Skip to content

feat(gms): add cross-node host-tier orchestration [GMS 08/18]#11052

Open
hutm wants to merge 1 commit into
review/gms-v2-latehost-08-3-1-host-tier-engine-corefrom
review/gms-v2-latehost-08-4-host-tier-offload-validation
Open

feat(gms): add cross-node host-tier orchestration [GMS 08/18]#11052
hutm wants to merge 1 commit into
review/gms-v2-latehost-08-3-1-host-tier-engine-corefrom
review/gms-v2-latehost-08-4-host-tier-offload-validation

Conversation

@hutm

@hutm hutm commented Jun 29, 2026

Copy link
Copy Markdown
Contributor

Summary

Coordinate KV content discovery and transfer across nodes while keeping the storage and transport implementations interchangeable.

Scope

Adds cross-node transfer orchestration and validation helpers on top of the dedicated GMS KV cache manager. Optional NIXL and Mooncake implementations are deliberately deferred to position 15.

Stack

Position 8/18 in the GMS-managed KV review train. This PR is based on review/gms-v2-latehost-08-3-1-host-tier-engine-core.

Parent: #10450
Tracking: #10455 #10458

Review migration

This is the current review entry replacing historical merged PR #10480. GitHub does not allow a merged PR to be retargeted; #10480 and all of its comments and reviews remain intact as the historical record.

The train is rebased on main at aeda23b483831df2f75b697b8ccf65f4ffd9f3d6. The reordered functionality is preserved while each PR boundary is independently buildable and lintable: compatibility call sites and the engine-facing placement adapter now appear at first use; missing type imports and test markers were added; repository-pinned formatting was applied; one unused constant was removed; and every commit carries a DCO sign-off. The complete range passes git diff --check and the full pre-commit suite.

Validation

  • Includes host-tier validation and stress coverage; full engine E2E recipes are isolated in positions 10 and 13.
  • The rewritten commit replayed cleanly on the updated GMS 07/18 boundary.
  • Existing GMS 09-18 patches are exactly equivalent under git range-diff.

@hutm hutm requested review from a team as code owners June 29, 2026 17:34
@copy-pr-bot

copy-pr-bot Bot commented Jun 29, 2026

Copy link
Copy Markdown

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@datadog-official

datadog-official Bot commented Jun 29, 2026

Copy link
Copy Markdown

Pipelines

⚠️ Warnings

🚦 2 Pipeline jobs failed

Docs link check | lychee   View in Datadog   GitHub Actions

Pre Merge | pre-merge-status-check   View in Datadog   GitHub Actions

This comment will be updated automatically if new data arrives.
🔗 Commit SHA: e68878e | Docs | Give us feedback!

@hutm hutm force-pushed the review/gms-v2-latehost-08-4-host-tier-offload-validation branch from 8e65d33 to 5710076 Compare June 29, 2026 17:45
@hutm hutm force-pushed the review/gms-v2-latehost-08-3-1-host-tier-engine-core branch from 579eb84 to 3a9fe9a Compare June 29, 2026 17:45
@hutm hutm changed the title [GMS 08/18] Cross-node host-tier transfer orchestration feat(gms): add cross-node host-tier orchestration [GMS 08/18] Jun 29, 2026
@github-actions github-actions Bot added the feat label Jun 29, 2026
@hutm hutm force-pushed the review/gms-v2-latehost-08-4-host-tier-offload-validation branch from 5710076 to a5269bc Compare June 29, 2026 17:57
@hutm hutm force-pushed the review/gms-v2-latehost-08-3-1-host-tier-engine-core branch from 3a9fe9a to 4e41f5a Compare June 29, 2026 17:57
@hutm hutm force-pushed the review/gms-v2-latehost-08-4-host-tier-offload-validation branch from a5269bc to 5275c65 Compare July 5, 2026 18:50
@hutm hutm force-pushed the review/gms-v2-latehost-08-3-1-host-tier-engine-core branch from 4e41f5a to e1165e0 Compare July 5, 2026 18:50
@hutm hutm force-pushed the review/gms-v2-latehost-08-4-host-tier-offload-validation branch from 5275c65 to 674209e Compare July 5, 2026 18:59
@hutm hutm force-pushed the review/gms-v2-latehost-08-3-1-host-tier-engine-core branch from e1165e0 to 13bbb00 Compare July 5, 2026 18:59
Compose typed cross-node transfer, bootstrap, staging, tombstone, and recovery RPC domains on top of the local daemon. Transport implementations remain optional and land later.

Signed-off-by: mkhadkevich <mkhadkevich@nvidia.com>
@hutm hutm force-pushed the review/gms-v2-latehost-08-4-host-tier-offload-validation branch from 674209e to e68878e Compare July 5, 2026 19:42
@hutm hutm force-pushed the review/gms-v2-latehost-08-3-1-host-tier-engine-core branch from 13bbb00 to 9349b97 Compare July 5, 2026 19:42
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant