Hello all, i've a question on using new runtime class inside a pod while that po . . .

k8suser:
Hello all, i’ve a question on using new runtime class inside a pod while that pod is part of a deployment ! as most of examples that i found on official k8s website are about just simple Pods and not Pods that are part of deployments!

Magnus Markling:
What is your question?

k8suser:
on all examples found i saw that we can add a new runtime into a simple Pod , but how you can add that into a deployment ?

Magnus Markling:
Can’t you specify a runtimeClassName in the pod spec for the deployment just as you would for a single pod?

k8suser:
no you cannot do that and even if you do what if the pod got deleted the replicaset will trigger a new one WITHOUT the runtime class

Magnus Markling:
That’s weird. So the “pod spec” is not respected when the deployment creates new pods? Then I’m out of ideas.

k8suser:
yup i need some help here @Mumshad Mannambeth @Vijin Palazhi (kodekloud)

Hamid Zorgani:
@k8suser did u try to add it as deploy spec ?

k8suser:
yup did not work

Hamid Zorgani:
can u share the error from the pod side or its been created with default runtime class.

Hamid Zorgani:
did you install runtime class on all nodes or you use node selector to land on the specific node with the new runtime class

Hamid Zorgani:
/ kubectl get pods -o customcolumns=NAME:metadata.name,STATUS:.status.phase,RUNTIME:.spec.runtimeClassName

k8suser:
there is only 1 worker node and master node is tainted so no need to use node selectors here

Hamid Zorgani:
@k8suser could u share error details witch runtime class ect plus the output of the command above.

Hamid Zorgani:
the container engine version used too