aws / aws/amazon-sagemaker-examples
Probable bug in MMS BYO example
- Dominant language
- Jupyter Notebook
- Stars
- 11k
- Forks
- 7k
- Avg merge
- 8h 29m
- Merged PRs (30d)
- 8
Description
The example on using `sagemaker-inference` toolkit for multi-model serving bring-your-own-container has some code which smells like a probable bug:
https://github.com/aws/amazon-sagemaker-examples/blob/cb28f5f09c536c8ce5db96c8cf06cd4913d762e9/advanced_functionality/multi_model_bring_your_own/container/dockerd-entrypoint.py#L27
The join-split done to `sys.argv[1:]` results in breaking up any arguments which contain whitespace into separate arguments; e.g. if one of the command-line arguments to the entry point is `"this is foobar"`, then `join` followed by `shlex.split` ends up splitting it into 3 separate arguments.
If this behavior is intentional, it would be excellent to have a comment above the referred line explaining why this is done, as this may reflect on how users should write their custom inference entry point scripts. If this behavior is not intentional, the code could be simplified and made less bug-prone by simply using `subprocess.check_call(sys.argv[1:])`
Contributor guide
Assessment
This issue has not been assessed yet.