Skip to content

Fix permission error for ORTModule lock file #7814

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Thiago Crepaldi (thiagocrepaldi) merged 3 commits into from
May 26, 2021
Merged

Fix permission error for ORTModule lock file #7814

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
52cdfce
Fix permission error for ORTModule lock file
May 24, 2021
aa37eee
Add support to use TORCH_CPP_BUILD_DIR env var to set torch cpp ext dir
May 25, 2021
ff70dc2
Generate documentation only if the previous taskjs succeeds
May 26, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
32 changes: 27 additions & 5 deletions orttraining/orttraining/python/training/ortmodule/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,27 +11,49 @@
################################################################################
ONNX_OPSET_VERSION = 12
MINIMUM_TORCH_VERSION_STR = '1.8.1'
TORCH_CPP_BUILD_DIR = os.path.join(os.path.dirname(__file__),'torch_inline_extensions')

# Use one of the available directories as Torch CPP extension in the following order:
# 1) Path at listed at TORCH_EXTENSIONS_DIR environment variable
# 2) Default Python package dir
# 3) <Home directory>/.cache
home_dir = os.path.expanduser("~")
python_package_dir = os.path.dirname(__file__)
torch_extensions_dir = os.environ.get('TORCH_EXTENSIONS_DIR')

TORCH_CPP_BUILD_DIR = os.path.join(python_package_dir,'torch_inline_extensions')
TORCH_CPP_BUILD_DIR_BACKUP = os.path.join(home_dir, '.cache', 'torch_ort_extensions')

if torch_extensions_dir is not None and os.access(torch_extensions_dir, os.X_OK | os.W_OK):
TORCH_CPP_BUILD_DIR = torch_extensions_dir
elif not os.access(python_package_dir, os.X_OK | os.W_OK):
Comment thread
snnn marked this conversation as resolved.
if os.access(home_dir, os.X_OK | os.W_OK):
TORCH_CPP_BUILD_DIR = TORCH_CPP_BUILD_DIR_BACKUP
else:
extra_message = ''
if torch_extensions_dir:
extra_message = 'or the path pointed by the TORCH_EXTENSIONS_DIR environment variable '
raise PermissionError('ORTModule could not find a writable directory to cache its internal files.',
Comment thread
thiagocrepaldi marked this conversation as resolved.
f'Make {python_package_dir} or {home_dir} {extra_message}writable and try again.')

# Check whether Torch C++ extension compilation was aborted in previous runs
if not os.path.exists(TORCH_CPP_BUILD_DIR):
os.makedirs(TORCH_CPP_BUILD_DIR, exist_ok = True)
elif os.path.exists(os.path.join(TORCH_CPP_BUILD_DIR,'lock')):
print("WARNING: ORTModule detected PyTorch CPP extension's lock file during initialization, "
"which can cause unexpected hangs. "
f"Delete {os.path.join(TORCH_CPP_BUILD_DIR,'lock')} to prevent unexpected behavior.")
f"Delete {os.path.join(TORCH_CPP_BUILD_DIR,'lock')} if a hang occurs.")

@snnn Changming Sun (snnn) May 27, 2021

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We should avoid using the word "hang" in our code. We can say it stops responding.


# Verify proper PyTorch is installed before proceding to ONNX Runtime initializetion
# Verify proper PyTorch is installed before proceding to ONNX Runtime initialization
try:
import torch
torch_version = version.parse(torch.__version__.split('+')[0])
minimum_torch_version = version.parse(MINIMUM_TORCH_VERSION_STR)
if torch_version < minimum_torch_version:
raise RuntimeError(
f'ONNXRuntime ORTModule frontend requires PyTorch version greater or equal to {MINIMUM_TORCH_VERSION_STR}, '
f'ONNX Runtime ORTModule frontend requires PyTorch version greater or equal to {MINIMUM_TORCH_VERSION_STR}, '
f'but version {torch.__version__} was found instead.')
except:
raise(f'PyTorch {MINIMUM_TORCH_VERSION_STR} must be installed in order to run ONNXRuntime ORTModule frontend!')
raise(f'PyTorch {MINIMUM_TORCH_VERSION_STR} must be installed in order to run ONNX Runtime ORTModule frontend!')

