Skip to content

chore(evidence): refresh gb200-eks + h100-gke-cos attestations#1427

Merged
mchmarny merged 1 commit into
NVIDIA:mainfrom
yuanchen8911:chore/refresh-gb200-eks-evidence
Jun 24, 2026
Merged

chore(evidence): refresh gb200-eks + h100-gke-cos attestations#1427
mchmarny merged 1 commit into
NVIDIA:mainfrom
yuanchen8911:chore/refresh-gb200-eks-evidence

Conversation

@yuanchen8911

@yuanchen8911 yuanchen8911 commented Jun 23, 2026

Copy link
Copy Markdown
Contributor

Summary

Refresh the conformance-evidence pointers for gb200-eks-ubuntu-training and h100-gke-cos-training. The recipe-evidence drift gate flagged both as stale after #1418 (aws-efa v0.5.26v0.5.29 + EKS overlay pin) changed their rendered digests. This updates each pointer to a freshly re-validated, signed bundle.

Motivation / Context

#1418 bumped aws-efa to v0.5.29, changing the rendered digest of EKS/GKE recipes that include aws-efa. The committed evidence pointers still referenced pre-bump bundles, so the drift gate reported them stale.

Re-validated on matching clusters — all phases passed:

  • gb200-eks-ubuntu-training — GB200 EKS (p6e-gb200.36xlarge, Ubuntu 24.04, K8s 1.34). Subject digest 5177d2ed…. Bundle ghcr.io/yuanchen8911/aicr-evidence:gb200-eks-ubuntu-training-a3fe1a7e7bcb, Rekor 1930094136.
  • h100-gke-cos-training — H100 GKE/COS (K8s 1.35). NCCL all-reduce 338 GB/s. Subject digest 064f27fa…. Bundle ghcr.io/yuanchen8911/aicr-evidence:h100-gke-cos-training-ee1657fedd59, Rekor 1931193974.

Related: #1418. Note: the GKE NCCL run surfaced a separate validator bug (JobSet staging image) tracked in #1430; the evidence here was captured with that worked around.

Type of Change

  • Build/CI/tooling (evidence refresh)

Component(s) Affected

  • Recipe engine / data (recipes/evidence)

Testing

aicr evidence verify recipes/evidence/gb200-eks-ubuntu-training.yaml   # bundle valid
aicr evidence verify recipes/evidence/h100-gke-cos-training.yaml       # bundle valid

Offline verification passes for both: signature ([email protected] via github oauth), manifest-hash-check, and predicate-parse all green.

@github-actions

github-actions Bot commented Jun 23, 2026

Copy link
Copy Markdown
Contributor

Recipe evidence check

Affected leaf overlays: 2

Recipe Pointer Verify Digest match
gb200-eks-ubuntu-training ✅ present ❌ bundle invalid ⚠️ skipped (verify did not surface a signed digest)
h100-gke-cos-training ✅ present ❌ bundle invalid ⚠️ skipped (verify did not surface a signed digest)

How to refresh evidence

Run on a cluster matching the recipe's criteria:

aicr snapshot -o snapshot.yaml
aicr validate \
  -r recipes/overlays/<slug>.yaml \
  -s snapshot.yaml \
  --emit-attestation ./out \
  --push ghcr.io/<your-fork>/aicr-evidence
cp ./out/pointer.yaml recipes/evidence/<slug>.yaml

This gate is warning-only and never blocks merge. See ADR-007 for the trust model.

@coderabbitai

coderabbitai Bot commented Jun 23, 2026

Copy link
Copy Markdown

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: ASSERTIVE

Plan: Enterprise

Run ID: 50014cf1-f19f-4a93-a65b-89ff732188ed

📥 Commits

Reviewing files that changed from the base of the PR and between 5a26d11 and 19e71b0.

📒 Files selected for processing (2)
  • recipes/evidence/gb200-eks-ubuntu-training.yaml
  • recipes/evidence/h100-gke-cos-training.yaml

📝 Walkthrough

Walkthrough

Both recipes/evidence/gb200-eks-ubuntu-training.yaml and recipes/evidence/h100-gke-cos-training.yaml receive single attestation entry updates. In each file, the attestedAt timestamp is replaced with a newer value, the bundle fields (digest and oci image reference) are replaced with new values, and the signer details (identity and rekorLogIndex) are updated. The issuer, predicateType, recipe, and schemaVersion fields remain unchanged.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~2 minutes

Possibly related PRs

  • NVIDIA/aicr#1166: Updates recipe-evidence-check.sh to recognize attestation entry changes in recipes/evidence/*.yaml pointer files as affected changes.

Suggested reviewers

  • mchmarny
🚥 Pre-merge checks | ✅ 4
✅ Passed checks (4 passed)
Check name Status Explanation
Title check ✅ Passed The PR title 'chore(evidence): refresh gb200-eks + h100-gke-cos attestations' directly and clearly summarizes the main change: updating attestation records for two recipes after their digests changed due to upstream dependencies.
Description check ✅ Passed The PR description is comprehensive and directly related to the changeset, explaining the motivation for the attestation updates, validation details, and verification results.
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.

✏️ 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.

@yuanchen8911 yuanchen8911 requested a review from njhensley June 23, 2026 19:04
@yuanchen8911 yuanchen8911 force-pushed the chore/refresh-gb200-eks-evidence branch from 97400d9 to ab5233e Compare June 23, 2026 19:05
@yuanchen8911 yuanchen8911 requested a review from mchmarny June 23, 2026 20:24
mchmarny
mchmarny previously approved these changes Jun 23, 2026
@mchmarny mchmarny enabled auto-merge (squash) June 23, 2026 20:25
@yuanchen8911 yuanchen8911 changed the title chore(evidence): refresh gb200-eks-ubuntu-training attestation chore(evidence): refresh gb200-eks + h100-gke-cos attestations Jun 23, 2026
@yuanchen8911 yuanchen8911 force-pushed the chore/refresh-gb200-eks-evidence branch from 8c2b818 to 5a26d11 Compare June 23, 2026 20:30
@yuanchen8911 yuanchen8911 requested a review from mchmarny June 23, 2026 21:39
Re-validated both recipes after the aws-efa v0.5.29 / eks-pin bump (NVIDIA#1418)
changed their rendered digests, and refreshed the evidence pointers to the
freshly signed bundles so the recipe-evidence drift gate no longer reports
them stale.

- gb200-eks-ubuntu-training: GB200 EKS, all phases passed; bundle
  ghcr.io/yuanchen8911/aicr-evidence:gb200-eks-ubuntu-training-a3fe1a7e7bcb
  (rekorLogIndex 1930094136)
- h100-gke-cos-training: H100 GKE/COS, all phases passed (NCCL 338 GB/s);
  bundle ghcr.io/yuanchen8911/aicr-evidence:h100-gke-cos-training-ee1657fedd59
  (rekorLogIndex 1931193974)
@yuanchen8911 yuanchen8911 force-pushed the chore/refresh-gb200-eks-evidence branch from 5a26d11 to 19e71b0 Compare June 24, 2026 00:05
@mchmarny mchmarny merged commit 9eb3ce9 into NVIDIA:main Jun 24, 2026
155 checks passed
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.

3 participants