Skip to content
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.

Sign up for GitHub

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

Jump to bottom

fix: set ForceAttemptHTTP2 to increase ARM throttling limit #3006

Merged
merged 1 commit into from Jan 3, 2023
Merged

fix: set ForceAttemptHTTP2 to increase ARM throttling limit #3006

merged 1 commit into from Jan 3, 2023

Conversation

umagnus
Copy link
Contributor

@umagnus umagnus commented Dec 30, 2022 •
edited by andyzhangx

What type of PR is this?

/kind bug

What this PR does / why we need it:

set ForceAttemptHTTP2 as false to respect custom dialer, after setting as false, the RoleInstance distribution is balanced a lot, this could also reduce ARM/VMSS throttling since ARM RoleInstance handling is distributed.

While HTTP/2 has many advantages abstractly, we have not been able to put in in mainline use in the control plane so far primarily because it doesn’t mesh well with the load-balanced architecture. You’d have to invest quite a bit in connection health and load-balancing management on the client side to get the robustness of HTTP/1.1 with lots of concurrent connections. 
To nail this down, try forcing your client to use HTTP/1.1 and see if the requests start getting distributed across many ARM FEs.
  • ForceAttemptHTTP2 as true
RoleInstance userAgent count_
FrontdoorWeb-vmss-fdweb_213 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 13,930
FrontdoorWeb-vmss-fdweb_107 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 12,190
FrontdoorWeb-vmss-fdweb_381 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 6,534
FrontdoorWeb-vmss-fdweb_285 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 599
FrontdoorWeb-vmss-fdweb_234 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 145
FrontdoorWeb-vmss-fdweb_300 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 100
FrontdoorWeb-vmss-fdweb_253 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 100
FrontdoorWeb-vmss-fdweb_186 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 100
FrontdoorWeb-vmss-fdweb_10 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 68
FrontdoorWeb-vmss-fdweb_122 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 7
  • ForceAttemptHTTP2 as false
RoleInstance userAgent count_
FrontdoorWeb-vmss-fdweb_134 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 1,139
FrontdoorWeb-vmss-fdweb_376 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 1,100
FrontdoorWeb-vmss-fdweb_66 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 979
FrontdoorWeb-vmss-fdweb_186 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 918
FrontdoorWeb-vmss-fdweb_278 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 886
FrontdoorWeb-vmss-fdweb_370 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 881
FrontdoorWeb-vmss-fdweb_85 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 829
FrontdoorWeb-vmss-fdweb_200 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 739
FrontdoorWeb-vmss-fdweb_227 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 680
FrontdoorWeb-vmss-fdweb_192 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 657
FrontdoorWeb-vmss-fdweb_153 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 647
FrontdoorWeb-vmss-fdweb_206 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 646
FrontdoorWeb-vmss-fdweb_36 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 615
FrontdoorWeb-vmss-fdweb_55 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 601
FrontdoorWeb-vmss-fdweb_234 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 585
FrontdoorWeb-vmss-fdweb_338 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 551
FrontdoorWeb-vmss-fdweb_142 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 435
FrontdoorWeb-vmss-fdweb_346 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 414
FrontdoorWeb-vmss-fdweb_221 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 381
FrontdoorWeb-vmss-fdweb_159 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 320
FrontdoorWeb-vmss-fdweb_276 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 309
FrontdoorWeb-vmss-fdweb_237 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 305
FrontdoorWeb-vmss-fdweb_318 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 296
FrontdoorWeb-vmss-fdweb_213 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 288
FrontdoorWeb-vmss-fdweb_295 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 287
FrontdoorWeb-vmss-fdweb_255 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 269
FrontdoorWeb-vmss-fdweb_46 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 268
FrontdoorWeb-vmss-fdweb_31 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 253
FrontdoorWeb-vmss-fdweb_341 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 252
FrontdoorWeb-vmss-fdweb_368 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 246
FrontdoorWeb-vmss-fdweb_290 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 237
FrontdoorWeb-vmss-fdweb_10 Go/go1.19.4 (amd64-linux) go-autorest/v14.2.1 disk.csi.azure.com/v1.26.0 OSS-kubectl 236

Which issue(s) this PR fixes:

Fixes #

Special notes for your reviewer:

Does this PR introduce a user-facing change?

fix: set ForceAttemptHTTP2 to increase ARM throttling limit

Additional documentation e.g., KEPs (Kubernetes Enhancement Proposals), usage docs, etc.:

fix: set ForceAttemptHTTP2 to increase ARM throttling limit

@k8s-ci-robot k8s-ci-robot added kind/bug Categorizes issue or PR as related to a bug. do-not-merge/release-note-label-needed Indicates that a PR should not merge because it's missing one of the release note labels. cncf-cla: yes Indicates the PR's author has signed the CNCF CLA. labels Dec 30, 2022
@netlify
Copy link

netlify bot commented Dec 30, 2022 •
edited

✅ Deploy Preview for kubernetes-sigs-cloud-provide-azure canceled.