# ORTModule must be loaded only after all validation passes
from .ortmodule import ORTModule
16 changes: 8 additions & 8 deletions tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -119,7 +119,7 @@ stages:

- task: CmdLine@2
displayName: 'Build Python Documentation'
condition: ne(variables['PythonVersion'], '3.9') # tensorflow not available on python 3.9
condition: and(succeeded(), ne(variables['PythonVersion'], '3.9')) # tensorflow not available on python 3.9
inputs:
script: |
mkdir -p $HOME/.onnx
Expand All @@ -137,7 +137,7 @@ stages:

- task: CopyFiles@2
displayName: 'Copy Python Documentation to: $(Build.ArtifactStagingDirectory)'
condition: ne(variables['PythonVersion'], '3.9') # tensorflow not available on python 3.9
condition: and(succeeded(), ne(variables['PythonVersion'], '3.9')) # tensorflow not available on python 3.9
inputs:
SourceFolder: '$(Build.BinariesDirectory)/docs/inference/html'
Contents: '**'
Expand Down Expand Up @@ -431,7 +431,7 @@ stages:

- task: CmdLine@2
displayName: 'Build Python Documentation'
condition: ne(variables['PythonVersion'], '3.9') # tensorflow not available on python 3.9
condition: and(succeeded(), ne(variables['PythonVersion'], '3.9')) # tensorflow not available on python 3.9
inputs:
script: |
mkdir -p $HOME/.onnx
Expand All @@ -447,7 +447,7 @@ stages:

- task: CopyFiles@2
displayName: 'Copy Python Documentation to: $(Build.ArtifactStagingDirectory)'
condition: ne(variables['PythonVersion'], '3.9') # tensorflow not available on python 3.9
condition: and(succeeded(), ne(variables['PythonVersion'], '3.9')) # tensorflow not available on python 3.9
inputs:
SourceFolder: '$(Build.BinariesDirectory)/docs/training/html'
Contents: '**'
Expand Down Expand Up @@ -588,7 +588,7 @@ stages:

- task: CmdLine@2
displayName: 'Build Python Documentation'
condition: ne(variables['PythonVersion'], '3.9') # tensorflow not available on python 3.9
condition: and(succeeded(), ne(variables['PythonVersion'], '3.9')) # tensorflow not available on python 3.9
inputs:
script: |
mkdir -p $HOME/.onnx
Expand All @@ -606,7 +606,7 @@ stages:

- task: CopyFiles@2
displayName: 'Copy Python Documentation to: $(Build.ArtifactStagingDirectory)'
condition: ne(variables['PythonVersion'], '3.9') # tensorflow not available on python 3.9
condition: and(succeeded(), ne(variables['PythonVersion'], '3.9')) # tensorflow not available on python 3.9
inputs:
SourceFolder: '$(Build.BinariesDirectory)/docs/training/html'
Contents: '**'
Expand Down Expand Up @@ -761,7 +761,7 @@ stages:

- task: CmdLine@2
displayName: 'Build Python Documentation'
condition: ne(variables['PythonVersion'], '3.9') # tensorflow not available on python 3.9
condition: and(succeeded(), ne(variables['PythonVersion'], '3.9')) # tensorflow not available on python 3.9
inputs:
script: |
mkdir -p $HOME/.onnx
Expand All @@ -779,7 +779,7 @@ stages:

- task: CopyFiles@2
displayName: 'Copy Python Documentation to: $(Build.ArtifactStagingDirectory)'
condition: ne(variables['PythonVersion'], '3.9') # tensorflow not available on python 3.9
condition: and(succeeded(), ne(variables['PythonVersion'], '3.9')) # tensorflow not available on python 3.9
inputs:
SourceFolder: '$(Build.BinariesDirectory)/docs/training/html'
Contents: '**'
Expand Down