Add priority flag to kube-proxy on Windows hosts

[ravisantoshgudimetla]

What would you like to be added:

Add priority flag to kube-proxy so that end-user can specify priority

Why is this needed:

On Windows hosts, there is a chance that the kube-proxy process won't get enough CPU cycles because of the default priority set for the Windows process.

xref: #95735 (comment)

Additional context:

xref: #95735

cc @marosset @jsturtevant @JocelynBerrendonner @aravindhp

Let me know if you guys are fine with it. I can create a PR for the same.

[ravisantoshgudimetla]

/sig network
/sig windows

[JocelynBerrendonner]

@sbangari, FYI

[thockin]

Why is this in kube-proxy and not whatever starts kube-proxy? I'd prefer NOT to add flags if we can avoid it...

[jayunit100]

/assign

[jayunit100]

re flag vs config -- The important thing is that we bump the priority, luckily its easy enough to do, via the golang windows library in the PR... , bc kube-proxy running at low-priority on windows causes alot of problems.... wether done by a flag or not is probably less of a big deal, right @ravisantoshgudimetla

Is there a reasonable default we can use, if its <= Normal priority? that might satisfy Tims suggestion

[marosset]

@thockin There are a few different ways to start processes with non-normal priority but all have some drawbacks (mostly due to how priority can only be set on running processes in Windows)

• start { /abovenormal | /high | /realtime } creates the process in a new shell Windows and makes it hard to capture / redirect stdout/stderr
• Using powershell to start the process and then set the priority has similar issues but not all environments/nodes may have powershell installed
• PsExec by SysInternals requires introducing a new tool and accepting a EULA
• Setting scheduled tasks to find specific processes and raise their priority is finicky

I think that adding an extra flag would be the solution that is the most re-usable across all of the different solutions already out there for setting up the nodes.

@jayunit100 / @ravisantoshgudimetla I'm asking some folks in the Windows container platform team if they think using AboveNormal or High would be the best priority class to recommend.

[marosset]

/milestone v1.21
/triage accepted
/priority important-soon

[jayunit100]

• @marosset impl detail, can we do this via the kube-proxy config map api ?
• spoke with @ravisantoshgudimetla and i think hes ok with this as well.
• @alejandrox1 and i can help on that front if minor refactors needed since we're digging cleaning up the no-op situation around kube proxy config flags anyways (see KEP below)

TLDR: kube-proxy flags, specifically, are not used at all anymore ... theyre in a zombie state and we're working to remove them, (with exception of master-override, config, and maybe a couple other inherited ones)

See kubernetes/enhancements#2158 if interested in the specifics, but some history -- it looks like the flags in kube-proxy are largely "appendages of the past"

[marosset]

@marosset impl detail, can we do this via the kube-proxy config map api ?

Setting this via the config map API would be fine - I think the important detail is to have the kube-proxy process set its own process priority. I don't have a preference on how kube-proxy is instructed to do so.

[ravisantoshgudimetla]

@marosset - Thank you for elaborating on why the kube-proxy process has to set it owns priority. I just want to add one more point related to having external entity set the priority, as community we need to give specification on how to set priority of the kube-proxy and kubelet if we're relying on external entities to do so.

@jayunit100 / @ravisantoshgudimetla I'm asking some folks in the Windows container platform team if they think using AboveNormal or High would be the best priority class to recommend.

I believe you're referring to recommending in the docs.

spoke with @ravisantoshgudimetla and i think hes ok with this as well.

Setting this via the config map API would be fine - I think the important detail is to have the kube-proxy process set its own process priority. I don't have a preference on how kube-proxy is instructed to do so.

+1

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-contributor-experience at kubernetes/community.
/lifecycle stale

[desponda]

Hi 👋 Bug Triage here, code freeze is tomorrow! This issue hasn't been updated in a while and has been marked for 1.21 milestone, is there a plan to continue working on this @immuzz

[desponda]

/milestone clear
@immuzz will this be targeted for a future milestone?

[danwinship]

/area kube-proxy

[LappleApple]

/help
/good-first-issue

[k8s-ci-robot]

@LappleApple:
This request has been marked as suitable for new contributors.

Please ensure the request meets the requirements listed here.

If this request no longer meets these requirements, the label can be removed
by commenting with the /remove-good-first-issue command.

In response to this:

/help
/good-first-issue

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.

[fejta-bot]

Stale issues rot after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.

If this issue is safe to close now please do so with /close.

Send feedback to sig-contributor-experience at kubernetes/community.
/lifecycle rotten

[fejta-bot]

Rotten issues close after 30d of inactivity.
Reopen the issue with /reopen.
Mark the issue as fresh with /remove-lifecycle rotten.

Send feedback to sig-contributor-experience at kubernetes/community.
/close

[k8s-ci-robot]

@fejta-bot: Closing this issue.

In response to this:

Rotten issues close after 30d of inactivity.
Reopen the issue with /reopen.
Mark the issue as fresh with /remove-lifecycle rotten.

Send feedback to sig-contributor-experience at kubernetes/community.
/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.

[aravindhp]

/reopen

This is still required.

[k8s-ci-robot]

@aravindhp: Reopened this issue.

In response to this:

/reopen

This is still required.

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.

[fejta-bot]

Rotten issues close after 30d of inactivity.
Reopen the issue with /reopen.
Mark the issue as fresh with /remove-lifecycle rotten.

Send feedback to sig-contributor-experience at kubernetes/community.
/close

[k8s-ci-robot]

@fejta-bot: Closing this issue.

In response to this:

Rotten issues close after 30d of inactivity.
Reopen the issue with /reopen.
Mark the issue as fresh with /remove-lifecycle rotten.

Send feedback to sig-contributor-experience at kubernetes/community.
/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.