open-mmlab/mmengine

[Bug] DVCLiveVisBackend initializing DVC workspace in subdirectories

Opened this issue · 1 comments

smarais commented

Prerequisite

Environment

OrderedDict([('sys.platform', 'linux'), ('Python', '3.10.8 (main, Nov 4 2022, 13:48:29) [GCC 11.2.0]'), ('CUDA available', True), ('MUSA available', False), ('numpy_random_seed', 2147483648), ('GPU 0', 'NVIDIA GeForce RTX 3080 Ti Laptop GPU'), ('CUDA_HOME', '/opt/conda'), ('NVCC', 'Cuda compilation tools, release 11.6, V11.6.124'), ('GCC', 'gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0'), ('PyTorch', '1.13.1'), ('PyTorch compiling details', 'PyTorch built with:\n - GCC 9.3\n - C++ Version: 201402\n - Intel(R) oneAPI Math Kernel Library Version 2021.4-Product Build 20210904 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v2.6.0 (Git Hash 52b5f107dd9cf10910aaa19cb47f3abf9b349815)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 11.6\n - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_61,code=sm_61;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_37,code=compute_37\n - CuDNN 8.3.2 (built against CUDA 11.5)\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.6, CUDNN_VERSION=8.3.2, CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, CXX_FLAGS= -fabi-version=11 -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wunused-local-typedefs -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Werror=cast-function-type -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.13.1, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, \n'), ('TorchVision', '0.14.1'), ('OpenCV', '4.9.0'), ('MMEngine', '0.10.3')])

Reproduces the problem - code sample

I am using DVC to run an MMPretrain task. I am enabling the DVCLiveVisBackend in any training config, like so:

visualizer = dict(
    type="Visualizer",
    vis_backends=[dict(type="DVCLiveVisBackend",save_dir="dvclive"],
)

With reference to: https://github.com/open-mmlab/mmengine/blame/v0.10.3/mmengine/visualization/vis_backend.py#L1185-L1189

This code block attempts to discover a Git repository starting from the current directory (os.curdir). If it fails to find Git information (i.e., if a KeyError is thrown), it then initializes a new DVC repository in the current directory without considering whether a DVC repository already exists at a higher level in the directory structure.

In my use case os.curdir will correspond to the dvc.yaml file location which is not in the dvc/git root directory.

Suggestions:

  1. Adjust the initialization logic to better detect existing DVC repositories (for example dvc root and avoid reinitializing DVC in subdirectories.
  2. Add a param to disable this behavior or provide the workspace directory

Reproduces the problem - command or script

In my case, I am tesing in mmpretrain.
dvc repro subfolder/dvc.yaml from the project root folder.

Example dvc.yaml:

stages:
  train:
    cmd: python /workspace/mmpretrain/tools/train.py --work-dir /workspace/myproject/output /workspace/mmpretrain/configs/resnet/resnet34_8xb32_in1k.py

Reproduces the problem - error message

No error message. Creating DVC workspaces in sub-directories causes issues for DVC in that it treats the subfolder as a separate project.

Additional information

No response

michaelgruner commented

I had a similar situation. After some debugging I found out that the problem was that no default_signature was found. Here's the test it performs:

path = pygit2.discover_repository(os.fspath(os.curdir), True, '')
pygit2.Repository(path).default_signature
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
KeyError: "config value 'user.name' was not found"

which made DVC initialize the new repository, breaking everything in its way. This is true for most docker containers.

After running:

git config --global user.name "Michael Gruner"
git config --global user.email "michael.gruner@ridgerun.ai"

everything works as expected. Hope it helps someone!