Skip to content

About

NVIDIA OpenShell for self-hosted Kubernetes research clusters: Kata microVM sandboxes, warm pools, cluster labels, signed images

Resources

Code of conduct

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

 
 

Latest commit

 

History

1,915 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

EnrichedLab OpenShell

A downstream build of NVIDIA OpenShell for self-hosted Kubernetes clusters that run research and operations agents. It tracks upstream releases and adds what a shared research cluster needs: microVM sandboxes, fast starts for many concurrent agents, cluster-owned labels, and a verifiable record of every run.

Upstream's own documentation is in README.md and at docs.nvidia.com/openshell. This fork is not affiliated with or endorsed by NVIDIA.

Base

Upstream release v0.1.5
Default branch enrichedlab/main (upstream release plus the changes below)
Images ghcr.io/richardstoker-usda/enrichedlab-openshell/{gateway,supervisor,sandbox,cli}, linux/amd64 and linux/arm64
Chart oci://ghcr.io/richardstoker-usda/enrichedlab-openshell/charts/helm-chart
License Apache 2.0, as upstream

Changes from upstream

Change Status
Images, the static CLI and the Helm chart built in this repository's CI, each with a signed SLSA build provenance attestation available
Merge-based tracking of upstream releases, with a preview branch on upstream main available
Sandbox Pods run in Kata Containers microVMs (Cloud Hypervisor) through a RuntimeClass in testing; configuration only, no code change
Agent Sandbox warm pools: claim a pre-started sandbox before creating a cold one in progress
Operator-fixed labels and annotations on every Sandbox, workload and supervisor Pod and workspace claim (fixed_pod_labels, fixed_pod_annotations), for clusters whose admission, telemetry or retention policy requires them available
Supervisor telemetry to the gateway's collector, a separate collector, or none (supervisor_telemetry), without forwarding the gateway's collector credentials available
Retention class and expiry on every sandbox object in progress
A run receipt per sandbox: policy and image digests, runtime, OCSF event counts, exit status, output checksums in progress

Changes that are useful beyond one site are offered upstream; this table shrinks as they land.

Verify an image

gh attestation verify oci://ghcr.io/richardstoker-usda/enrichedlab-openshell/gateway@sha256:<digest> \
  --repo RichardStoker-USDA/enrichedlab-openshell

Deploy by digest, never by tag.

Maintaining

Branch model, sync script and release tags: FORK.md.

About

NVIDIA OpenShell for self-hosted Kubernetes research clusters: Kata microVM sandboxes, warm pools, cluster labels, signed images

Resources

Code of conduct

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages