Skip to content

ci: bump Linux CUDA slim container to gha-convention base runner - #1075

Merged
ndizazzo merged 1 commit into
mainfrom
ci/cuda-slim-digest-bump-gha-convention
Jul 26, 2026
Merged

ci: bump Linux CUDA slim container to gha-convention base runner#1075
ndizazzo merged 1 commit into
mainfrom
ci/cuda-slim-digest-bump-gha-convention

Conversation

@ndizazzo

@ndizazzo ndizazzo commented Jul 25, 2026

Copy link
Copy Markdown
Collaborator

Summary

Pins the Linux CUDA slim matrix entry's container: image to the new digest published by Mesh-LLM/mesh-llm-runner-images #8 — the refactor to FROM ghcr.io/actions/actions-runner base + GHA convention /__e symlinks.

Before After
sha256:c5b85ef5... sha256:295341c6c9f17c9eb69281fd454bda953799406d6915f472c914fb5f024a88ed

Why

The old image was built from mcr.microsoft.com/dotnet/runtime-deps:8.0-noble and was missing the GHA convention mount point /__e/node24/bin/node. Once the ARC runner pod has a working docker CLI (baked by the new base image) and the host-path docker socket from carrack mounted at /var/run/docker.sock, the actions/runner agent execs marketplace JS actions via:

docker exec <container> /__e/node24/bin/node <runner-temp>/actions/checkout/dist/index.js

The old image lacks /__e/node24/bin/node, so this fails with:

OCI runtime exec failed: exec failed: unable to start container process: exec: "/__e/node24/bin/node": stat /__e/node24/bin/node: no such file or directory
``\]

— confirmed at `Linux CUDA slim` step 3 (`actions/checkout@v5`) on run 30146423269 job 89658307286 (in stanton-patio51 cluster build environment).

Reference: [actions/runner-container-hooks#347](https://github.com/actions/runner-container-hooks/issues/347) — exact same failure mode on K8s self-hosted DinD.

The new image inherits `ghcr.io/actions/actions-runner:latest` (Ubuntu 24.04 with built-in `runner` user, `/home/runner/externals/{node20,node24}/bin/node`, `/usr/bin/docker` CLI, and `/__e/node24` symlinks) and layers CUDA 12.9.2 + Rust + just + sccache + pnpm on top.

## Pairs with

- **stanton-patio51** commit [`31a661b`](https://github.com/ndizazzo/crusader-patio51/commit/31a661b) — bumps the corresponding self-hosted ARC runner pod image to `sha256:2dd446840d607f893ab1a747668a1554ba5cece64ec5a6dbbaa725bf3479c2f3` (self-hosted cuda12 amd64), drops the now-redundant docker-cli-installer initContainer, keeps the supplementalGroups + host-path docker socket mount.

## Verification plan

- This PR's follow-up push triggers PR Builds (this workflow).
- Expect: `Linux CUDA slim` job runs to completion (no longer errors at `actions/checkout@v5`) on stanton-patio51 self-hosted AMD64 scale-set runner.
- Other matrix cells unchanged (ROCm slim, Vulkan, macOS, Linux CPU) — they use `runs_on: [ubuntu-24.04]` (GitHub-hosted) and are not affected by this digest swap.


<!-- This is an auto-generated comment: release notes by coderabbit.ai -->

## Summary by CodeRabbit

* **Chores**
  * Updated the CUDA build environment to use a refreshed, verified container image.
  * Improved reliability and consistency for Linux CUDA build checks.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->

Pins the matrix CUDA slim container to the new public cuda12 amd64
image digest sha256:295341c6c9f17c9eb69281fd454bda953799406d6915f472c914fb5f024a88ed
from Mesh-LLM/mesh-llm-runner-images PR #8 (FROM ghcr.io/actions/actions-runner
base image refactor). The new image exposes /__e/node24/bin/node for the
actions/runner agent to exec marketplace JS actions via docker exec against
the host-path /var/run/docker.sock mounted into the self-hosted ARC runner
pod on carrack.

Prior image (sha256:c5b85ef...) had no /__e GHA-convention mount path,
causing Linux CUDA slim jobs to fail at actions/checkout@v5 with
`stat /__e/node24/bin/node: no such file or directory` even after
docker CLI became available (stanton-patio51 commit 97dd4be).

Pairs with stanton-patio51 31a661b which bumps the self-hosted ARC runner
pod image to the matching self-hosted cuda12 amd64 digest
sha256:2dd446840d607f893ab1a747668a1554ba5cece64ec5a6dbbaa725bf3479c2f3.
Together these are the downstream follow-ups to runner-images PR #8.
@coderabbitai

coderabbitai Bot commented Jul 25, 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: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 4160455f-5ee4-4573-b3e1-5fc43ca1fdad

📥 Commits

Reviewing files that changed from the base of the PR and between cdd793b and a26e6b8.

📒 Files selected for processing (1)
  • .github/workflows/pr_builds.yml

📝 Walkthrough

Walkthrough

The CUDA slim Linux build matrix entry now references an updated pinned digest for the mesh-llm-cuda-runner container image.

Changes

CUDA runner update

Layer / File(s) Summary
Update CUDA runner image pin
.github/workflows/pr_builds.yml
The CUDA slim matrix entry uses a new pinned container image digest.

Estimated code review effort: 1 (Trivial) | ~2 minutes

Suggested reviewers: i386, michaelneale

🚥 Pre-merge checks | ✅ 5
✅ Passed checks (5 passed)
Check name Status Explanation
Description Check ✅ Passed Check skipped - CodeRabbit’s high-level summary is enabled.
Title check ✅ Passed The title clearly describes the CI change to the Linux CUDA slim container and its new base runner image.
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.
✨ Finishing Touches
🧪 Generate unit tests (beta)
  • Create PR with unit tests
  • Commit unit tests in branch ci/cuda-slim-digest-bump-gha-convention

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

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

@ndizazzo
ndizazzo merged commit 7de3420 into main Jul 26, 2026
23 checks passed
@ndizazzo
ndizazzo deleted the ci/cuda-slim-digest-bump-gha-convention branch July 26, 2026 21:48
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