Keda

Описание

KEDA (Kubernetes Event-driven Autoscaling) позволяет автоматически масштабировать Kubernetes-объекты на основе событий, таких как нагрузка на процессор, память или другие метрики.

Основные возможности:

• Позволяет масштабировать приложения в реальном времени в зависимости от количества событий, поступающих в систему. Это позволяет оптимизировать использование ресурсов и обеспечивать отзывчивость системы.
• Интегрируется с различными поставщиками событий, такими как Apache Kafka, Azure Queue Storage, RabbitMQ и другими. Это обеспечивает гибкость в выборе источника данных для вашего приложения.
• Интегрируется с механизмом масштабирования Horizontal Pod Autoscaler в Kubernetes, что позволяет использовать их вместе для оптимального управления нагрузкой.
• Предоставляет различные параметры конфигурации, такие как минимальное и максимальное количество реплик, метрики для масштабирования и т. д., что позволяет настраивать его поведение под конкретные потребности приложения.*

Подключение модуля Keda

Описание Yaml

Внимание!

Values без описания (1) являются продвинутыми настройками и редактировать их не рекомендовано

1. Описание

apiVersion: addon.bootsman.tech/v1alpha1
kind: Config
metadata:
  name: CLUSTER_NAME-keda
  namespace: CLUSTER_NAMESPACE
spec:
  enabled: true (1)
  values:
    image:
      keda:
        registry: harbor.bootsman.host/bootsman-nimbus/common-artifacts
        repository: keda
        tag: 2.15.1
      metricsApiServer:
        registry: harbor.bootsman.host/bootsman-nimbus/common-artifacts
        repository: keda-metrics-apiserver
        tag: 2.15.1
      webhooks:
        registry: harbor.bootsman.host/bootsman-nimbus/common-artifacts
        repository: keda-admission-webhooks
        tag: 2.15.1

1. True - включено.

   False - выключено

Настройка в UI

Все Values

Продвинутые настройки

Ниже представлены тонкие настройки модуля.

Используйте их для расширения конфигурации модуля, если потребуется.

Документация

Более полная документация по модулю:
Keda Docs
Keda Chart

keda Values

  values:
    global:
      image:
        # -- Global image registry of KEDA components
        registry: null

    image:
      keda:
        # -- Image registry of KEDA operator
        registry: harbor.bootsman.host/bootsman-nimbus/common-artifacts
        # -- Image name of KEDA operator
        repository: keda
        # -- Image tag of KEDA operator. Optional, given app version of Helm chart is used by default
        tag: 2.15.1
      metricsApiServer:
        # -- Image registry of KEDA Metrics API Server
        registry: harbor.bootsman.host/bootsman-nimbus/common-artifacts
        # -- Image name of KEDA Metrics API Server
        repository: keda-metrics-apiserver
        # -- Image tag of KEDA Metrics API Server. Optional, given app version of Helm chart is used by default
        tag: 2.15.1
      webhooks:
        # -- Image registry of KEDA admission-webhooks
        registry: harbor.bootsman.host/bootsman-nimbus/common-artifacts
        # -- Image name of KEDA admission-webhooks
        repository: keda-admission-webhooks
        # -- Image tag of KEDA admission-webhooks . Optional, given app version of Helm chart is used by default
        tag: 2.15.1
      # -- Image pullPolicy for all KEDA components
      pullPolicy: Always

    # -- Kubernetes cluster name. Used in features such as emitting CloudEvents
    clusterName: kubernetes-default

    # -- Kubernetes cluster domain
    clusterDomain: cluster.local

    crds:
      # -- Defines whether the KEDA CRDs have to be installed or not.
      install: true

      # -- Custom annotations specifically for CRDs
      additionalAnnotations:
        {}
        # foo: bar

    # -- Defines Kubernetes namespaces to watch to scale their workloads. Default watches all namespaces
    watchNamespace: ""

