Lightning-AI / Lightning-AI/pytorch-lightning
Checkpoints Saving with different permissions to account defaults
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 31.4k
- Forks
- 3.8k
- Avg merge
- 6d 7h
- Merged PRs (30d)
- 6
Description
### Bug description
When working on a managed cluster all our files save with "-rw-r--r--" permissions. However the checkpoint files from lightning are being saved with "-rw-------" permissions. We've gone round in circles trying to debug this and do not understand. Is there anything in the lightning code base that is causing this/is it configurable?
### Environment
Current environment
```
#- PyTorch Lightning Version (e.g., 2.4.0): 2.4.0
#- PyTorch Version (e.g., 2.4): 2.4
#- Python version (e.g., 3.12): 3.12
#- OS (e.g., Linux): Ubuntu 22
#- CUDA/cuDNN version: 12.5
#- GPU models and configuration: `8xH100`
#- How you installed Lightning(`conda`, `pip`, source): `conda`
```
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Reproduce checkpoint saving with Lightning 2.4.0 in the reported Ubuntu 22, Python 3.12, PyTorch 2.4 environment and compare its permissions with other saved files. Trace the checkpoint-saving entry point to determine whether Lightning sets the restrictive permissions, then identify whether the behavior is configurable and document the observed result.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- operating-systems
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100