Autoscaling selenium grid on kubernetes with scaledjobs #1854

msvticket · 2023-05-23T11:44:11Z

Description

This PR builds on #1714 but adds a few features:

KEDA can be installed automatically
Support for scaling using ScaledJobs (which is the default when autoscaling is enabled)
Set environment variables SE_NODE_GRID_URL and DRAIN_AFTER_SESSION_COUNT automatically
Set graphqlurl automatically based on configuration
Refactor out pod templates to named template
Conditionally adding preStop hook for draining node on termination

Motivation and Context

Compared to #1714 I want to make it easier to get started with autoscaling. The basic setup should be achieved with just setting autoscaling.enabled to true.

Types of changes

Bug fix (non-breaking change which fixes an issue)
New feature (non-breaking change which adds functionality)
Breaking change (fix or feature that would cause existing functionality to change)

Checklist

I have read the contributing document.
My change requires a change to the documentation.
I have updated the documentation accordingly.
I have added tests to cover my changes.
All new and existing tests passed.

msvticket · 2023-05-24T09:09:44Z

To test the selenium-grid helm chart with the contribution of this PR you can get it from the repo https://jenkins-x-charts.github.io/repo. The latest version as of now is 4.12.0.

Autoscale selenium browser nodes running in kubernetes based on the request pending in session queue using KEDA. Toggle autoscaling on/off using 'autoscalingEnabled' option in helm charts.

and make them the default KEDA scaling type install keda automatically set SE_NODE_GRID_URL and DRAIN_AFTER_SESSION_COUNT automatically Set graphqlurl automatically refactor out pod templates to named template conditionally adding preStop hook for deregistering node Signed-off-by: Mårten Svantesson <[email protected]>

win5923 · 2023-05-26T07:06:05Z

When setting 'autoscaling.enabled' to true, KEDA is not automatically installed.

Error: UPGRADE FAILED: [resource mapping not found for name: "selenium-chrome-node" namespace: "devops-tools" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first, resource mapping not found for name: "selenium-edge-node" namespace: "devops-tools" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first, resource mapping not found for name: "selenium-firefox-node" namespace: "devops-tools" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first]

msvticket · 2023-05-26T07:11:10Z

When setting 'autoscaling.enabled' to true, KEDA is not automatically installed.

Error: UPGRADE FAILED: [resource mapping not found for name: "selenium-chrome-node" namespace: "devops-tools" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first, resource mapping not found for name: "selenium-edge-node" namespace: "devops-tools" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first, resource mapping not found for name: "selenium-firefox-node" namespace: "devops-tools" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first]

Well, it works for me. What are your steps for installing selenium-grid?

win5923 · 2023-05-26T07:14:03Z

install helm chart and get value.

helm repo add myrepo https://jenkins-x-charts.github.io/repo/
helm install selenium-grid myrepo/selenium-grid
helm get values selenium-grid -a > value.yaml

change "autoscaling.enabled" to true and upgrade.

helm upgrade selenium-grid myrepo/selenium-grid -f value.yaml

win5923 · 2023-05-29T08:24:57Z

install helm chart and get value.

helm repo add myrepo https://jenkins-x-charts.github.io/repo/
helm install selenium-grid myrepo/selenium-grid
helm get values selenium-grid -a > value.yaml

change "autoscaling.enabled" to true and upgrade.

helm upgrade selenium-grid myrepo/selenium-grid -f value.yaml

sorry this is my new error when autoscaling is enabled, Is something wrong with my steps?

Error: UPGRADE FAILED: template: selenium-grid/templates/firefox-node-deployment.yaml:41:19: executing "selenium-grid/templates/firefox-node-deployment.yaml" at <tpl (toYaml .) $>: error calling tpl: error during tpl function execution for "- name: DRAIN_AFTER_SESSION_COUNT\n  value: '{{- and (eq (include \"seleniumGrid.useKEDA\" .) \"true\") (eq .Values.autoscaling.scalingType\n    \"job\") | ternary \"1\" \"0\" -}}'\n- name: SE_NODE_GRID_URL\n  value: '{{ include \"seleniumGrid.url\" .}}'": template: selenium-grid/templates/firefox-node-deployment.yaml:2:23: executing "selenium-grid/templates/firefox-node-deployment.yaml" at <include "seleniumGrid.useKEDA" .>: error calling include: template: no template "seleniumGrid.useKEDA" associated with template "gotpl"

msvticket · 2023-05-30T11:13:08Z

Well, I'm not really familiar with using the helm command directly to install stuff. I have always used a declarative approach through Jenkins X. Others use similar tools like ArgoCD. These tools work around the problems with handling CRDs that the helm commands have.

