Skip to content

fix(helm): propagate common split service settings - #2537

Open
charlesbluca wants to merge 2 commits into
NVIDIA:mainfrom
charlesbluca:split-security-settings
Open

fix(helm): propagate common split service settings#2537
charlesbluca wants to merge 2 commits into
NVIDIA:mainfrom
charlesbluca:split-security-settings

Conversation

@charlesbluca

Copy link
Copy Markdown
Collaborator

Description

Fix split-topology Helm rendering so common service.* settings propagate to the gateway, realtime, and batch workloads.

The split Deployment branch was missing the standalone renderers common metadata, scheduling, container security, environment, volume, and shutdown settings. It also defaulted termination grace periods to 15 seconds and left worker wait-for-gateway init containers without the configured container security context.

This change restores parity, adds a sentinel-based Helm render regression test, and updates Helm, OpenShift, and release-note documentation.

Validation

  • helm template with split-mode sentinel values and manifest assertions.
  • helm lint nemo_retriever/helm --set nims.enabled=false --set serviceConfig.vectordb.enabled=false --set topology.mode=split.
  • git -c core.whitespace=cr-at-eol diff --check.
  • Not run: pytest and strict MkDocs build are unavailable in this environment.

Checklist

  • I am familiar with the Contributing Guidelines.
  • New or existing tests cover these changes.
  • The documentation is up to date with these changes.

@charlesbluca
charlesbluca marked this pull request as ready for review August 17, 2026 22:15
@charlesbluca
charlesbluca requested review from a team as code owners August 17, 2026 22:15
@charlesbluca
charlesbluca requested a review from jioffe502 August 17, 2026 22:15
@greptile-apps

greptile-apps Bot commented Aug 17, 2026

Copy link
Copy Markdown
Contributor

Greptile Summary

The PR restores common service.* settings across split-topology gateway, realtime, and batch Deployments.

  • Propagates Pod metadata, scheduling, security, environment, volume, and shutdown settings.
  • Applies the configured container security context to worker gateway-wait init containers.
  • Adds Helm rendering regression coverage and updates Helm, OpenShift, and release-note documentation.

Confidence Score: 5/5

The PR appears safe to merge.

No blocking failure remains.

Important Files Changed

Filename Overview
nemo_retriever/helm/templates/deployment.yaml Propagates common service settings into all split-topology workloads without an eligible follow-up defect.
nemo_retriever/tests/test_helm_split_common_service_settings.py Adds sentinel-based assertions covering common split Deployment settings and worker init-container security contexts.
nemo_retriever/helm/README.md Documents the common service settings inherited by split-topology workloads.
nemo_retriever/helm/openshift.md Clarifies security-context propagation for split service and gateway-wait containers on OpenShift.
docs/docs/extraction/releasenotes.md Records the split-topology Helm rendering correction.

Reviews (2): Last reviewed commit: "style: format split Helm regression test" | Re-trigger Greptile

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.

1 participant