Skip to content

[BUG]: colossalai run does not behave like a normal process #2447

@FrankLeeeee

Description

@FrankLeeeee

🐛 Describe the bug

When using colossalai run, no matter whether the training program works or not, the CLI will always treat it like a process which runs fine. This is because that the multi-node runner runs the code as a subprocess, therefore, any error in the subprocess does not surface back to the main program.

Environment

No response

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions