rpc in Pytorch Windows. How to bypass this issue? Thank you!

The entire log is as below:

C:\Development\monailabel>monailabel start_server --app apps/radiology --studies datasets/Task09_Spleen/imagesTr --conf models segmentation
Using PYTHONPATH=C:\Users\yzhou\AppData\Local\miniconda3\envs;C:\Users\yzhou\AppData\Local\miniconda3\envs;
""
[2023-06-08 16:45:15,172] [5692] [MainThread] [INFO] (main:285) - USING:: version = False
[2023-06-08 16:45:15,172] [5692] [MainThread] [INFO] (main:285) - USING:: app = C:\Development\monailabel\apps\radiology
[2023-06-08 16:45:15,173] [5692] [MainThread] [INFO] (main:285) - USING:: studies = C:\Development\monailabel\datasets\Task09_Spleen\imagesTr
[2023-06-08 16:45:15,174] [5692] [MainThread] [INFO] (main:285) - USING:: verbose = INFO
[2023-06-08 16:45:15,174] [5692] [MainThread] [INFO] (main:285) - USING:: conf = [['models', 'segmentation']]
[2023-06-08 16:45:15,174] [5692] [MainThread] [INFO] (main:285) - USING:: host = 0.0.0.0
[2023-06-08 16:45:15,174] [5692] [MainThread] [INFO] (main:285) - USING:: port = 8000
[2023-06-08 16:45:15,174] [5692] [MainThread] [INFO] (main:285) - USING:: uvicorn_app = monailabel.app:app
[2023-06-08 16:45:15,175] [5692] [MainThread] [INFO] (main:285) - USING:: ssl_keyfile = None
[2023-06-08 16:45:15,175] [5692] [MainThread] [INFO] (main:285) - USING:: ssl_certfile = None
[2023-06-08 16:45:15,175] [5692] [MainThread] [INFO] (main:285) - USING:: ssl_keyfile_password = None
[2023-06-08 16:45:15,175] [5692] [MainThread] [INFO] (main:285) - USING:: ssl_ca_certs = None
[2023-06-08 16:45:15,175] [5692] [MainThread] [INFO] (main:285) - USING:: workers = None
[2023-06-08 16:45:15,175] [5692] [MainThread] [INFO] (main:285) - USING:: limit_concurrency = None
[2023-06-08 16:45:15,176] [5692] [MainThread] [INFO] (main:285) - USING:: access_log = False
[2023-06-08 16:45:15,176] [5692] [MainThread] [INFO] (main:285) - USING:: root_path = /
[2023-06-08 16:45:15,176] [5692] [MainThread] [INFO] (main:285) - USING:: log_level = info
[2023-06-08 16:45:15,176] [5692] [MainThread] [INFO] (main:285) - USING:: log_config = None
[2023-06-08 16:45:15,176] [5692] [MainThread] [INFO] (main:285) - USING:: dryrun = False
[2023-06-08 16:45:15,176] [5692] [MainThread] [INFO] (main:285) - USING:: action = start_server
[2023-06-08 16:45:15,177] [5692] [MainThread] [INFO] (main:296) -
Allow Origins: ['*']
[2023-06-08 16:45:16,657] [5692] [MainThread] [INFO] (uvicorn.error:74) - Started server process [5692]
[2023-06-08 16:45:16,657] [5692] [MainThread] [INFO] (uvicorn.error:48) - Waiting for application startup.
[2023-06-08 16:45:16,657] [5692] [MainThread] [INFO] (monailabel.interfaces.utils.app:37) - Initializing App from: C:\Development\monailabel\apps\radiology; studies: C:\Development\monailabel\datasets\Task09_Spleen\imagesTr; conf: {'models': 'segmentation'}
[2023-06-08 16:45:17,863] [5692] [MainThread] [INFO] (monailabel.utils.others.class_utils:57) - Subclass for MONAILabelApp Found: <class 'main.MyApp'>
[2023-06-08 16:45:17,878] [5692] [MainThread] [ERROR] (uvicorn.error:123) - Traceback (most recent call last):
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\starlette\routing.py", line 677, in lifespan
async with self.lifespan_context(app) as maybe_state:
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\starlette\routing.py", line 566, in aenter
await self.router.startup()
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\starlette\routing.py", line 654, in startup
await handler()
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\monailabel\app.py", line 108, in startup_event
instance = app_instance()
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\monailabel\interfaces\utils\app.py", line 50, in app_instance
app = c(app_dir=app_dir, studies=studies, conf=conf)
File "C:\Development\monailabel\apps\radiology\main.py", line 49, in init
for c in get_class_names(lib.configs, "TaskConfig"):
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\monailabel\utils\others\class_utils.py", line 164, in get_class_names
module = importlib.import_module("." + name, package=current_module_name)
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\importlib_init.py", line 127, in import_module
return bootstrap.gcd_import(name[level:], package, level)
File "", line 1030, in gcd_import
File "", line 1007, in find_and_load
File "", line 986, in find_and_load_unlocked
File "", line 680, in load_unlocked
File "", line 850, in exec_module
File "", line 228, in call_with_frames_removed
File "C:\Development\monailabel\apps\radiology\lib\configs\deepedit.py", line 17, in
import lib.trainers
File "C:\Development\monailabel\apps\radiology\lib\trainers_init.py", line 12, in
from .deepedit import DeepEdit
File "C:\Development\monailabel\apps\radiology\lib\trainers\deepedit.py", line 45, in
from monailabel.tasks.train.basic_train import BasicTrainTask, Context
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\monailabel\tasks\train\basic_train.py", line 23, in
import ignite
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\ignite_init.py", line 3, in
import ignite.engine
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\ignite\engine_init.py", line 10, in
from ignite.metrics import Metric
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\ignite\metrics_init.py", line 7, in
from ignite.metrics.frequency import Frequency
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\ignite\metrics\frequency.py", line 7, in
from ignite.handlers.timing import Timer
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\ignite\handlers_init.py", line 5, in
from ignite.handlers.checkpoint import Checkpoint, DiskSaver, ModelCheckpoint
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\ignite\handlers\checkpoint.py", line 17, in
from torch.distributed.optim import ZeroRedundancyOptimizer
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\torch\distributed\optim_init.py", line 8, in
from .optimizer import DistributedOptimizer
File "C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\torch\distributed\optim\optimizer.py", line 9, in
from torch.distributed.rpc import RRef
ImportError: cannot import name 'RRef' from 'torch.distributed.rpc' (C:\Users\yzhou\AppData\Local\miniconda3\envs\TEST\lib\site-packages\torch\distributed\rpc_init.py)

[2023-06-08 16:45:17,879] [5692] [MainThread] [ERROR] (uvicorn.error:59) - Application startup failed. Exiting.

[tangy5]

Emmm, it looks like the PyTorch Distributed package does not support Windows well on your machine. You could refer to Pytorch installation guide.

Or you could try use the monailabel docker, maybe a better choice to make it compatible with your machine without bothering too many packages installation issues.

docker run -it --rm --gpus all --ipc=host --net=host -v ~:/workspace/ projectmonai/monailabel:latest bash

[yiluzhou1]

@tangy5 Thank you so much for your advice!