KEP5598: update docs for Opportunistic Batching feature #52899

dom4ha · 2025-10-23T00:17:41Z

Description

Issue

Enhancements issue: kubernetes/enhancements#5598

netlify · 2025-10-23T00:17:47Z

👷 Deploy Preview for kubernetes-io-vnext-staging processing.

Name	Link
🔨 Latest commit	`97116b1`
🔍 Latest deploy log	https://app.netlify.com/projects/kubernetes-io-vnext-staging/deploys/692f75406dade20008d6b5e0

netlify · 2025-10-23T00:27:48Z

✅ Pull request preview available for checking

Built without sensitive environment variables

Name	Link
🔨 Latest commit	`97116b1`
🔍 Latest deploy log	https://app.netlify.com/projects/kubernetes-io-main-staging/deploys/692f75404f6fc40008016250
😎 Deploy Preview	https://deploy-preview-52899--kubernetes-io-main-staging.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

To edit notification comments on pull requests, go to your Netlify project configuration.

Urvashi0109 · 2025-11-17T08:21:47Z

Hello @dom4ha 👋, v1.35 Docs Team here again!

We are closing in on the deadline to get your PR ready for review before Tuesday 18th November 2025, so I'm sending a second reminder. Please take a look at the Documenting for a release - PR Ready for Review document to get your PR ready for review before the deadline.

Please also let us know once your PR is fully Ready for Review -- meaning all documentation updates are complete and it's awaiting reviewer feedback -- so we can update our tracking.

Thank you!

AnshumanTripathi · 2025-11-21T18:09:09Z

Hello @dom4ha 👋! I'm reaching out from the Docs team. Just checking in as we approach Docs Freeze on 3rd December 2025, 12:00 UTC.
This documentation appears to still be under review. To meet the Docs Freeze, this PR must have a technical review as well as lgtm and approve labels applied, without any unaddressed comments or concerns from SIG Docs. The status of this enhancement is marked as at risk for docs freeze. Thank you!

content/en/docs/concepts/scheduling-eviction/scheduler-perf-tuning.md

sanposhiho · 2025-11-26T04:06:14Z

content/en/docs/concepts/scheduling-eviction/scheduler-perf-tuning.md

+1. Scheduling profile needs to disable [default topology spread](/docs/concepts/scheduling-eviction/topology-spread-constraints/#internal-default-constraints)
+1. `IgnorePreferredTermsOfExistingPods` of [InterPodAffinityArgs](/docs/reference/config-api/kube-scheduler-config.v1/#kubescheduler-config-k8s-io-v1-InterPodAffinityArgs)
+can be set to `true` to make the batching more efficient.


Let's explain those two separately? because they are "what a cluster admin has to configure their cluster", while others are "what kind of pods can get the benefits from this feature".

So, overall, I'd prefer the explanation to be like:

This feature takes effect only for specific pods for now: - No topology spread constraints. - No DRA. - ... Also, to enable this feature in your scheduler, you have to configure the scheduler to: - Change `IgnorePreferredTermsOfExistingPods` to `true` - Set empty on default topology spread.

Good point. Broke it down into 4 categories

macsko · 2025-11-26T14:46:21Z

/sig scheduling

sanposhiho · 2025-12-01T07:21:24Z

content/en/docs/concepts/scheduling-eviction/scheduler-perf-tuning.md

+This feature takes effect only in specific conditions:
+1. Pods need to have equivalent scheduling constraints
+1. Pods need to be scheduled at the same time (cache expires afer 0.5s)
+1. Pods cannot be interleaved with other pods
+1. Only one batched pod can fit into one node (placing 2 pods on one node invalidates the cache)


I feel like it's easier to explain the basic flow first, and then explain these conditions without bullet points. Maybe like:

Suggested change

This feature takes effect only in specific conditions:

1. Pods need to have equivalent scheduling constraints

1. Pods need to be scheduled at the same time (cache expires afer 0.5s)

1. Pods cannot be interleaved with other pods

1. Only one batched pod can fit into one node (placing 2 pods on one node invalidates the cache)

Basically, this feature works like:

1. The scheduler schedules pod-1 and caches the scheduling result.

1. The scheduler schedules pod-2, 3, ... with the cached results.

1. The cache expires after 0.5 second. The scheduler schedules the next pod without cache.

Pods with equivalent scheduling constraints have to come to the scheduling cycle back to back. When the scheduler schedules a pod with different constraints, the cache is flushed at that point.

Also, I'd move 1. Only one batched pod can fit into one node (placing 2 pods on one node invalidates the cache) to the next section.

sanposhiho · 2025-12-01T07:25:18Z

content/en/docs/concepts/scheduling-eviction/scheduler-perf-tuning.md

+And specific pods that do not use:
+1. Inter pod affinity/anti-affinity
+1. Topology spread constraints
+1. DRA (cannot have any Resource Claims)


Suggested change

And specific pods that do not use:

1. Inter pod affinity/anti-affinity

1. Topology spread constraints

1. DRA (cannot have any Resource Claims)

We apply this batching scheduling to specific pods that:

1. don't have inter pod affinity/anti-affinity

1. don't have tpology spread constraints

1. don't have DRA (i.e., don't have any Resource Claims)

1. scheduled exclusively on nodes (i.e., placing more than one pods on one node invalidates the cache)

sanposhiho

/lgtm
/approve

sig-scheduling tech review

k8s-ci-robot · 2025-12-02T03:11:13Z

LGTM label has been added.

Details

Git tree hash: 384e2eea1d5e29c88040d70e1e6dcdc2bb07517c

natalisucks

review from docs, some nits, some required – thanks for this PR!

content/en/docs/concepts/scheduling-eviction/scheduler-perf-tuning.md

dom4ha · 2025-12-02T23:27:55Z

I applied all your comments.
Thanks @natalisucks for the review.

divya-mohan0209

LGTM from the docs side. Also, re-adding the approval from SIG Scheduling since that was removed in a previous commit due to stylistic changes suggested from our side.
/approve
/lgtm

k8s-ci-robot · 2025-12-03T02:37:04Z

LGTM label has been added.

Details

Git tree hash: 7247c1f099688ddc06995063030fbbf7bb33aac2

k8s-ci-robot · 2025-12-03T02:37:12Z

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: divya-mohan0209, sanposhiho

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

~~content/en/docs/OWNERS~~ [divya-mohan0209]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

k8s-ci-robot added this to the 1.35 milestone Oct 23, 2025

k8s-ci-robot added the do-not-merge/work-in-progress Indicates that a PR should not merge because it is a work in progress. label Oct 23, 2025

k8s-ci-robot added cncf-cla: yes Indicates the PR's author has signed the CNCF CLA. size/XS Denotes a PR that changes 0-9 lines, ignoring generated files. labels Oct 23, 2025

pacoxu mentioned this pull request Nov 4, 2025

Opportunistic batching kubernetes/enhancements#5598

Open

10 tasks

dom4ha force-pushed the DocsUpdateKep5598 branch from cad4cbc to d96f8a0 Compare November 18, 2025 03:12

k8s-ci-robot removed the size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files. label Nov 18, 2025

dom4ha force-pushed the DocsUpdateKep5598 branch from d96f8a0 to 13d70c4 Compare November 25, 2025 01:23

macsko reviewed Nov 25, 2025

View reviewed changes

sanposhiho reviewed Nov 26, 2025

View reviewed changes

k8s-ci-robot added the sig/scheduling Categorizes an issue or PR as relevant to SIG Scheduling. label Nov 26, 2025

github-project-automation bot added this to SIG Scheduling Nov 26, 2025

dom4ha force-pushed the DocsUpdateKep5598 branch from 3426b75 to 157bd0c Compare November 27, 2025 10:34

sanposhiho reviewed Dec 1, 2025

View reviewed changes

dom4ha force-pushed the DocsUpdateKep5598 branch 2 times, most recently from b7fd102 to f80338e Compare December 1, 2025 15:21

sanposhiho approved these changes Dec 2, 2025

View reviewed changes

k8s-ci-robot assigned sanposhiho Dec 2, 2025

k8s-ci-robot added the lgtm "Looks good to me", indicates that a PR is ready to be merged. label Dec 2, 2025

natalisucks reviewed Dec 2, 2025

View reviewed changes

helayoty moved this to Needs Final Approver in SIG Scheduling Dec 2, 2025

Docs update for Opportunistic Batching feature

97116b1

dom4ha force-pushed the DocsUpdateKep5598 branch from f80338e to 97116b1 Compare December 2, 2025 23:24

k8s-ci-robot removed the lgtm "Looks good to me", indicates that a PR is ready to be merged. label Dec 2, 2025

k8s-ci-robot requested a review from sanposhiho December 2, 2025 23:24

divya-mohan0209 reviewed Dec 3, 2025

View reviewed changes

k8s-ci-robot assigned divya-mohan0209 Dec 3, 2025

k8s-ci-robot added the lgtm "Looks good to me", indicates that a PR is ready to be merged. label Dec 3, 2025

k8s-ci-robot added the approved Indicates a PR has been approved by an approver from all required OWNERS files. label Dec 3, 2025

k8s-ci-robot merged commit 26c750f into kubernetes:dev-1.35 Dec 3, 2025
6 checks passed

github-project-automation bot moved this from Needs Final Approver to Done in SIG Scheduling Dec 3, 2025

-And specific pods that do not use:
-. Inter pod affinity/anti-affinity
-. Topology spread constraints
-. DRA (cannot have any Resource Claims)
+We apply this batching scheduling to specific pods that:
+. don't have inter pod affinity/anti-affinity
+. don't have tpology spread constraints
+. don't have DRA (i.e., don't have any Resource Claims)
+. scheduled exclusively on nodes (i.e., placing more than one pods on one node invalidates the cache)

KEP5598: update docs for Opportunistic Batching feature #52899

KEP5598: update docs for Opportunistic Batching feature #52899

Conversation

dom4ha commented Oct 23, 2025

Description

Issue

Uh oh!

netlify bot commented Oct 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

👷 Deploy Preview for kubernetes-io-vnext-staging processing.

Uh oh!

netlify bot commented Oct 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

✅ Pull request preview available for checking

Uh oh!

Urvashi0109 commented Nov 17, 2025

Uh oh!

AnshumanTripathi commented Nov 21, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

sanposhiho Nov 26, 2025

Choose a reason for hiding this comment

Uh oh!

dom4ha Nov 27, 2025

Choose a reason for hiding this comment

Uh oh!

macsko commented Nov 26, 2025

Uh oh!

sanposhiho Dec 1, 2025

Choose a reason for hiding this comment

Uh oh!

sanposhiho Dec 1, 2025

Choose a reason for hiding this comment

Uh oh!

sanposhiho left a comment

Choose a reason for hiding this comment

Uh oh!

k8s-ci-robot commented Dec 2, 2025

Uh oh!

natalisucks left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

dom4ha commented Dec 2, 2025

Uh oh!

divya-mohan0209 left a comment

Choose a reason for hiding this comment

Uh oh!

k8s-ci-robot commented Dec 3, 2025

Uh oh!

k8s-ci-robot commented Dec 3, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

11 participants

netlify bot commented Oct 23, 2025 •

edited

Loading

netlify bot commented Oct 23, 2025 •

edited

Loading