    # -- Name of secret to use to pull images to use to pull Docker images
    imagePullSecrets: []

    networkPolicy:
      # -- Enable network policies
      enabled: false
      # -- Flavor of the network policies (cilium)
      flavor: "cilium"
      # -- Allow use of extra egress rules for cilium network policies
      cilium:
        operator:
          extraEgressRules: []

    operator:
      # -- Name of the KEDA operator
      name: keda-operator
      # -- ReplicaSets for this Deployment you want to retain (Default: 10)
      revisionHistoryLimit: 10
      # -- Capability to configure the number of replicas for KEDA operator.
      # While you can run more replicas of our operator, only one operator instance will be the leader and serving traffic.
      # You can run multiple replicas, but they will not improve the performance of KEDA, it could only reduce downtime during a failover.
      # Learn more in [our documentation](https://keda.sh/docs/latest/operate/cluster/#high-availability).
      replicaCount: 1
      # --Disable response compression for k8s restAPI in client-go.
      # Disabling compression simply means that turns off the process of making data smaller for K8s restAPI in client-go for faster transmission.
      disableCompression: true
      # -- [Affinity] for pod scheduling for KEDA operator. Takes precedence over the `affinity` field
      affinity: {}
        # podAntiAffinity:
        #   requiredDuringSchedulingIgnoredDuringExecution:
        #   - labelSelector:
        #       matchExpressions:
        #       - key: app
        #         operator: In
        #         values:
        #         - keda-operator
        #     topologyKey: "kubernetes.io/hostname"
      # -- Additional containers to run as part of the operator deployment
      extraContainers: []
        # - name: hello-many
        # args:
        #   - -c
        #   - "while true; do echo hi; sleep 300; done"
        # command:
        #   - /bin/sh
        # image: 'busybox:glibc'
      # -- Additional init containers to run as part of the operator deployment
      extraInitContainers: []
        # - name: hello-once
        # args:
        #   - -c
        #   - "echo 'Hello World!'"
        # command:
        #   - /bin/sh
        # image: 'busybox:glibc'
      # -- Liveness probes for operator ([docs](https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/))
      livenessProbe:
        initialDelaySeconds: 25
        periodSeconds: 10
        timeoutSeconds: 1
        failureThreshold: 3
        successThreshold: 1
      # -- Readiness probes for operator ([docs](https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/#define-readiness-probes))
      readinessProbe:
        initialDelaySeconds: 20
        periodSeconds: 3
        timeoutSeconds: 1
        failureThreshold: 3
        successThreshold: 1

    metricsServer:
      # -- ReplicaSets for this Deployment you want to retain (Default: 10)
      revisionHistoryLimit: 10
      # -- Capability to configure the number of replicas for KEDA metric server.
      # While you can run more replicas of our metric server, only one instance will used and serve traffic.
      # You can run multiple replicas, but they will not improve the performance of KEDA, it could only reduce downtime during a failover.
      # Learn more in [our documentation](https://keda.sh/docs/latest/operate/cluster/#high-availability).
      replicaCount: 1
      # --Disable response compression for k8s restAPI in client-go.
      # Disabling compression simply means that turns off the process of making data smaller for K8s restAPI in client-go for faster transmission.
      disableCompression: true
      # use ClusterFirstWithHostNet if `useHostNetwork: true` https://kubernetes.io/docs/concepts/services-networking/dns-pod-service/#pod-s-dns-policy
      # -- Defined the DNS policy for the metric server
      dnsPolicy: ClusterFirst
      # -- Enable metric server to use host network
      useHostNetwork: false
      # -- [Affinity] for pod scheduling for Metrics API Server. Takes precedence over the `affinity` field
      affinity: {}
        # podAntiAffinity:
        #   requiredDuringSchedulingIgnoredDuringExecution:
        #   - labelSelector:
        #       matchExpressions:
        #       - key: app
        #         operator: In
        #         values:
        #         - keda-operator-metrics-apiserver
        #     topologyKey: "kubernetes.io/hostname"
      # -- Liveness probes for Metrics API Server ([docs](https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/))
      livenessProbe:
        initialDelaySeconds: 5
        periodSeconds: 10
        timeoutSeconds: 1
        failureThreshold: 3
        successThreshold: 1
      # -- Readiness probes for Metrics API Server ([docs](https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/#define-readiness-probes))
      readinessProbe:
        initialDelaySeconds: 5
        periodSeconds: 3
        timeoutSeconds: 1
        failureThreshold: 3
        successThreshold: 1

