This issue is currently awaiting triage.

If a SIG or subproject determines this is a relevant issue, they will accept it by applying the triage/accepted label and provide further guidance.

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

/area controller-manager

@chymy: The label(s) sig/appa cannot be applied, because the repository doesn't have them.

/sig apps

/cc @ahg-g

/sig autoscaling

Pod deletion cost does not offer any guarantees on pod deletion order.

For more details, please refer https://kubernetes.io/docs/concepts/workloads/controllers/replicaset/#pod-deletion-cost

Pod deletion cost does not offer any guarantees on pod deletion order.

I can't understand why, is it because the cache update is slow?

Hi @chymy,
if you set the annotation controller.kubernetes.io/pod-deletion-cost with positive value for the pod: analysis-3-8hplt then this pod(analysis-3-fbcdl) is preferred to be deleted before pod(analysis-3-8hplt) with deletion cost.

So @chymy Can you briefly explain more about, Why do you expect the pod:analysis-3-8hplt to be deleted

So @chymy Can you briefly explain more about, Why do you expect the pod:analysis-3-8hplt to be deleted

Because I set the pod-deletion-cost annotation for analysis-3-8hplt, I plan to scale down analysis-3-8hplt first. My environment has three master nodes. I think the reason might be that the time interval between setting the annotation and scaling down is too short, causing some apiserver caches to not fully synchronize.

I set the controller.kubernetes.io/pod-deletion-cost: "-1" annotation for analysis-3-8hplt

Pod deletion cost does not offer any guarantees on pod deletion order.

Taking "best effort basis" here to mean, "support may or may not be added to older controllers like ReplicationController," is quite the stretch. "Best effort" indicates that there may be situations where it is not possible to provide a guarantee.

If the RC controller is lacking support for pod deletion costs, then a best effort has not been made. There may indeed be a timing issue, as @chymy posited. The bottom line is that "best effort basis" still requires effort; hiding behind "best effort" instead of engaging with a bug report is not an effective long-term strategy.

So @chymy Can you briefly explain more about, Why do you expect the pod:analysis-3-8hplt to be deleted

@chymy Is correct. The default pod deletion cost is 0 in the case where the annotation has not been set. Pods with lower deletion cost are preferred to be deleted before pods with higher deletion cost. Out of two replicas with the following pod deletion costs, analysis-3-8hplt should have been deleted first, with analysis-3-fbcdl as the 2nd preference:

The Kubernetes project currently lacks enough contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle stale
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale

The Kubernetes project currently lacks enough active contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle rotten
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle rotten