NVIDIA / NVIDIA/OpenShell

feat(helm): make gateway PVC size and StorageClass configurable

Open
#3,216 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

state:accepted
Dominant language
Rust
Stars
8.7k
Forks
1.3k
Avg merge
2d 11h
Merged PRs (30d)
253

Description

User Story

As a Kubernetes platform operator deploying the OpenShell gateway through Helm, I want to select the gateway PVC's size and StorageClass, so that encrypted gateway state lands on the cluster's retained stateful storage with capacity appropriate to the deployment.

Problem Statement

OpenShell Helm chart 0.0.116 hard-codes the StatefulSet volumeClaimTemplates request to 1Gi and does not set or expose storageClassName. The chart exposes workspace PVC size/class values, but not the gateway database claim.

Impact / Why This Matters

Without chart values, operators must accept the cluster's default StorageClass and fixed capacity or carry a post-render patch. This can put the gateway's encrypted credential database and control-plane state on an unintended storage tier. A later correction may require StatefulSet/PVC migration because volumeClaimTemplates fields are immutable in common upgrade paths. Post-render patches are also coupled to the template's claim order.

Proposed Design

Expose a small gateway persistence block in Helm values, for example:

persistence:
  size: 1Gi
  storageClassName: ""

The existing observable behavior should remain the default: a 1Gi claim using the cluster's default StorageClass when no values are set. Setting these values should render the requested capacity and storageClassName into the gateway StatefulSet. Naming is illustrative; the important user workflow is being able to declare both fields without a post-renderer.

Acceptance Criteria

  • The gateway PVC request size is configurable through a documented chart value.
  • The gateway PVC StorageClass is configurable through a documented chart value.
  • Omitting the new values preserves the current 1Gi/default-StorageClass behavior.
  • helm template tests cover default and overridden values.
  • Upgrade documentation notes the StatefulSet/PVC immutability implications for existing releases.

Alternatives Considered

  • Use the cluster's default StorageClass and expand the claim later. This does not guarantee the required retention/storage tier and depends on expansion support.
  • Pre-create a PVC. The StatefulSet currently owns a volumeClaimTemplate and has no existing-claim workflow.
  • Carry a Flux/Helm post-render patch. This works, but is coupled to the StatefulSet resource name and claim index, and it shifts a normal storage choice out of the chart's supported interface.

Agent Investigation

  • Verified the behavior against the released chart 0.0.116 at OCI digest sha256:df55cd1538bdfb7836834c30dfcf8373b85ffea83bbfd70d50dbe69407a0d2b3.
  • Rendering with production values still produces resources.requests.storage: 1Gi and no storageClassName until a post-render patch is applied.
  • Searched open and closed issues for Helm persistence, volumeClaimTemplates, storage size, and StorageClass; no existing issue matched this gateway PVC gap.

Checklist

  • I've reviewed existing issues and the architecture docs
  • This is a design proposal, not a "please build this" request

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by locating the gateway StatefulSet volumeClaimTemplates and the existing workspace PVC values in the Helm chart, using released chart 0.0.116 as the baseline. Run helm template with default and overridden values to verify the gateway claim size and StorageClass, then cover those cases in Helm tests and document StatefulSet/PVC immutability implications for upgrades.

Written by the indexing model from the issue text.

Assessment

Tech stack
helm, kubernetes
Domain
devops, infrastructure
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
52/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.