    webhooks:
      # -- Enable admission webhooks (this feature option will be removed in v2.12)
      enabled: true
      # -- Port number to use for KEDA admission webhooks. Default is 9443.
      port: ""
      # -- Port number to use for KEDA admission webhooks health probe
      healthProbePort: 8081
      # -- Liveness probes for admission webhooks ([docs](https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/))
      livenessProbe:
        initialDelaySeconds: 25
        periodSeconds: 10
        timeoutSeconds: 1
        failureThreshold: 3
        successThreshold: 1
      # -- Readiness probes for admission webhooks ([docs](https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/#define-readiness-probes))
      readinessProbe:
        initialDelaySeconds: 20
        periodSeconds: 3
        timeoutSeconds: 1
        failureThreshold: 3
        successThreshold: 1
      # -- Enable webhook to use host network, this is required on EKS with custom CNI
      useHostNetwork: false
      # -- Name of the KEDA admission webhooks
      name: keda-admission-webhooks
      # -- ReplicaSets for this Deployment you want to retain (Default: 10)
      revisionHistoryLimit: 10
      # -- Capability to configure the number of replicas for KEDA admission webhooks
      replicaCount: 1
      # -- [Affinity] for pod scheduling for KEDA admission webhooks. Takes precedence over the `affinity` field
      affinity: {}
        # podAntiAffinity:
        #   requiredDuringSchedulingIgnoredDuringExecution:
        #   - labelSelector:
        #       matchExpressions:
        #       - key: app
        #         operator: In
        #         values:
        #         - keda-admission-webhooks
        #     topologyKey: "kubernetes.io/hostname"

      # -- [Failure policy](https://kubernetes.io/docs/reference/access-authn-authz/extensible-admission-controllers/#failure-policy) to use with KEDA admission webhooks
      failurePolicy: Ignore

    upgradeStrategy:
      # -- Capability to configure [Deployment upgrade strategy] for operator
      operator: {}
        # type: RollingUpdate
        # rollingUpdate:
        #   maxUnavailable: 1
        #   maxSurge: 1

      # -- Capability to configure [Deployment upgrade strategy] for Metrics Api Server
      metricsApiServer: {}
        # type: RollingUpdate
        # rollingUpdate:
        #   maxUnavailable: 1
        #   maxSurge: 1

      # -- Capability to configure [Deployment upgrade strategy] for Admission webhooks
      webhooks: {}
        # type: RollingUpdate
        # rollingUpdate:
        #   maxUnavailable: 1
        #   maxSurge: 1

    podDisruptionBudget:
      # -- Capability to configure [Pod Disruption Budget]
      operator: {}
        #  minAvailable: 1
        #  maxUnavailable: 1

      # -- Capability to configure [Pod Disruption Budget]
      metricServer: {}
        #  minAvailable: 1
        #  maxUnavailable: 1

      # -- Capability to configure [Pod Disruption Budget]
      webhooks: {}
        #  minAvailable: 1
        #  maxUnavailable: 1

    # -- Custom labels to add into metadata
    additionalLabels:
      {}
      # foo: bar

    # -- Custom annotations to add into metadata
    additionalAnnotations:
      {}
      # foo: bar

    podAnnotations:
      # -- Pod annotations for KEDA operator
      keda: {}
      # -- Pod annotations for KEDA Metrics Adapter
      metricsAdapter: {}
      # -- Pod annotations for KEDA Admission webhooks
      webhooks: {}
    podLabels:
      # -- Pod labels for KEDA operator
      keda: {}
      # -- Pod labels for KEDA Metrics Adapter
      metricsAdapter: {}
      # -- Pod labels for KEDA Admission webhooks
      webhooks: {}

    rbac:
      # -- Specifies whether RBAC should be used
      create: true
      # -- Specifies whether RBAC for CRDs should be [aggregated](https://kubernetes.io/docs/reference/access-authn-authz/rbac/#aggregated-clusterroles) to default roles (view, edit, admin)
      aggregateToDefaultRoles: false

