Documentation about imagePullPolicy confuses readers

[rinocata]

The section on imagePullPolicy seems to be misleading. Always seems to, well, always pull images no matter what despite what the description says.

[k8s-ci-robot]

@rinocata: This issue is currently awaiting triage.

SIG Docs takes a lead on issue triage for this website, but any Kubernetes member can accept issues by applying the triage/accepted label.

The triage/accepted label can be added by org members by writing /triage accepted in a comment.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[mengjiao-liu]

page: https://kubernetes.io/docs/concepts/containers/images/#image-pull-policy

/kind support

[utkarsh-singh1]

/language en

Hi @rinocata, always value of imagePullPolicy ensures that the latest image of container is present every time you create, deploy or run the pod.

You can find more details about the default value of imagePullPolicy here - https://kubernetes.io/docs/concepts/containers/images/#imagepullpolicy-defaulting

[rinocata]

You seem to be correct @utkarsh-singh1 but the page you linked contradicts the one @mengjiao-liu (thanks) posted.

If the kubelet has a container image with that exact digest cached locally, the kubelet uses its cached image

This seems to be the behavior of IfNotPresent, not Always.

[rinocata]

Also, it's not dependent on container runtime, same behavior occurs with containerd and CRI-O. My guess is simply unverified or outdated information.

[sftim]

Also, it's not dependent on container runtime, same behavior occurs with containerd and CRI-O. My guess is simply unverified or outdated information.

Although two container runtimes both behave a particular way, that doesn't mean we can conclude they all behave that way.

Issue 37926 is very similar to this.

/remove-kind support
/triage duplicate

[rinocata]

Fair enough, the documentation is misleading in any case.

[sftim]

/retitle Documentation about imagePullPolicy confuses readers

[sftim]

I'd like to see some evidence that the existing text is factually incorrect. I'm very willing to accept that it's not easy to understand, but I believe that it's not actually wrong.

[rinocata]

Would kubectl describe pod logs constitute evidence?

[sftim]

Would kubectl describe pod logs constitute evidence?

The way I'm imagining this, you'd also need to show the state of the image cache for the container runtime used on that node before the Pod was scheduled there.

[sftim]

Please add any evidence to #37926 (which this issue pretty much duplicates)

[rinocata]

show the state of the image cache for the container runtime

Would crictl output be enough?

[sftim]

Let's continue discussion in issue #37926

[mehabhalodiya]

Per this comment, I am closing this issue. Thanks!

/close

[k8s-ci-robot]

@mehabhalodiya: Closing this issue.

In response to this:

Per this comment, I am closing this issue. Thanks!

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.