Skip to content

README.md

Latest commit

 

History

History
64 lines (48 loc) · 2.38 KB

README.md

File metadata and controls

64 lines (48 loc) · 2.38 KB
Raw

Task Helm Chart Sample

Helm chart that deploys the Task Simple Sample container as a Kubernetes Job. Use it to validate that a self-hosted NVCF cluster can launch a task from a Helm artifact.

Prerequisites

Build and push the task-simple-sample container to an OCI registry your self-hosted NVCF cluster can access. See task-simple-sample/README.md for the build and push flow.

Package and upload the chart

helm package task-helmchart-test/

Push the resulting task-helmchart-test-<version>.tgz to a chart registry your cluster can pull from (for example nvcr.io) and register Helm pull credentials with nvcf-cli registry-credential add --artifact-type HELM.

The chart declares CPU and memory requests and limits because self-managed NVCF validates resource limits before admitting Helm task workloads. Override resources in a values file if the task container needs different amounts.

Launch on self-hosted NVCF

Resolve the cluster gateway and generate an invocation API key via nvcf-cli:

export GATEWAY_ADDR=$(kubectl get gateway nvcf-gateway -n envoy-gateway -o jsonpath='{.status.addresses[0].value}')
export NVCF_API_KEY=$(nvcf-cli api-key generate --description "task-helmchart-sample" --json | jq -r .apiKey)
export ORG_ID=<your-org-id>

Submit the task with a Helm chart reference through the NVCT API:

curl --request POST \
  --url "http://${GATEWAY_ADDR}/v2/orgs/${ORG_ID}/nvct/tasks" \
  --header "Host: api.${GATEWAY_ADDR}" \
  --header "Authorization: Bearer ${NVCF_API_KEY}" \
  --header "Content-Type: application/json" \
  --data '{
    "name": "task-helmchart-sample",
    "helmChart": "<your-registry>/<namespace>/task-helmchart-test:<version>",
    "gpuSpecification": {
      "gpu": "T10",
      "instanceType": "g6.full",
      "backend": "GFN"
    },
    "maxRuntimeDuration": "PT1H",
    "maxQueuedDuration": "PT2H",
    "terminationGracePeriodDuration": "PT15M",
    "resultHandlingStrategy": "NONE"
  }'

Local smoke test

To deploy the chart directly against a local Kubernetes cluster without NVCF:

kubectl create secret docker-registry <image-pull-secret-name> \
  --docker-server=<your-registry> \
  --docker-username=<user> \
  --docker-password=<pass>

helm install <release-name> task-helmchart-test/ \
  --set ngcImagePullSecretName=<image-pull-secret-name>