      # -- Whether RBAC for configured CRDs that can have a `scale` subresource should be created
      enabledCustomScaledRefKinds: true
      # -- List of custom resources that support the `scale` subresource and can be referenced by `scaledobject.spec.scaleTargetRef`.
      # The feature needs to be also enabled by `enabledCustomScaledRefKinds`.
      # If left empty, RBAC for `apiGroups: *` and `resources: *, */scale` will be created
      # note: Deployments and StatefulSets are supported out of the box
      scaledRefKinds:
      - apiGroup: "*"
        kind: "*"
      # - apiGroup: argoproj.io
      #   kind: Rollout

    serviceAccount:
      operator:
        # -- Specifies whether a service account should be created
        create: true
        # -- The name of the service account to use.
        name: keda-operator
        # -- Specifies whether a service account should automount API-Credentials
        automountServiceAccountToken: true
        # -- Annotations to add to the service account
        annotations: {}
      metricServer:
        # -- Specifies whether a service account should be created
        create: true
        # -- The name of the service account to use.
        name: keda-metrics-server
        # -- Specifies whether a service account should automount API-Credentials
        automountServiceAccountToken: true
        # -- Annotations to add to the service account
        annotations: {}
      webhooks:
        # -- Specifies whether a service account should be created
        create: true
        # -- The name of the service account to use.
        name: keda-webhook
        # -- Specifies whether a service account should automount API-Credentials
        automountServiceAccountToken: true
        # -- Annotations to add to the service account
        annotations: {}

    podIdentity:
      azureWorkload:
        # -- Set to true to enable Azure Workload Identity usage.
        # See https://keda.sh/docs/concepts/authentication/#azure-workload-identity
        # This will be set as a label on the KEDA service account.
        enabled: false
        # Set to the value of the Azure Active Directory Client and Tenant Ids
        # respectively. These will be set as annotations on the KEDA service account.
        # -- Id of Azure Active Directory Client to use for authentication with Azure Workload Identity. ([docs](https://keda.sh/docs/concepts/authentication/#azure-workload-identity))
        clientId: ""
        # -- Id Azure Active Directory Tenant to use for authentication with for Azure Workload Identity. ([docs](https://keda.sh/docs/concepts/authentication/#azure-workload-identity))
        tenantId: ""
        # Set to the value of the service account token expiration duration.
        # This will be set as an annotation on the KEDA service account.
        # -- Duration in seconds to automatically expire tokens for the service account. ([docs](https://keda.sh/docs/concepts/authentication/#azure-workload-identity))
        tokenExpiration: 3600
      aws:
        irsa:
          # -- Specifies whether [AWS IAM Roles for Service Accounts (IRSA)](https://docs.aws.amazon.com/eks/latest/userguide/iam-roles-for-service-accounts.html) is to be enabled or not.
          enabled: false
          # -- Sets the token audience for IRSA.
          # This will be set as an annotation on the KEDA service account.
          audience: "sts.amazonaws.com"
          # -- Set to the value of the ARN of an IAM role with a web identity provider.
          # This will be set as an annotation on the KEDA service account.
          roleArn: ""
          # -- Sets the use of an STS regional endpoint instead of global.
          # Recommended to use regional endpoint in almost all cases.
          # This will be set as an annotation on the KEDA service account.
          stsRegionalEndpoints: "true"
          # -- Set to the value of the service account token expiration duration.
          # This will be set as an annotation on the KEDA service account.
          tokenExpiration: 86400
      gcp:
        # -- Set to true to enable GCP Workload Identity.
        # See https://keda.sh/docs/2.10/authentication-providers/gcp-workload-identity/
        # This will be set as a annotation on the KEDA service account.
        enabled: false
        # -- GCP IAM Service Account Email which you would like to use for workload identity.
        gcpIAMServiceAccount: ""

    # -- Set this if you are using an external scaler and want to communicate
    # over TLS (recommended). This variable holds the name of the secret that
    # will be mounted to the /grpccerts path on the Pod
    grpcTLSCertsSecret: ""

    # -- Set this if you are using HashiCorp Vault and want to communicate
    # over TLS (recommended). This variable holds the name of the secret that
    # will be mounted to the /vault path on the Pod
    hashiCorpVaultTLS: ""