Name Link
🔨 Latest commit 2aa57e3
🔍 Latest deploy log https://app.netlify.com/sites/kubernetes-sigs-cloud-provide-azure/deploys/63ae963a8c0f3b000897e137

@k8s-ci-robot
Copy link
Contributor

Hi @umagnus. Thanks for your PR.

I'm waiting for a kubernetes-sigs member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

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.

@k8s-ci-robot k8s-ci-robot added the needs-ok-to-test Indicates a PR that requires an org member to verify it is safe to test. label Dec 30, 2022
@k8s-ci-robot k8s-ci-robot added the size/XS Denotes a PR that changes 0-9 lines, ignoring generated files. label Dec 30, 2022
@coveralls
Copy link

Coverage Status

Coverage: 79.389%. Remained the same when pulling 2aa57e3 on umagnus:reset-ForceAttemptHTTP2 into 9863f3e on kubernetes-sigs:master.

@andyzhangx
Copy link
Member

/ok-to-test

@k8s-ci-robot k8s-ci-robot added ok-to-test Indicates a non-member PR verified by an org member that is safe to test. and removed needs-ok-to-test Indicates a PR that requires an org member to verify it is safe to test. labels Dec 30, 2022
@andyzhangx andyzhangx changed the title set ForceAttemptHTTP2 as false to respect custom dialer fix: set ForceAttemptHTTP2 as false to respect custom dialer Dec 30, 2022
@k8s-ci-robot k8s-ci-robot added release-note Denotes a PR that will be considered when it comes time to generate release notes. and removed do-not-merge/release-note-label-needed Indicates that a PR should not merge because it's missing one of the release note labels. labels Dec 30, 2022
@andyzhangx
Copy link
Member

/retest

andyzhangx
andyzhangx approved these changes Jan 3, 2023
View reviewed changes
Copy link
Member

@andyzhangx andyzhangx left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

/lgtm

@k8s-ci-robot k8s-ci-robot added the lgtm "Looks good to me", indicates that a PR is ready to be merged. label Jan 3, 2023
@k8s-ci-robot
Copy link
Contributor

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: andyzhangx, umagnus

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

@k8s-ci-robot k8s-ci-robot added the approved Indicates a PR has been approved by an approver from all required OWNERS files. label Jan 3, 2023
@nilo19
Copy link
Contributor

nilo19 commented Jan 3, 2023 •
edited

Do we need to test the performance of the cloud provider as well? As this config is globally false, not false for csi drivers only.

@andyzhangx
Copy link
Member

andyzhangx commented Jan 3, 2023 •
edited

While HTTP/2 has many advantages abstractly, we have not been able to put in in mainline use in the control plane so far primarily because it doesn’t mesh well with the load-balanced architecture. You’d have to invest quite a bit in connection health and load-balancing management on the client side to get the robustness of HTTP/1.1 with lots of concurrent connections.
To nail this down, try forcing your client to use HTTP/1.1 and see if the requests start getting distributed across many ARM FEs.

@nilo19 actually the auto trust(proxy) used by aks ccm and csi driver already set this value as false, this change would make value in open source as identical as managed ccm and csi driver.

@andyzhangx
Copy link
Member

/retest

@andyzhangx andyzhangx changed the title fix: set ForceAttemptHTTP2 as false to respect custom dialer fix: set ForceAttemptHTTP2 to increase ARM throttling limit Jan 3, 2023
@k8s-ci-robot k8s-ci-robot merged commit ea01ac8 into kubernetes-sigs:master Jan 3, 2023
@serbrech
Copy link
Member

serbrech commented Jan 3, 2023

What about the impact on SNAT exhaustion @andyzhangx ? has this been conisdered/tested?

@andyzhangx
Copy link
Member

andyzhangx commented Jan 4, 2023 •
edited

What about the impact on SNAT exhaustion @andyzhangx ? has this been conisdered/tested?

@serbrech there is MaxIdleConnsPerHost: 100 setting which means it would occupy at most 100 ports at most, and in the csi driver load test, it's 10 Azure RoleInstances for true value, and 97 Azure RoleInstances for false value, so it's increasing 97-10=87 ports during load test, that should be acceptable in most cases since it's increasing ~10x ARM throttling limit.

This was referenced Jan 4, 2023
Merged
Merged
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@andyzhangx andyzhangx andyzhangx approved these changes

@MartinForReal MartinForReal Awaiting requested review from MartinForReal

@nilo19 nilo19 Awaiting requested review from nilo19

Assignees

@andyzhangx andyzhangx

Labels
approved Indicates a PR has been approved by an approver from all required OWNERS files. cncf-cla: yes Indicates the PR's author has signed the CNCF CLA. kind/bug Categorizes issue or PR as related to a bug. lgtm "Looks good to me", indicates that a PR is ready to be merged. ok-to-test Indicates a non-member PR verified by an org member that is safe to test. release-note Denotes a PR that will be considered when it comes time to generate release notes. size/XS Denotes a PR that changes 0-9 lines, ignoring generated files.
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

None yet
6 participants
@umagnus @k8s-ci-robot @coveralls @andyzhangx @nilo19 @serbrech