Back-off pulling image "alpine:3.10"

[lizu18xz]

What happened?

What command can be used to modify the image name in initContainers?

What did you expect to happen?

modify the image name in initContainers

Environment

Kubernetes version:

$ kubectl version

Training Operator version:

$ kubectl get pods -n kubeflow -l control-plane=kubeflow-training-operator -o jsonpath="{.items[*].spec.containers[*].image}"
kubeflow/training-operator:v1-855e096

Training Operator Python SDK version:

$ pip show kubeflow-training

Impacted by this bug?

Give it a 👍 We prioritize the issues with most 👍

[wucb]

Modify training-operator config
set --pytorch-init-container-image=xxx in commd

[tenzen-y]

Modify training-operator config set --pytorch-init-container-image=xxx in commd

Yes, you can pass an arbitrary initContainer image via the operator command:

training-operator/cmd/training-operator.v1/main.go

Lines 98 to 99 in ff19a10

	flag.StringVar(&config.Config.PyTorchInitContainerImage, "pytorch-init-container-image",
	config.PyTorchInitContainerImageDefault, "The image for pytorch init container")

/remove-kind bug
/kind question

[lizu18xz]

training-operator config

training-operator config Where is it? I found deploymenty.yaml 。But I don't know how to set pytorch-init-container-image

[lizu18xz]

Modify training-operator config set --pytorch-init-container-image=xxx in commd

/remove-kind bug

Modify training-operator config set --pytorch-init-container-image=xxx in commd

Yes, you can pass an arbitrary initContainer image via the operator command:

training-operator/cmd/training-operator.v1/main.go

Lines 98 to 99 in ff19a10

	flag.StringVar(&config.Config.PyTorchInitContainerImage, "pytorch-init-container-image",
	config.PyTorchInitContainerImageDefault, "The image for pytorch init container")

/remove-kind bug /kind question

How to use this command? I only found deploymenty.yaml.

I couldn't find the training operator config, use : kubectl get cm -A
Can you explain it in detail? Thank you

[Syulin7]

@lizu18xz

kubectl edit deploy training-operator -n kubeflow

    spec:
      containers:
      - command:
        - /manager
        - --pytorch-init-container-image=your-image

[andreyvelich]

/remove-label lifecycle/needs-triage