Skip to content

fix(ci): add --platform to aiperf-bench E2E docker build#674

Merged
mchmarny merged 2 commits into
NVIDIA:mainfrom
xdu31:fix/e2e-aiperf-bench-platform
Apr 24, 2026
Merged

fix(ci): add --platform to aiperf-bench E2E docker build#674
mchmarny merged 2 commits into
NVIDIA:mainfrom
xdu31:fix/e2e-aiperf-bench-platform

Conversation

@xdu31

@xdu31 xdu31 commented Apr 24, 2026

Copy link
Copy Markdown
Contributor

Summary

Add --platform linux/amd64 to the docker build for aiperf-bench in the E2E composite action to prevent BuildKit from resolving the wrong architecture manifest for the python multi-arch base image.

Motivation / Context

The E2E action intermittently fails building aiperf-bench with:

exec: "/bin/sh": stat /bin/sh: no such file or directory

The Go validator images (deployment, performance, conformance) are unaffected because they are COPY-only into distroless — no RUN steps, so /bin/sh is never invoked. The release workflows (on-push.yaml, on-tag.yaml) are also unaffected because they use docker/build-push-action with explicit platforms: matrix values.

Fixes #673

Type of Change

  • Bug fix (non-breaking change that fixes an issue)

Components Affected

  • .github/actions/e2e/action.yml

Testing

  • Verify E2E pipeline passes with the --platform flag

Risk Assessment

Low — additive flag on an existing docker build command. No behavioral change on correctly-configured runners; prevents failures on misconfigured ones.

Checklist

  • My code follows the project style guidelines
  • I have performed a self-review
  • Changes are limited to what's necessary for this fix

@xdu31
fix(ci): add --platform to aiperf-bench E2E docker build
7aaec51 
The aiperf-bench image build in the E2E action intermittently fails
with "/bin/sh: no such file or directory" because the bare docker build
lacks an explicit --platform flag. Without it, BuildKit may resolve the
wrong architecture manifest for the python multi-arch base image.

The Go validator images are unaffected (COPY-only, no RUN steps) and
the release workflows already use docker/build-push-action with
explicit platform matrix values.

Fixes NVIDIA#673
@xdu31 xdu31 added the bug label Apr 24, 2026
@xdu31 xdu31 requested a review from a team as a code owner April 24, 2026 18:35
@coderabbitai

coderabbitai Bot commented Apr 24, 2026

Copy link
Copy Markdown

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info
⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Enterprise

Run ID: 5c3b5d0f-abcb-45f6-996f-46d8f9a55917

📥 Commits

Reviewing files that changed from the base of the PR and between 97be223 and 7aaec51.

📒 Files selected for processing (1)
  • .github/actions/e2e/action.yml
📝 Walkthrough

Walkthrough

The E2E action's Docker build command for aiperf-bench now explicitly specifies --platform linux/amd64 when constructing the image. Previously, the build command lacked an explicit platform flag, allowing BuildKit to potentially resolve an incorrect architecture manifest for the multi-architecture python:3.12-slim base image. This manifested as intermittent /bin/sh not found errors during the pip install step in the container.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

🚥 Pre-merge checks | ✅ 4
✅ Passed checks (4 passed)
Check name Status Explanation
Title check ✅ Passed The title accurately and specifically describes the main change: adding the --platform flag to the aiperf-bench E2E docker build command.
Description check ✅ Passed The description is clearly related to the changeset, providing motivation, context, and technical rationale for the docker build platform flag addition.
Linked Issues check ✅ Passed The pull request directly addresses issue #673 by implementing the proposed fix: adding --platform linux/amd64 to the E2E docker build for aiperf-bench to resolve incorrect architecture manifest selection.
Out of Scope Changes check ✅ Passed All changes are scoped to the stated objective: the modification adds only the --platform flag to the aiperf-bench docker build in .github/actions/e2e/action.yml, with no extraneous alterations.

✏️ Tip: You can configure your own custom pre-merge checks in the settings.

✨ Finishing Touches
🧪 Generate unit tests (beta)
  • Create PR with unit tests

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

@mchmarny mchmarny enabled auto-merge (squash) April 24, 2026 18:42
@mchmarny mchmarny disabled auto-merge April 24, 2026 18:45
@mchmarny mchmarny merged commit 8eac760 into NVIDIA:main Apr 24, 2026
36 of 38 checks passed
lockwobr pushed a commit that referenced this pull request Apr 28, 2026
@xdu31 @mchmarny @lockwobr
fix(ci): add --platform to aiperf-bench E2E docker build (#674)
95ce17e 
Co-authored-by: Mark Chmarny <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@mchmarny mchmarny mchmarny approved these changes

Assignees

@xdu31 xdu31

Labels

area/ci size/XS

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

E2E: aiperf-bench docker build fails with /bin/sh: no such file or directory

2 participants

@xdu31 @mchmarny