[🐛 Bug]: Autoscaling jobs issue after keda 2.16.1 upgrade

[amardeep2006]

What happened?

I tried upgrading the grid 4.27.0 from helm version 0.38.0 to 0.38.2(trunk branch) and the KEDA does not seems to be picking pending sessions from queue.
I am passing just on capability in my tests browserName: 'chrome'

capabilities: [{
    browserName: 'chrome',
    'se:downloadsEnabled': true
}],

Autoscaling type is Job. I have tried both default and accurate strategy. Is there some breaking change in 0.38.2.

Command used to start Selenium Grid with Docker (or Kubernetes)

Using helm 0.38.2 
Note : I have taken the helm chart from trunk branch today and it has latest commit https://github.com/SeleniumHQ/docker-selenium/commit/d01680cba3feb3d050d9ff667aaa9816fca8e33a


global:
  seleniumGrid:
    # Image registry for all selenium components
    imageRegistry: myrepo/selenium-grid
    # Image tag for all selenium components
    imageTag: "4.27.0-20241225"
    # Image tag for browser's nodes
    nodesImageTag: "4.27.0-20241225"
    # Image tag for browser's video recorder
    imagePullSecret: ""
    # Log level for all components. Possible values describe here: https://www.selenium.dev/documentation/grid/configuration/cli_options/#logging
    logLevel: INFO
    # -- Whether to enable structured logging
    structuredLogs: true    
    # kubectl image is used to execute kubectl commands in utility jobs
    kubectlImage: myrepo/bitnami/kubectl:latest
isolateComponents: false
# Basic auth settings for Selenium Grid
basicAuth:
  # Enable or disable basic auth
  enabled: true
  # -- Username for basic auth
  username: $GRID_USERNAME
  # -- Password for basic auth
  password: $GRID_PASSWORD  
  # -- Embed the basic auth "u:p@" in few URLs e.g. SE_NODE_GRID_URL.
  embeddedUrl: false
autoscaling:
  enabled: true
  scalingType: job
  scaledOptions:
    minReplicaCount: 0
    maxReplicaCount: $MAX_REPLICAS_COUNT
    pollingInterval: 20  
  # terminationGracePeriodSeconds: 5400 #default
  # Options for KEDA ScaledJobs (only used when scalingType is set to "job"). See https://keda.sh/docs/latest/concepts/scaling-jobs/#scaledjob-spec
  scaledJobOptions:
    scalingStrategy:
      # Change this from "default" to "accurate" or "eager" when the calculation problem is fixed
      # -- Scaling strategy for KEDA ScaledJob
      strategy: default  
customLabels: {"app-id": "selgrid", "app-tier": "application", "app-name": "selgrid"}
tls:
  ingress:
    enabled: true
ingress:
  # Name of ingress class to select which controller will implement ingress resource
  # Custom annotations for ingress resource
  annotations:
    nginx.ingress.kubernetes.io/proxy-connect-timeout: "900"
    nginx.ingress.kubernetes.io/proxy-read-timeout: "900"
    nginx.ingress.kubernetes.io/proxy-send-timeout: "900"
    nginx.ingress.kubernetes.io/proxy-body-size: 100m
  # Default host for the ingress resource
  hostname: $INGRESS_DOMAIN
  tls:
    - secretName: sel-grid-tls-secret-dynamic
      hosts:
        - $INGRESS_DOMAIN
router:
  imagePullPolicy: Always
distributor:
  imagePullPolicy: Always  
eventBus:
  imagePullPolicy: Always
sessionMap:
  imagePullPolicy: Always
sessionQueue:
  imagePullPolicy: Always    
hub:
