Lepton Executor fails with `exec /bin/sh: argument list too long`
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 258
- Forks
- 113
- Avg merge
- 1d 3h
- Merged PRs (30d)
- 5
Description
Describe the bug
When using the lepton executor there's a data mover step that uploads the code prior to kicking off the job. This is limited to just 128kb because it's using busybox:1.37.0 container which is a very low ARG_MAX limit. This fails with the error:
exec /bin/sh: argument list too long
Steps/Code to reproduce bug
The data copy command) embeds the entire job directory as a base64-encoded tarball and run it as a shell command: cmd = f"echo {encoded_data} | base64 -d > {dest_path}/archive.tar.gz && tar -xzf ..." causing this to fail.
Expected behavior
This should work for arbitrary amount of files.
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
Start in nemo_run/core/execution/lepton.py around the data mover at line 101 and inspect how the encoded job directory is passed to busybox:1.37.0. Reproduce with a job directory large enough to exceed the 128 KB argument limit; done means the Lepton executor can upload and extract arbitrarily many files without the argument-list error.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, shell
- Domain
- cloud, infrastructure
- Issue type
- Bug
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 62/100