early stop worker got Exception Error

[kamirdin]

🐛 Describe the bug

dp = ...
dp = dp.sharding_filter()
rs = MultiProcessingReadingService(num_workers=4)
dataloader = DataLoader2(dp, reading_service=rs)

for _ in dataloader:
    pass

dataloader.shutdown()

Exception: Can not expect any response without submitted request
This exception is thrown by __iter__ of _IterateQueueDataPipes(datapipes=[QueueWrapper, QueueWrapper, QueueWrapper, QueueWrapper])

related to 901b483

Versions

[pip3] numpy==1.26.4
[pip3] pytorch-wpe==0.0.1
[pip3] rotary-embedding-torch==0.6.4
[pip3] torch==2.1.1
[pip3] torch-complex==0.4.4
[pip3] torchaudio==2.1.1
[pip3] torchdata==0.7.1

[andrewkho]

Hi @kamirdin, please see #1196 as we won't be supporting dataloader2 nor datapipes in torchdata moving forward