Skip to content

Add 'local' launch script for graph store mode - so we can simulate the cluster on dev boxes#587

Merged
kmontemayor2-sc merged 5 commits intomainfrom
kmonte/launch-gs-local-script
May 7, 2026
Merged

Add 'local' launch script for graph store mode - so we can simulate the cluster on dev boxes#587
kmontemayor2-sc merged 5 commits intomainfrom
kmonte/launch-gs-local-script

Conversation

@kmontemayor2-sc
Copy link
Copy Markdown
Collaborator

Scope of work done

Where is the documentation for this feature?: N/A

Did you add automated tests or write a test plan?

Updated Changelog.md? NO

Ready for code review?: NO

Copy link
Copy Markdown
Collaborator

@mkolodner-sc mkolodner-sc left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should this go in scripts? Looking our scripts folder currently, the majority of scripts here are to help setup GiGL. This to me feels like something that would be better in an experimental folder since I imagine this is primarily used for testing/debugging -- WDYT?

@kmontemayor2-sc
Copy link
Copy Markdown
Collaborator Author

I don't think this belongs in "experimental" as it's not experimental - it's production ready and meant to be used (potentially for experiments).

We could put it elsewhere, maybe gigl/scripts/ as this is more buisness-logic-y? Happy to put it under gigl/scripts if you prefer

Copy link
Copy Markdown
Collaborator

@mkolodner-sc mkolodner-sc left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks!

Comment thread tests/unit/scripts/launch_graph_store_local_test.py Outdated
Comment thread scripts/launch_graph_store_local.py
Comment thread scripts/launch_graph_store_local.py
Comment thread scripts/launch_graph_store_local.py
_build_cluster_spec previously hard-coded :2222 for every entry. The
per-entry ports are not consumed at runtime (cluster spec consumers in
gigl/distributed/utils/networking.py only read len(...)), but populating
them with real free ports makes the manifest representative of what
Vertex AI would produce. main() now requests 1 + world_size ports up
front: the first becomes MASTER_PORT, the rest are threaded through
build_node_processes to populate workerpool0/1/2 entries indexed by
global rank.

Also rename tests/unit/scripts -> tests/unit/scripts_test and add an
__init__.py so 'make unit_test_py' discovers the launcher test without
shadowing the top-level scripts/ package.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@kmontemayor2-sc kmontemayor2-sc marked this pull request as ready for review May 7, 2026 16:21
@kmontemayor2-sc kmontemayor2-sc added this pull request to the merge queue May 7, 2026
Merged via the queue into main with commit 655240a May 7, 2026
7 checks passed
@kmontemayor2-sc kmontemayor2-sc deleted the kmonte/launch-gs-local-script branch May 7, 2026 17:58
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.

4 participants