  # Custom sub path for the hub deployment
  # subPath: /selenium  
  imagePullPolicy: Always
  # Resources for container
  resources:
    requests:
      memory: "200Mi"
      cpu: "100m"
    limits:
      memory: "9Gi"
      cpu: "4"
  extraEnvironmentVariables:
    - name: SE_JAVA_OPTS
      value: "-Xmx8192m"  
chromeNode:
# Number of chrome nodes Only used when Autoscaling is false
  replicas: $FIXED_CHROME_REPLICAS
  imagePullPolicy: Always
  # /dev/shm volume
  dshmVolumeSizeLimit: "2Gi"
  # Resources for chrome-node container
  resources:
    requests:
      memory: "100Mi"
      cpu: "100m"
    limits:
      memory: "2Gi"
      cpu: "2"
  extraEnvironmentVariables: 
    - name: "SE_NODE_ENABLE_MANAGED_DOWNLOADS"
      value: "true"
    # - name: "SE_VNC_NO_PASSWORD"
    #   value: "1"
    # - name: "SE_VNC_VIEW_ONLY"
    #   value: "1"
  terminationGracePeriodSeconds: 5400
edgeNode:
  replicas: $FIXED_EDGE_REPLICAS
  imagePullPolicy: Always
  # /dev/shm volume
  dshmVolumeSizeLimit: "2Gi"
  # Resources for edge-node container
  resources:
    requests:
      memory: "100Mi"
      cpu: "100m"
    limits:
      memory: "2Gi"
      cpu: "2"
  extraEnvironmentVariables: 
    - name: "SE_NODE_ENABLE_MANAGED_DOWNLOADS"
      value: "true"
    # - name: "SE_VNC_NO_PASSWORD"
    #   value: "1"
    # - name: "SE_VNC_VIEW_ONLY"
    #   value: "1"
  terminationGracePeriodSeconds: 5400
firefoxNode:
  enabled: false
  imagePullPolicy: Always
  # /dev/shm volume
  dshmVolumeSizeLimit: "2Gi"
  # Resources for firefox-node container
  resources:
    requests:
      memory: "1Gi"
      cpu: "1"
    limits:
      memory: "2Gi"
      cpu: "2"
  extraEnvironmentVariables: 
    - name: "SE_NODE_ENABLE_MANAGED_DOWNLOADS"
      value: "true"
    # - name: "SE_VNC_NO_PASSWORD"
    #   value: "1"
    # - name: "SE_VNC_VIEW_ONLY"
    #   value: "1"
  autoscaling:
    scaledOptions:
      minReplicaCount: 0
      maxReplicaCount: 3
  terminationGracePeriodSeconds: 5400
keda:
  image:  
    keda:
      registry: myrepo
      # -- Image name of KEDA operator
      repository: myrepo/kedacore/keda
      # -- Image tag of KEDA operator. Optional, given app version of Helm chart is used by default
      tag: $KEDA_VERSION
    metricsApiServer:
      registry: myrepo
      # -- Image name of KEDA Metrics API Server
      repository: myrepo/kedacore/keda-metrics-apiserver
      # -- Image tag of KEDA Metrics API Server. Optional, given app version of Helm chart is used by default
      tag: $KEDA_VERSION
    webhooks:
      registry: myrepo
      # -- Image name of KEDA admission-webhooks
      repository: myrepo/kedacore/keda-admission-webhooks
      # -- Image tag of KEDA admission-webhooks . Optional, given app version of Helm chart is used by default
      tag: $KEDA_VERSION
    # -- Image pullPolicy for all KEDA components
  podLabels:
    # -- Pod labels for KEDA operator
    keda: {"app-id": "selgrid", "app-tier": "application", "app-name": "selgrid"}
    # -- Pod labels for KEDA Metrics Adapter
    metricsAdapter: {"app-id": "selgrid", "app-tier": "application", "app-name": "selgrid"}
    # -- Pod labels for KEDA Admission webhooks
    webhooks: {"app-id": "selgrid", "app-tier": "application", "app-name": "selgrid"}

Relevant log output

2024-12-27T10:06:00Z	INFO	setup	maxprocs: Updating GOMAXPROCS=1: determined from CPU quota
2024-12-27T10:06:00Z	INFO	setup	Starting manager
2024-12-27T10:06:00Z	INFO	setup	KEDA Version: 2.16.1
2024-12-27T10:06:00Z	INFO	setup	Git Commit: ce14b239e0300f388b0425aef68154d8070cd66f
2024-12-27T10:06:00Z	INFO	setup	Go Version: go1.23.4
2024-12-27T10:06:00Z	INFO	setup	Go OS/Arch: linux/amd64
2024-12-27T10:06:00Z	INFO	setup	Running on Kubernetes 1.28+	{"version": "v1.28.15-eks-7f9249a"}
2024-12-27T10:06:00Z	INFO	setup	WARNING: KEDA 2.16.1 hasn't been tested on Kubernetes v1.28.15-eks-7f9249a
2024-12-27T10:06:00Z	INFO	setup	You can check recommended versions on https://keda.sh
2024-12-27T10:06:00Z	INFO	starting server	{"name": "health probe", "addr": "[::]:8081"}
I1227 10:06:00.649846       1 leaderelection.go:254] attempting to acquire leader lease mer-merselgrid-dev-mer-sel-grid/operator.keda.sh...
I1227 10:06:17.014943       1 leaderelection.go:268] successfully acquired lease mer-merselgrid-dev-mer-sel-grid/operator.keda.sh
2024-12-27T10:06:17Z	INFO	Starting EventSource	{"controller": "scaledobject", "controllerGroup": "keda.sh", "controllerKind": "ScaledObject", "source": "kind source: *v1alpha1.ScaledObject"}
2024-12-27T10:06:17Z	INFO	Starting EventSource	{"controller": "scaledobject", "controllerGroup": "keda.sh", "controllerKind": "ScaledObject", "source": "kind source: *v2.HorizontalPodAutoscaler"}
2024-12-27T10:06:17Z	INFO	Starting Controller	{"controller": "scaledobject", "controllerGroup": "keda.sh", "controllerKind": "ScaledObject"}
2024-12-27T10:06:17Z	INFO	Starting EventSource	{"controller": "triggerauthentication", "controllerGroup": "keda.sh", "controllerKind": "TriggerAuthentication", "source": "kind source: *v1alpha1.TriggerAuthentication"}
2024-12-27T10:06:17Z	INFO	Starting Controller	{"controller": "triggerauthentication", "controllerGroup": "keda.sh", "controllerKind": "TriggerAuthentication"}
2024-12-27T10:06:17Z	INFO	Starting EventSource	{"controller": "scaledjob", "controllerGroup": "keda.sh", "controllerKind": "ScaledJob", "source": "kind source: *v1alpha1.ScaledJob"}
2024-12-27T10:06:17Z	INFO	Starting Controller	{"controller": "scaledjob", "controllerGroup": "keda.sh", "controllerKind": "ScaledJob"}
2024-12-27T10:06:17Z	INFO	Starting EventSource	{"controller": "cloudeventsource", "controllerGroup": "eventing.keda.sh", "controllerKind": "CloudEventSource", "source": "kind source: *v1alpha1.CloudEventSource"}
2024-12-27T10:06:17Z	INFO	Starting Controller	{"controller": "cloudeventsource", "controllerGroup": "eventing.keda.sh", "controllerKind": "CloudEventSource"}
2024-12-27T10:06:17Z	INFO	Starting EventSource	{"controller": "clustertriggerauthentication", "controllerGroup": "keda.sh", "controllerKind": "ClusterTriggerAuthentication", "source": "kind source: *v1alpha1.ClusterTriggerAuthentication"}
2024-12-27T10:06:17Z	INFO	Starting Controller	{"controller": "clustertriggerauthentication", "controllerGroup": "keda.sh", "controllerKind": "ClusterTriggerAuthentication"}
2024-12-27T10:06:17Z	INFO	Starting EventSource	{"controller": "clustercloudeventsource", "controllerGroup": "eventing.keda.sh", "controllerKind": "ClusterCloudEventSource", "source": "kind source: *v1alpha1.ClusterCloudEventSource"}
2024-12-27T10:06:17Z	INFO	Starting Controller	{"controller": "clustercloudeventsource", "controllerGroup": "eventing.keda.sh", "controllerKind": "ClusterCloudEventSource"}
2024-12-27T10:06:17Z	INFO	Starting EventSource	{"controller": "cert-rotator", "source": "kind source: *v1.Secret"}
2024-12-27T10:06:17Z	INFO	Starting EventSource	{"controller": "cert-rotator", "source": "kind source: *unstructured.Unstructured"}
2024-12-27T10:06:17Z	INFO	Starting EventSource	{"controller": "cert-rotator", "source": "kind source: *unstructured.Unstructured"}
2024-12-27T10:06:17Z	INFO	Starting Controller	{"controller": "cert-rotator"}
2024-12-27T10:06:17Z	INFO	cert-rotation	starting cert rotator controller
2024-12-27T10:06:17Z	INFO	cert-rotation	no cert refresh needed
2024-12-27T10:06:17Z	INFO	cert-rotation	certs are ready in /certs
2024-12-27T10:06:17Z	INFO	Starting workers	{"controller": "cert-rotator", "worker count": 1}
2024-12-27T10:06:17Z	INFO	cert-rotation	no cert refresh needed
2024-12-27T10:06:17Z	ERROR	cert-rotation	Webhook not found. Unable to update certificate.	{"name": "keda-admission", "gvk": "admissionregistration.k8s.io/v1, Kind=ValidatingWebhookConfiguration", "error": "ValidatingWebhookConfiguration.admissionregistration.k8s.io \"keda-admission\" not found"}
github.com/open-policy-agent/cert-controller/pkg/rotator.(*ReconcileWH).ensureCerts
	/workspace/vendor/github.com/open-policy-agent/cert-controller/pkg/rotator/rotator.go:822
github.com/open-policy-agent/cert-controller/pkg/rotator.(*ReconcileWH).Reconcile
	/workspace/vendor/github.com/open-policy-agent/cert-controller/pkg/rotator/rotator.go:791
sigs.k8s.io/controller-runtime/pkg/internal/controller.(*Controller[...]).Reconcile
	/workspace/vendor/sigs.k8s.io/controller-runtime/pkg/internal/controller/controller.go:116
sigs.k8s.io/controller-runtime/pkg/internal/controller.(*Controller[...]).reconcileHandler
	/workspace/vendor/sigs.k8s.io/controller-runtime/pkg/internal/controller/controller.go:303
sigs.k8s.io/controller-runtime/pkg/internal/controller.(*Controller[...]).processNextWorkItem
	/workspace/vendor/sigs.k8s.io/controller-runtime/pkg/internal/controller/controller.go:263
sigs.k8s.io/controller-runtime/pkg/internal/controller.(*Controller[...]).Start.func2.2
	/workspace/vendor/sigs.k8s.io/controller-runtime/pkg/internal/controller/controller.go:224
2024-12-27T10:06:17Z	INFO	cert-rotation	Ensuring CA cert	{"name": "v1beta1.external.metrics.k8s.io", "gvk": "apiregistration.k8s.io/v1, Kind=APIService", "name": "v1beta1.external.metrics.k8s.io", "gvk": "apiregistration.k8s.io/v1, Kind=APIService"}
2024-12-27T10:06:17Z	INFO	Starting workers	{"controller": "triggerauthentication", "controllerGroup": "keda.sh", "controllerKind": "TriggerAuthentication", "worker count": 1}
2024-12-27T10:06:17Z	INFO	Starting workers	{"controller": "scaledjob", "controllerGroup": "keda.sh", "controllerKind": "ScaledJob", "worker count": 1}
2024-12-27T10:06:17Z	INFO	Starting workers	{"controller": "scaledobject", "controllerGroup": "keda.sh", "controllerKind": "ScaledObject", "worker count": 5}
2024-12-27T10:06:17Z	INFO	Reconciling ScaledJob	{"controller": "scaledjob", "controllerGroup": "keda.sh", "controllerKind": "ScaledJob", "ScaledJob": {"name":"org-selgrid-selenium-node-chrome","namespace":"mer-merselgrid-dev-mer-sel-grid"}, "namespace": "mer-merselgrid-dev-mer-sel-grid", "name": "org-selgrid-selenium-node-chrome", "reconcileID": "c7dd5fba-37ed-495f-8796-8286dad16274"}
2024-12-27T10:06:17Z	INFO	Starting workers	{"controller": "clustercloudeventsource", "controllerGroup": "eventing.keda.sh", "controllerKind": "ClusterCloudEventSource", "worker count": 1}
2024-12-27T10:06:17Z	INFO	Starting workers	{"controller": "cloudeventsource", "controllerGroup": "eventing.keda.sh", "controllerKind": "CloudEventSource", "worker count": 1}
2024-12-27T10:06:17Z	INFO	Starting workers	{"controller": "clustertriggerauthentication", "controllerGroup": "keda.sh", "controllerKind": "ClusterTriggerAuthentication", "worker count": 1}
2024-12-27T10:06:17Z	INFO	KubeAPIWarningLogger	unknown field "status.authenticationsTypes"
2024-12-27T10:06:17Z	INFO	KubeAPIWarningLogger	unknown field "status.triggersTypes"
2024-12-27T10:06:17Z	INFO	RolloutStrategy: immediate, No jobs owned by the previous version of the scaledJob	{"controller": "scaledjob", "controllerGroup": "keda.sh", "controllerKind": "ScaledJob", "ScaledJob": {"name":"org-selgrid-selenium-node-chrome","namespace":"mer-merselgrid-dev-mer-sel-grid"}, "namespace": "mer-merselgrid-dev-mer-sel-grid", "name": "org-selgrid-selenium-node-chrome", "reconcileID": "c7dd5fba-37ed-495f-8796-8286dad16274"}
2024-12-27T10:06:17Z	INFO	Initializing Scaling logic according to ScaledJob Specification	{"controller": "scaledjob", "controllerGroup": "keda.sh", "controllerKind": "ScaledJob", "ScaledJob": {"name":"org-selgrid-selenium-node-chrome","namespace":"mer-merselgrid-dev-mer-sel-grid"}, "namespace": "mer-merselgrid-dev-mer-sel-grid", "name": "org-selgrid-selenium-node-chrome", "reconcileID": "c7dd5fba-37ed-495f-8796-8286dad16274"}
2024-12-27T10:06:17Z	INFO	scaleexecutor	Scaling Jobs	{"scaledJob.Name": "org-selgrid-selenium-node-chrome", "scaledJob.Namespace": "mer-merselgrid-dev-mer-sel-grid", "Number of running Jobs": 0}
2024-12-27T10:06:17Z	INFO	scaleexecutor	Scaling Jobs	{"scaledJob.Name": "org-selgrid-selenium-node-chrome", "scaledJob.Namespace": "mer-merselgrid-dev-mer-sel-grid", "Number of pending Jobs": 0}
2024-12-27T10:06:17Z	INFO	Reconciling ScaledJob	{"controller": "scaledjob", "controllerGroup": "keda.sh", "controllerKind": "ScaledJob", "ScaledJob": {"name":"org-selgrid-selenium-node-edge","namespace":"mer-merselgrid-dev-mer-sel-grid"}, "namespace": "mer-merselgrid-dev-mer-sel-grid", "name": "org-selgrid-selenium-node-edge", "reconcileID": "0c1ff265-5fa8-41f4-a218-776a7bab2dd8"}
2024-12-27T10:06:17Z	INFO	RolloutStrategy: immediate, No jobs owned by the previous version of the scaledJob	{"controller": "scaledjob", "controllerGroup": "keda.sh", "controllerKind": "ScaledJob", "ScaledJob": {"name":"org-selgrid-selenium-node-edge","namespace":"mer-merselgrid-dev-mer-sel-grid"}, "namespace": "mer-merselgrid-dev-mer-sel-grid", "name": "org-selgrid-selenium-node-edge", "reconcileID": "0c1ff265-5fa8-41f4-a218-776a7bab2dd8"}
2024-12-27T10:06:17Z	INFO	Initializing Scaling logic according to ScaledJob Specification	{"controller": "scaledjob", "controllerGroup": "keda.sh", "controllerKind": "ScaledJob", "ScaledJob": {"name":"org-selgrid-selenium-node-edge","namespace":"mer-merselgrid-dev-mer-sel-grid"}, "namespace": "mer-merselgrid-dev-mer-sel-grid", "name": "org-selgrid-selenium-node-edge", "reconcileID": "0c1ff265-5fa8-41f4-a218-776a7bab2dd8"}
2024-12-27T10:06:17Z	INFO	scaleexecutor	Scaling Jobs	{"scaledJob.Name": "org-selgrid-selenium-node-edge", "scaledJob.Namespace": "mer-merselgrid-dev-mer-sel-grid", "Number of running Jobs": 0}
2024-12-27T10:06:17Z	INFO	scaleexecutor	Scaling Jobs	{"scaledJob.Name": "org-selgrid-selenium-node-edge", "scaledJob.Namespace": "mer-merselgrid-dev-mer-sel-grid", "Number of pending Jobs": 0}

Operating System

v1.28.15-eks-7f9249a

Docker Selenium version (image tag)

4.27.0-20241225

Selenium Grid chart version (chart version)

0.38.2

[VietND96]

Lets read my PR kedacore/keda#6437
Now the client should send the request with set capability platformName matches with KEDA scaler metadata

[github-actions]

@amardeep2006, thank you for creating this issue. We will troubleshoot it as soon as we can.

---

Info for maintainers

Triage this issue by using labels.

If information is missing, add a helpful comment and then I-issue-template label.

If the issue is a question, add the I-question label.

If the issue is valid but there is no time to troubleshoot it, consider adding the help wanted label.

If the issue requires changes or fixes from an external project (e.g., ChromeDriver, GeckoDriver, MSEdgeDriver, W3C), add the applicable G-* label, and it will provide the correct link and auto-close the issue.

After troubleshooting the issue, please add the R-awaiting answer label.

Thank you!

[VietND96]

If you want scaler trigger against request without cap platformName. In chart, under config hpa of each Node, set metadata platformName to empty (by default it is Linux in chart).

[amardeep2006]

This has left me bit confused. I thought hpa was applicable for only Deployments. Will it work for the Jobs as well?

[VietND96]

You also can create multiple scalers with different metadata easily under config crossBrowsers, it is an array map with each item scheme is same as each node structure. Checkout values file cross-browsers-values.yaml for sample

[VietND96]

Actually hpa is the config key name, it doesn't form the scaling type Job or Deployment, it keeps the scaler trigger params to pass to ScaledObject or ScaledJob. Only autoscaling.scalingType to decide that.

[amardeep2006]

Thanks for your blazing fast response @VietND96 . I am testing both suggestions : passing platformName in capabilities and helm chart config one by one.
I will reply back on this issue.

[VietND96]

As the target I mentioned in the PR is the Grid with autoscaling Nodes, non-autoscaling Nodes, relay Nodes, etc. The scaler needs to isolate and count the exact number of ongoing sessions + pending sessions (without overlapping if multiple ScaledJobs exist at a time) and then send it to KEDA. The rest of the work for metrics to K8s HPA that KEDA will take care.
The scaling behavior is only correct when the scaler implementation can count and expose the correct number to KEDA.
To ensure that counting is correct, we need to implement the condition to trigger the scaler strictly.
Config in 3 areas we need to match (avoid other existing Nodes impact) request capabilities, node stereotypes, and scaler trigger params.

[amardeep2006]

I confirm that in my initial set of sanity testing both suggestions worked. Appreciate your help. For now I will stick with keeping platformName as empty in value file as there are many teams using the solutions without platformName in capabilities and we donot have relay/windows node use case for now. I will share my long term observations over a week.

[VietND96]

Yes, I also will try to get time to write down all the details that users need to know to scale the Grid with KEDA 2.16.1+