    logging:
      operator:
        # -- Logging level for KEDA Operator.
        # allowed values: `debug`, `info`, `error`, or an integer value greater than 0, specified as string
        level: info
        # -- Logging format for KEDA Operator.
        # allowed values: `json` or `console`
        format: console
        # -- Logging time encoding for KEDA Operator.
        # allowed values are `epoch`, `millis`, `nano`, `iso8601`, `rfc3339` or `rfc3339nano`
        timeEncoding: rfc3339
        # -- If enabled, the stack traces will be also printed
        stackTracesEnabled: false
      metricServer:
        # -- Logging level for Metrics Server.
        # allowed values: `0` for info, `4` for debug, or an integer value greater than 0, specified as string
        level: 0
        # -- Logging stderrthreshold for Metrics Server
        # allowed values: 'DEBUG','INFO','WARN','ERROR','ALERT','EMERG'
        stderrthreshold: ERROR
      webhooks:
        # -- Logging level for KEDA Operator.
        # allowed values: `debug`, `info`, `error`, or an integer value greater than 0, specified as string
        level: info
        # -- Logging format for KEDA Admission webhooks.
        # allowed values: `json` or `console`
        format: console
        # -- Logging time encoding for KEDA Operator.
        # allowed values are `epoch`, `millis`, `nano`, `iso8601`, `rfc3339` or `rfc3339nano`
        timeEncoding: rfc3339

    # -- [Security context] for all containers
    # @default -- [See below](#KEDA-is-secure-by-default)
    securityContext:
      # -- [Security context] of the operator container
      # @default -- [See below](#KEDA-is-secure-by-default)
      operator:
        capabilities:
          drop:
          - ALL
        allowPrivilegeEscalation: false
        readOnlyRootFilesystem: true
        seccompProfile:
          type: RuntimeDefault
      # -- [Security context] of the metricServer container
      # @default -- [See below](#KEDA-is-secure-by-default)
      metricServer:
        capabilities:
          drop:
          - ALL
        allowPrivilegeEscalation: false
        readOnlyRootFilesystem: true
        seccompProfile:
          type: RuntimeDefault
      # -- [Security context] of the admission webhooks container
      # @default -- [See below](#KEDA-is-secure-by-default)
      webhooks:
        capabilities:
          drop:
          - ALL
        allowPrivilegeEscalation: false
        readOnlyRootFilesystem: true
        seccompProfile:
          type: RuntimeDefault

    # --  [Pod security context] for all pods
    # @default -- [See below](#KEDA-is-secure-by-default)
    podSecurityContext:
      # -- [Pod security context] of the KEDA operator pod
      # @default -- [See below](#KEDA-is-secure-by-default)
      operator:
        runAsNonRoot: true
        # runAsUser: 1000
        # runAsGroup: 1000
        # fsGroup: 1000

      # -- [Pod security context] of the KEDA metrics apiserver pod
      # @default -- [See below](#KEDA-is-secure-by-default)
      metricServer:
        runAsNonRoot: true
        # runAsUser: 1000
        # runAsGroup: 1000
        # fsGroup: 1000

      # -- [Pod security context] of the KEDA admission webhooks
      # @default -- [See below](#KEDA-is-secure-by-default)
      webhooks:
        runAsNonRoot: true
        # runAsUser: 1000
        # runAsGroup: 1000
        # fsGroup: 1000

    service:
      # -- KEDA Metric Server service type
      type: ClusterIP
      # -- HTTPS port for KEDA Metric Server service
      portHttps: 443
      # -- HTTPS port for KEDA Metric Server container
      portHttpsTarget: 6443
      # -- Annotations to add the KEDA Metric Server service
      annotations: {}

    # We provides the default values that we describe in our docs:
    # https://keda.sh/docs/latest/operate/cluster/
    # If you want to specify the resources (or totally remove the defaults), change or comment the following
    # lines, adjust them as necessary, or simply add the curly braces after 'operator' and/or 'metricServer'
    # and remove/comment the default values
    resources:
      # -- Manage [resource request & limits] of KEDA operator pod
      operator:
        limits:
          cpu: 1
          memory: 1000Mi
        requests:
          cpu: 100m
          memory: 100Mi
      # -- Manage [resource request & limits] of KEDA metrics apiserver pod
      metricServer:
        limits:
          cpu: 1
          memory: 1000Mi
        requests:
          cpu: 100m
          memory: 100Mi
      # -- Manage [resource request & limits] of KEDA admission webhooks pod
      webhooks:
        limits:
          cpu: 1
          memory: 1000Mi
        requests:
          cpu: 100m
          memory: 100Mi
    # -- Node selector for pod scheduling ([docs](https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod-node/))
    nodeSelector: {}
    # -- Tolerations for pod scheduling ([docs](https://kubernetes.io/docs/concepts/scheduling-eviction/taint-and-toleration/))
    tolerations: []

    topologySpreadConstraints:
      # -- [Pod Topology Constraints] of KEDA operator pod
      operator: []
      # -- [Pod Topology Constraints] of KEDA metrics apiserver pod
      metricsServer: []
      # -- [Pod Topology Constraints] of KEDA admission webhooks pod
      webhooks: []

    # -- [Affinity] for pod scheduling for KEDA operator, Metrics API Server and KEDA admission webhooks.
    affinity: {}
      # podAntiAffinity:
      #   requiredDuringSchedulingIgnoredDuringExecution:
      #   - labelSelector:
      #       matchExpressions:
      #       - key: app
      #         operator: In
      #         values:
      #         - keda-operator
      #         - keda-operator-metrics-apiserver
      #         - keda-admission-webhooks
      #     topologyKey: "kubernetes.io/hostname"

    # -- priorityClassName for all KEDA components
    priorityClassName: ""

    ## The default HTTP timeout in milliseconds that KEDA should use
    ## when making requests to external services. Removing this defaults to a
    ## reasonable default
    http:
      # -- The default HTTP timeout to use for all scalers that use raw HTTP clients (some scalers use SDKs to access target services. These have built-in HTTP clients, and the timeout does not necessarily apply to them)
      timeout: 3000
      keepAlive:
        # -- Enable HTTP connection keep alive
        enabled: true
      # -- The minimum TLS version to use for all scalers that use raw HTTP clients (some scalers use SDKs to access target services. These have built-in HTTP clients, and this value does not necessarily apply to them)
      minTlsVersion: TLS12

    ## This setting lets you enable profiling for all of the components of KEDA and in the specific port you choose
    ## This can be useful when trying to investigate errors like memory leaks or CPU or even look at goroutines to understand better
    ## This setting is disabled by default
    profiling:
      operator:
        # -- Enable profiling for KEDA operator
        enabled: false
        # -- Expose profiling on a specific port
        port: 8082
      metricsServer:
        # -- Enable profiling for KEDA metrics server
        enabled: false
        # -- Expose profiling on a specific port
        port: 8083
      webhooks:
        # -- Enable profiling for KEDA admission webhook
        enabled: false
        # -- Expose profiling on a specific port
        port: 8084


    ## Extra KEDA Operator and Metrics Adapter container arguments
    extraArgs:
      # -- Additional KEDA Operator container arguments
      keda: {}
      # -- Additional Metrics Adapter container arguments
      metricsAdapter: {}

    # -- Additional environment variables that will be passed onto all KEDA components
    env: []
    # - name: ENV_NAME
    #   value: 'ENV-VALUE'

    # Extra volumes and volume mounts for the deployment. Optional.
    volumes:
      keda:
        # -- Extra volumes for KEDA deployment
        extraVolumes: []
        # -- Extra volume mounts for KEDA deployment
        extraVolumeMounts: []

      metricsApiServer:
        # -- Extra volumes for metric server deployment
        extraVolumes: []
        # -- Extra volume mounts for metric server deployment
        extraVolumeMounts: []

      webhooks:
        # -- Extra volumes for admission webhooks deployment
        extraVolumes: []
        # -- Extra volume mounts for admission webhooks deployment
        extraVolumeMounts: []

    prometheus:
      metricServer:
        # -- Enable metric server Prometheus metrics expose
        enabled: false
        # -- HTTP port used for exposing metrics server prometheus metrics
        port: 8080
        # -- HTTP port name for exposing metrics server prometheus metrics
        portName: metrics
        serviceMonitor:
          # -- Enables ServiceMonitor creation for the Prometheus Operator
          enabled: false
          # -- JobLabel selects the label from the associated Kubernetes service which will be used as the job label for all metrics. [ServiceMonitor Spec]
          jobLabel: ""
          # -- TargetLabels transfers labels from the Kubernetes `Service` onto the created metrics
          targetLabels: []
          # -- PodTargetLabels transfers labels on the Kubernetes `Pod` onto the created metrics
          podTargetLabels: []
          # -- Name of the service port this endpoint refers to. Mutually exclusive with targetPort
          port: metrics
          # -- Name or number of the target port of the Pod behind the Service, the port must be specified with container port property. Mutually exclusive with port
          targetPort: ""
          # -- Interval at which metrics should be scraped If not specified Prometheus’ global scrape interval is used.
          interval: ""
          # -- Timeout after which the scrape is ended If not specified, the Prometheus global scrape timeout is used unless it is less than Interval in which the latter is used
          scrapeTimeout: ""
          # -- DEPRECATED. List of expressions that define custom relabeling rules for metric server ServiceMonitor crd (prometheus operator). [RelabelConfig Spec]
          relabellings: []
          # -- List of expressions that define custom relabeling rules for metric server ServiceMonitor crd (prometheus operator). [RelabelConfig Spec]
          relabelings: []
          # -- List of expressions that define custom  metric relabeling rules for metric server ServiceMonitor crd after scrape has happened (prometheus operator). [RelabelConfig Spec]
          metricRelabelings: []
          # --  Additional labels to add for metric server using ServiceMonitor crd (prometheus operator)
          additionalLabels: {}
          # -- HTTP scheme used for scraping. Defaults to `http`
          scheme: http
          # -- TLS configuration for scraping metrics
          tlsConfig: {}
            # caFile: /etc/prom-certs/root-cert.pem
            # certFile: /etc/prom-certs/cert-chain.pem
            # insecureSkipVerify: true
            # keyFile: /etc/prom-certs/key.pem
        podMonitor:
          # -- Enables PodMonitor creation for the Prometheus Operator
          enabled: false
          # -- Scraping interval for metric server using podMonitor crd (prometheus operator)
          interval: ""
          # -- Scraping timeout for metric server using podMonitor crd (prometheus operator)
          scrapeTimeout: ""
          # -- Scraping namespace for metric server using podMonitor crd (prometheus operator)
          namespace: ""
          # -- Additional labels to add for metric server using podMonitor crd (prometheus operator)
          additionalLabels: {}
          # -- List of expressions that define custom relabeling rules for metric server podMonitor crd (prometheus operator)
          relabelings: []
          # -- List of expressions that define custom  metric relabeling rules for metric server PodMonitor crd after scrape has happened (prometheus operator). [RelabelConfig Spec]
          metricRelabelings: []
      operator:
        # -- Enable KEDA Operator prometheus metrics expose
        enabled: false
        # -- Port used for exposing KEDA Operator prometheus metrics
        port: 8080
        serviceMonitor:
          # -- Enables ServiceMonitor creation for the Prometheus Operator
          enabled: false
          # -- JobLabel selects the label from the associated Kubernetes service which will be used as the job label for all metrics. [ServiceMonitor Spec]
          jobLabel: ""
          # -- TargetLabels transfers labels from the Kubernetes `Service` onto the created metrics
          targetLabels: []
          # -- PodTargetLabels transfers labels on the Kubernetes `Pod` onto the created metrics
          podTargetLabels: []
          # -- Name of the service port this endpoint refers to. Mutually exclusive with targetPort
          port: metrics
          # -- Name or number of the target port of the Pod behind the Service,
          # the port must be specified with container port property. Mutually exclusive with port
          targetPort: ""
          # -- Interval at which metrics should be scraped If not specified Prometheus’ global scrape interval is used.
          interval: ""
          # -- Timeout after which the scrape is ended If not specified, the Prometheus global scrape timeout is used unless it is less than Interval in which the latter is used
          scrapeTimeout: ""
          # -- DEPRECATED. List of expressions that define custom relabeling rules for metric server ServiceMonitor crd (prometheus operator). [RelabelConfig Spec]
          relabellings: []
          # -- List of expressions that define custom relabeling rules for metric server ServiceMonitor crd (prometheus operator). [RelabelConfig Spec]
