Skip to content

build: bump UCX to v1.21.0 for v1.4.0 (OPS-7781) - #12401

Merged
dagil-nvidia merged 4 commits into
mainfrom
pvijayakrish/bump-ucx-version
Jul 30, 2026
Merged

build: bump UCX to v1.21.0 for v1.4.0 (OPS-7781)#12401
dagil-nvidia merged 4 commits into
mainfrom
pvijayakrish/bump-ucx-version

Conversation

@pvijayakrish

@pvijayakrish pvijayakrish commented Jul 29, 2026

Copy link
Copy Markdown
Contributor
  • Pin nixl_ucx_ref to immutable tag v1.21.0 (was branch v1.20.x)
  • Drop vestigial nixl_ucx_efa_ref (consumed nowhere; ancestor of v1.21.0)
  • Bump UCX compliance attribution in lockstep (native_packages.yaml)
  • Remove Intel llm-scaler XPU ucx patch: both hunks upstream since v1.21.0, git apply fails fatally on the new tag

Open in Devin Review

Summary by CodeRabbit

  • Chores
    • Updated the bundled UCX component to version 1.21.0.
    • Simplified build configuration by relying on the upstream UCX support for Intel GPU data transfers.
    • Removed obsolete patch handling and an outdated UCX-specific configuration override.
    • Preserved existing licensing, source, image, and related dependency settings.

- Pin nixl_ucx_ref to immutable tag v1.21.0 (was branch v1.20.x)
- Drop vestigial nixl_ucx_efa_ref (consumed nowhere; ancestor of v1.21.0)
- Bump UCX compliance attribution in lockstep (native_packages.yaml)
- Remove Intel llm-scaler XPU ucx patch: both hunks upstream since v1.21.0,
  git apply fails fatally on the new tag
@pvijayakrish
pvijayakrish requested a review from a team as a code owner July 29, 2026 22:45
@coderabbitai

coderabbitai Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

Review Change Stack

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info
⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 8e696637-3fd7-4ca1-96e9-16b0404b4e85

📥 Commits

Reviewing files that changed from the base of the PR and between cde3ca9 and bcae0b4.

📒 Files selected for processing (3)
  • container/compliance/native_packages.yaml
  • container/context.yaml
  • container/templates/wheel_builder.Dockerfile

Walkthrough

UCX references are updated to v1.21.0, the NIXL UCX EFA pin is removed, and the wheel builder’s obsolete XPU patch download and application steps are replaced with upstream availability comments.

Changes

UCX upgrade

Layer / File(s) Summary
UCX version pinning
container/compliance/native_packages.yaml, container/context.yaml
The UCX package and CUDA 13 NIXL UCX reference now use v1.21.0; the NIXL UCX EFA reference is removed.
Obsolete patch workflow removal
container/templates/wheel_builder.Dockerfile
The build no longer downloads or conditionally applies the XPU UCX patch and documents its upstream availability since UCX v1.21.0.

Estimated code review effort: 2 (Simple) | ~10 minutes

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name Status Explanation Resolution
Description check ⚠️ Warning The description lists the changes but does not follow the required template or include the required Related Issues section. Add the required Overview, Details, and Where should the reviewer start? sections, and include the Related Issues section with the issue link or no-issue confirmation.
✅ Passed checks (4 passed)
Check name Status Explanation
Title check ✅ Passed The title clearly states the UCX bump to v1.21.0 and matches the main change.
Docstring Coverage ✅ Passed No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check ✅ Passed Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check ✅ Passed Check skipped because no linked issues were found for this pull request.

Comment @coderabbitai help to get the list of available commands.

@devin-ai-integration devin-ai-integration Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Devin Review found 1 potential issue.

Open in Devin Review

Comment thread container/templates/wheel_builder.Dockerfile
@datadog-official

datadog-official Bot commented Jul 29, 2026

Copy link
Copy Markdown

🎯 Code Coverage (details)
Patch Coverage: 100.00%
Overall Coverage: 39.49% (-8.64%)

This comment will be updated automatically if new data arrives.
🔗 Commit SHA: e6ee8ea | Docs | Datadog PR Page | Give us feedback!

pvijayakrish and others added 2 commits July 29, 2026 16:53
The dynamo-runtime image build in PR CI goes through dynamo-pipeline.yml,
whose build_timeout_minutes default of 30 is half the 60 min every other
framework build gets directly from pr.yaml. The build now takes slightly
over 30 min, so the "Build and Push Image" step gets cancelled at exactly
the timeout. Raise the default to 60 to match the sibling builds; this
also covers post-merge-ci.yml (nightly already overrides to 120).

Signed-off-by: pvijayakrish <pvijayakrish@nvidia.com>
@pvijayakrish
pvijayakrish enabled auto-merge (squash) July 30, 2026 01:54
@dagil-nvidia
dagil-nvidia disabled auto-merge July 30, 2026 01:57
@dagil-nvidia
dagil-nvidia merged commit 4bde30c into main Jul 30, 2026
105 of 107 checks passed
@dagil-nvidia
dagil-nvidia deleted the pvijayakrish/bump-ucx-version branch July 30, 2026 01:57
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.

2 participants