Skip to content

feat: unify jobs + mmt under jobs command - #68

Draft
tejapulagam wants to merge 1 commit into
mainfrom
feat/unify-mmt-under-jobs
Draft

feat: unify jobs + mmt under jobs command#68
tejapulagam wants to merge 1 commit into
mainfrom
feat/unify-mmt-under-jobs

Conversation

@tejapulagam

Copy link
Copy Markdown
Contributor

Summary

  • Route multi-machine training through lightning job so one CLI covers single- and multi-machine jobs.
  • job run --num-machines N (default 1) creates an MMT when N > 1.
  • job list includes both Jobs and MMTs (num_machines column).
  • job inspect / logs / ssh accept --rank for MMT workers.
  • job stop / delete resolve Job or MMT and act on the whole resource (no per-rank stop/delete).
  • Shared resolution helpers in resource_resolution (resolve_job_or_mmt, resolve_mmt_machine).
  • lightning mmt left in place for now (deprecation later).

Test plan

  • job run with default --num-machines 1 still creates a Job
  • job run --num-machines 2 creates an MMT
  • job list shows Jobs and MMTs with correct num_machines
  • job inspect / logs / ssh work for Jobs; with --rank for MMT workers
  • job stop / delete stop/delete an entire MMT (not a single rank)
  • Existing lightning mmt commands still work
  • Unit tests: python/tests/cli/job/ + test_resource_resolution.py

@tejapulagam tejapulagam changed the title feat: unify jobs + mmt under jobs command wip feat: unify jobs + mmt under jobs command Jul 30, 2026
@tejapulagam tejapulagam changed the title wip feat: unify jobs + mmt under jobs command feat: unify jobs + mmt under jobs command Jul 30, 2026
@tejapulagam
tejapulagam marked this pull request as draft July 30, 2026 22:27
@justusschock

Copy link
Copy Markdown
Member

This should match with the SDK though, so we'd need to merge both classes as well (which would be doable).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants