https://rancher.com/ logo
Join SlackCommunities
Powered by
This message was deleted.
# k3s
a
This message was deleted.
c
Did you add nvidia runtimeClass resources to the cluster, and set the runtimeClassName for your pod?
You must still add a RuntimeClass definition to your cluster, and deploy Pods that explicitly request the appropriate runtime by setting 
runtimeClassName: nvidia
in the Pod spec
just because the runtimes are there, doesn’t mean that your pods are using them. You have to specifically ask for it.
w
So if I want the NVIDIA Device Plugin pods to run correctly, then I must set the 
runtimeClassName: nvidia
via a configmap (or otherwise) in the Helm install?
The Helm cmdline is:
Copy code
# helm upgrade -i nvdp nvdp/nvidia-device-plugin \
    --version=0.14.3 \
    --namespace nvidia-device-plugin \
    --create-namespace \
    --set-file config.map.nvdp-config=/root/nvdp-config.yaml
with the configmap currently as:
Copy code
# cat nvdp-config.yaml
version: v1
flags:
  migStrategy: "none"
  failOnInitError: true
  nvidiaDriverRoot: "/"
  plugin:
    passDeviceSpecs: false
    deviceListStrategy: envvar
    deviceIDStrategy: uuid
We did add the RuntimeClass to the cluster as so:
c
I’m not sure how to set the runtimeClassName via that chart, but I think you’re on the right path
w
FYI @creamy-pencil-82913 -- 
--set runtimeClassName=nvidia
does the trick 🙂
(they support alternate runtimes in their Helm chart thankfully)
79 Views
Open in Slack
PreviousNext
Powered by