That said, it's not that strange that it doesn't work since you use the upgrade command when you want a chart to be installed. Had you added --set autoscaling.enabled=true to the install command it would have worked.
For some more details see:

https://helm.sh/docs/chart_best_practices/custom_resource_definitions/#method-1-let-helm-do-it-for-you

As for the error no template "seleniumGrid.useKEDA" associated with template "gotpl" I don't get that at all. Could have to do with the version of helm? What are you using?

win5923 · 2023-05-31T01:19:13Z

I understand what you said, and I used 'helm install' with --set autoscaling.enabled=true, but the issue still remains about CRDs.

helm version: v3.11.3

helm install selenium-grid myrepo/selenium-grid -n devops-tools --set autoscaling.enabled=true

Error: INSTALLATION FAILED: unable to build kubernetes objects from release manifest: [resource mapping not found for name: "selenium-chrome-node" namespace: "devops-tools" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first, resource mapping not found for name: "selenium-edge-node" namespace: "devops-tools" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first, resource mapping not found for name: "selenium-firefox-node" namespace: "devops-tools" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first]

helm seem to normally install subchart after current chart so trying this work around

msvticket · 2023-05-31T09:36:07Z

I understand what you said, and I used 'helm install' with --set autoscaling.enabled=true, but the issue still remains about CRDs.

helm version: v3.11.3

helm install selenium-grid myrepo/selenium-grid -n devops-tools --set autoscaling.enabled=true

Error: INSTALLATION FAILED: unable to build kubernetes objects from release manifest: [resource mapping not found for name: "selenium-chrome-node" namespace: "devops-tools" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first, resource mapping not found for name: "selenium-edge-node" namespace: "devops-tools" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first, resource mapping not found for name: "selenium-firefox-node" namespace: "devops-tools" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first]

Ah, there seem to be another limitation in helm: Sub charts are installed after the main chart. But there is a way to delay the installation of resources. I have added this now, so if you do helm repo update and then try again it should work.

msvticket · 2023-06-28T14:14:47Z

While working on support for video recording I noticed how impractical it was that I had put default values in extraEnvironmentVariables. That means that things will break when overriding extraEnvironmentVariables without setting all values. I'll change that.

diemol

Thank you for this, @msvticket!

diemol · 2023-06-29T11:59:28Z

Due to the previous conflicts between this PR and trunk and how the updates have been propagated, seems quite complicated to merge the commit history as it is.

I will squash and merge, and this will keep the authors there.

Also, thank you, @prashanth-volvocars!

DeepKandey · 2023-07-18T04:26:26Z

@msvticket Tried to install Helm with autoscaling enabled. Keda is not getting installed.

E0718 09:52:28.784838 25660 memcache.go:287] couldn't get resource list for external.metrics.k8s.io/v1beta1: the server is currently unable to handle the request
E0718 09:52:28.902582 25660 memcache.go:121] couldn't get resource list for external.metrics.k8s.io/v1beta1: the server is currently unable to handle the request
Error: INSTALLATION FAILED: failed post-install: unable to build kubernetes object for deleting hook selenium-grid/templates/chrome-node-scaledjobs.yaml: resource mapping not found for name: "selenium-chrome-node" namespace: "default" from "": no matches for kind "ScaledJob" in version "keda.sh/v1alpha1"
ensure CRDs are installed first

I think the workaround is still not working

DeepKandey · 2023-08-01T04:05:15Z

Hey, can someone please help me with this issue?

win5923 · 2023-08-01T07:27:45Z

run kubectl get apiservice v1beta1.external.metrics.k8s.io -o yaml to see what's wrong with the service.

DeepKandey · 2023-08-01T16:59:26Z

Got below response @win5923

apiVersion: apiregistration.k8s.io/v1
kind: APIService
metadata:
annotations:
meta.helm.sh/release-name: selenium-grid
meta.helm.sh/release-namespace: default
creationTimestamp: "2023-07-18T04:22:28Z"
labels:
app.kubernetes.io/component: operator
app.kubernetes.io/instance: selenium-grid
app.kubernetes.io/managed-by: Helm
app.kubernetes.io/name: v1beta1.external.metrics.k8s.io
app.kubernetes.io/part-of: keda-operator
app.kubernetes.io/version: 2.10.1
helm.sh/chart: keda-2.10.2
name: v1beta1.external.metrics.k8s.io
resourceVersion: "217776"
uid: b49b2159-07fd-49b1-8f41-992d827bb6ab
spec:
caBundle: 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

group: external.metrics.k8s.io
groupPriorityMinimum: 100
service:
name: keda-operator-metrics-apiserver
namespace: default
port: 443
version: v1beta1
versionPriority: 100
status:
conditions:

lastTransitionTime: "2023-08-01T16:07:02Z"
message: all checks passed
reason: Passed
status: "True"
type: Available

win5923 · 2023-08-01T17:38:49Z

The issue is not with the APIService. it might be a problem with your cluster network.

