Skip to content

Nova Linux build job runs as root inside its Docker container #5091

@huydhn

Description

@huydhn

The container in question https://hub.docker.com/r/pytorch/manylinux-builder/tags.

There are reports from ExecuTorch and Torchtune about the problematic way of running Nova Linux build job as root inside the container.

  1. On ExecuTorch, buck2 refuses to run as root https://github.com/pytorch/executorch/actions/runs/8655199802/job/23733768589#step:14:125, which blocks ET from using the workflow. For the context, ET use buck2 to gather the source files before passing them to cmake.
  2. On TorchTune, running the build as root leaves some artifacts like the conda folder on the runner, which could prevent subsequent jobs to clean it up as @clee2000 discovered it https://github.com/pytorch/torchtune/actions/runs/8639310064/job/23685409883?pr=688.

Usually, running thing as root is bad, and we should figure out a way to not do it anymore

cc @seemethere @atalman @kit1980 @clee2000 @dbort @malfet

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    Status

    Cold Storage

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions