-
Notifications
You must be signed in to change notification settings - Fork 15.1k
Add prefer-align-cpus-by-uncorecache feature gate #48340
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add prefer-align-cpus-by-uncorecache feature gate #48340
Conversation
|
Welcome @wongchar! |
👷 Deploy Preview for kubernetes-io-vnext-staging processing.
|
✅ Pull request preview available for checkingBuilt without sensitive environment variables
To edit notification comments on pull requests, go to your Netlify site configuration. |
ae9f6d5 to
6989bb5
Compare
|
This commit should be rebased onto |
6989bb5 to
35e7e1d
Compare
|
Hi @sftim ! My understanding is we have time to the doc freeze and Nov 19 was the deadline to have documentation PRs submitted and ready for review (not merged!) If my understanding is right I'm willing to help to kickoff the aforementioned concepts page. |
|
@ffromani that's right. Help is welcome! |
ffromani
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM from sig-node. Technical content looks correct to me.
|
Technical lgtm provided above /lgtm |
|
@reylejano when you get the chance we just need your final review and approval for this one. Thanks! |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If we can get #48797 merged and merged into v1.32 docs, then the changes here can go there instead (which is what I recommend doing).
See #48340 (review)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Something else we can do: merge this as-is, but track some work to do after the release to make the docs more readable.
Adding this extra detail harms the quality and readability of the existing task page, so we should plan that work, but we can accept that harm and not block adding this alpha feature on having it appropriately documented.
Ideally we do block beta graduation on getting the docs redone, though. Beta features should have docs that we think are GA quality (and we ship them to find out).
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I can and will work on getting #48797 ready anyway (https://kubernetes.slack.com/archives/C0BP8PW9G/p1732284959539839). If we decide to not merge for 1.32 due to the too strict timeline, I'll rebase once the other linked content landed, and once cpumanager content is settled I'll move on to memory manager and others:
- cpumanager
- memory manager
- topology manager
- (perhaps) device manager, this component doesn't have policies so it can use some refactoring but probably doesn't belong here.
|
Waiting for docs LGTM /lgtm cancel |
b96af01 to
ace1c3b
Compare
ffromani
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
from sig-node
| : Allocate virtual cores, sometimes called hardware threads, across different physical cores (available since Kubernetes v1.31) | ||
| `distribute-cpus-across-numa` (alpha, hidden by default) | ||
| : spread CPUs across different NUMA domains, aiming for an even balance between the selected domains (available since Kubernetes v1.23) | ||
| : Spread CPUs across different NUMA domains, aiming for an even balance between the selected domains (available since Kubernetes v1.23) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this is unrelated fix I asked @wongchar to pull in
| `strict-cpu-reservation` (alpha, hidden by default) | ||
| : Prevent all the pods regardless of their Quality of Service class to run on reserved CPUs (available since Kubernetes v1.32) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Same, fixing a rebase artifact of my previous #48797
| defaultValue: false | ||
| fromVersion: "1.32" | ||
| --- | ||
| When `PreferAlignCpusByUncoreCache` is enabled while the CPU Manager Policy is set to `static`, containers within a `Guaranteed` pod will individually be aligned to an uncore cache group at a best-effort policy. This feature can optimize performance for certain cache-sensitive workloads by minimizing the cpu allocation across uncore caches. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
fixed in the last upload after rebase on top of #48797
|
/lgtm /assign @chanieljdan |
|
LGTM label has been added. Git tree hash: 8bbd308be16094fd1c9f8c8c63b695d845fb8d3b
|
|
tech and docs lgtm provided /approve |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: chanieljdan The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
Description
Add documentation for new alpha feature gate: prefer-align-cpus-by-uncorecache
Issue
kubernetes/enhancements#4800
kubernetes/kubernetes#126750