Verbiage on Slide 6.9 is incorrect

BrianRagazzi

Slide 6.9 states in part "This identity remains with the node, regardless to which node the Pod is running on at any one time." which contradicts the linked page on k8s.io which states "StatefulSet Pods have a unique identity that is comprised of an ordinal, a stable network identity, and stable storage. The identity sticks to the Pod, regardless of which node it’s (re)scheduled on.".
Emphasis mine.

serewicz

Hello,

Thank you for the comment. The slide 6.9 points out that the scheduler will try to make the Pod stay on the same node, if that node remains Ready. More can be seen here: https://blog.thecodeteam.com/2017/08/16/technical-dive-statefulsets-deployments-kubernetes/ So yes, the Pod does stay on the same node whenever possible. Should the node itself fail the StatefulSet Pod would retain unique information. You mention a k8s.io page.

Could you let me know the full URL and about where on the page you found the contradiction. You said k8s.io, but the page I'm looking at has github and a google doc linked. I did a quick general search and found lots of possible items and want to make sure I'm looking at the same item as you are.

Regards,

BrianRagazzi

Thank you for the quick reply and the clarifying link!
The quote I used comes from https://kubernetes.io/docs/concepts/workloads/controllers/statefulset/#pod-identity.

I think I understand now that the Pods in a StatefulSet will get a unique identity and that the scheduler will attempt to recreate a given pod on the same node. However, the verbiage on slide 6.9 seems to suggest that the Pod identity remains with the node; perhaps the second sentence in the second paragraph should state that the pod identity remains with the pod instead?

serewicz

Hello,
I will definitely take a look and consider making it more clear for the next edit, which will be a bit.

Thanks again for taking the time to reach out.

Regards,