msvticket force-pushed the k8s-dynamic-grid branch from 606b961 to 54d86da Compare May 23, 2023 15:44

msvticket marked this pull request as ready for review May 23, 2023 15:47

msvticket changed the title ~~Kubernetes dynamic grid with scaledjobs~~ Autoscaling selenium grid on kubernetes with scaledjobs May 24, 2023

msvticket mentioned this pull request May 24, 2023

Dynamic Selenium 4 grid on kubernetes SeleniumHQ/selenium#9845

Closed

prashanth-volvocars and others added 3 commits May 25, 2023 18:45

Autoscaling selenium grid on kubernetes

12484e3

Autoscale selenium browser nodes running in kubernetes based on the request pending in session queue using KEDA. Toggle autoscaling on/off using 'autoscalingEnabled' option in helm charts.

Added all suggested changes

4f73df9

msvticket force-pushed the k8s-dynamic-grid branch from 54d86da to 9d4c6ef Compare May 25, 2023 16:45

fix: work around for limitations in helm

49d7107

helm seem to normally install subchart after current chart so trying this work around

jcputney mentioned this pull request Jun 23, 2023

Autoscaling selenium grid on kubernetes #1714

Closed

8 tasks

msvticket and others added 3 commits June 28, 2023 16:58

Merge branch 'trunk' into k8s-dynamic-grid

10dfc5b

fix: format

7d17924

fix: extracting default node env variables to configmap

04ff13d

msvticket force-pushed the k8s-dynamic-grid branch from dea3c07 to 04ff13d Compare June 28, 2023 15:31

diemol approved these changes Jun 29, 2023

View reviewed changes

diemol merged commit f0bbfe0 into SeleniumHQ:trunk Jun 29, 2023

VietND96 mentioned this pull request Dec 5, 2023

[🐛 Bug]: invalid session id after upgrading to 4.15 (helm-charts on k8s) #2039

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Autoscaling selenium grid on kubernetes with scaledjobs #1854

Autoscaling selenium grid on kubernetes with scaledjobs #1854

msvticket commented May 23, 2023 •

edited

Loading

msvticket commented May 24, 2023

win5923 commented May 26, 2023

msvticket commented May 26, 2023

win5923 commented May 26, 2023

win5923 commented May 29, 2023 •

edited

Loading

msvticket commented May 30, 2023

win5923 commented May 31, 2023 •

edited

Loading

msvticket commented May 31, 2023

msvticket commented Jun 28, 2023

diemol left a comment

diemol commented Jun 29, 2023

DeepKandey commented Jul 18, 2023

DeepKandey commented Aug 1, 2023

win5923 commented Aug 1, 2023

DeepKandey commented Aug 1, 2023

win5923 commented Aug 1, 2023 •

edited

Loading

Autoscaling selenium grid on kubernetes with scaledjobs #1854

Autoscaling selenium grid on kubernetes with scaledjobs #1854

Conversation

msvticket commented May 23, 2023 • edited Loading

Description

Motivation and Context

Types of changes

Checklist

msvticket commented May 24, 2023

win5923 commented May 26, 2023

msvticket commented May 26, 2023

win5923 commented May 26, 2023

win5923 commented May 29, 2023 • edited Loading

msvticket commented May 30, 2023

win5923 commented May 31, 2023 • edited Loading

msvticket commented May 31, 2023

msvticket commented Jun 28, 2023

diemol left a comment

Choose a reason for hiding this comment

diemol commented Jun 29, 2023

DeepKandey commented Jul 18, 2023

DeepKandey commented Aug 1, 2023

win5923 commented Aug 1, 2023

DeepKandey commented Aug 1, 2023

win5923 commented Aug 1, 2023 • edited Loading

msvticket commented May 23, 2023 •

edited

Loading

win5923 commented May 29, 2023 •

edited

Loading

win5923 commented May 31, 2023 •

edited

Loading

win5923 commented Aug 1, 2023 •

edited

Loading