          relabelings: []
          # -- List of expressions that define custom  metric relabeling rules for metric server ServiceMonitor crd after scrape has happened (prometheus operator). [RelabelConfig Spec]
          metricRelabelings: []
          # -- Additional labels to add for metric server using ServiceMonitor crd (prometheus operator)
          additionalLabels: {}
          # -- HTTP scheme used for scraping. Defaults to `http`
          scheme: http
          # -- TLS configuration for scraping metrics
          tlsConfig: {}
            # caFile: /etc/prom-certs/root-cert.pem
            # certFile: /etc/prom-certs/cert-chain.pem
            # insecureSkipVerify: true
            # keyFile: /etc/prom-certs/key.pem
        podMonitor:
          # -- Enables PodMonitor creation for the Prometheus Operator
          enabled: false
          # -- Scraping interval for KEDA Operator using podMonitor crd (prometheus operator)
          interval: ""
          # -- Scraping timeout for KEDA Operator using podMonitor crd (prometheus operator)
          scrapeTimeout: ""
          # -- Scraping namespace for KEDA Operator using podMonitor crd (prometheus operator)
          namespace: ""
          # -- Additional labels to add for KEDA Operator using podMonitor crd (prometheus operator)
          additionalLabels: {}
          # --  List of expressions that define custom relabeling rules for KEDA Operator podMonitor crd (prometheus operator)
          relabelings: []
          # -- List of expressions that define custom  metric relabeling rules for metric server PodMonitor crd after scrape has happened (prometheus operator). [RelabelConfig Spec]
          metricRelabelings: []
        prometheusRules:
          # -- Enables PrometheusRules creation for the Prometheus Operator
          enabled: false
          # -- Scraping namespace for KEDA Operator using prometheusRules crd (prometheus operator)
          namespace: ""
          # -- Additional labels to add for KEDA Operator using prometheusRules crd (prometheus operator)
          additionalLabels: {}
          # -- Additional alerts to add for KEDA Operator using prometheusRules crd (prometheus operator)
          alerts:
            []
            # - alert: KedaScalerErrors
            #   annotations:
            #     description: Keda scaledObject {{ $labels.scaledObject }} is experiencing errors with {{ $labels.scaler }} scaler
            #     summary: Keda Scaler {{ $labels.scaler }} Errors
            #   expr: sum by ( scaledObject , scaler) (rate(keda_metrics_adapter_scaler_errors[2m]))  > 0
            #   for: 2m
            #   labels:
      webhooks:
        # -- Enable KEDA admission webhooks prometheus metrics expose
        enabled: false
        # -- Port used for exposing KEDA admission webhooks prometheus metrics
        port: 8080
        serviceMonitor:
          # -- Enables ServiceMonitor creation for the Prometheus webhooks
          enabled: false
          # -- jobLabel selects the label from the associated Kubernetes service which will be used as the job label for all metrics. [ServiceMonitor Spec]
          jobLabel: ""
          # -- TargetLabels transfers labels from the Kubernetes `Service` onto the created metrics
          targetLabels: []
          # -- PodTargetLabels transfers labels on the Kubernetes `Pod` onto the created metrics
          podTargetLabels: []
          # -- Name of the service port this endpoint refers to. Mutually exclusive with targetPort
          port: metrics
          # -- Name or number of the target port of the Pod behind the Service, the port must be specified with container port property. Mutually exclusive with port
          targetPort: ""
          # -- Interval at which metrics should be scraped If not specified Prometheus’ global scrape interval is used.
          interval: ""
          # -- Timeout after which the scrape is ended If not specified, the Prometheus global scrape timeout is used unless it is less than Interval in which the latter is used
          scrapeTimeout: ""
          # -- DEPRECATED. List of expressions that define custom relabeling rules for metric server ServiceMonitor crd (prometheus operator). [RelabelConfig Spec]
          relabellings: []
          # -- List of expressions that define custom relabeling rules for metric server ServiceMonitor crd (prometheus operator). [RelabelConfig Spec]
          relabelings: []
          # -- List of expressions that define custom  metric relabeling rules for metric server ServiceMonitor crd after scrape has happened (prometheus operator). [RelabelConfig Spec]
          metricRelabelings: []
          # -- Additional labels to add for metric server using ServiceMonitor crd (prometheus operator)
          additionalLabels: {}
          # -- HTTP scheme used for scraping. Defaults to `http`
          scheme: http
          # -- TLS configuration for scraping metrics
          tlsConfig: {}
            # caFile: /etc/prom-certs/root-cert.pem
            # certFile: /etc/prom-certs/cert-chain.pem
            # insecureSkipVerify: true
            # keyFile: /etc/prom-certs/key.pem
        prometheusRules:
          # -- Enables PrometheusRules creation for the Prometheus Operator
          enabled: false
          # -- Scraping namespace for KEDA admission webhooks using prometheusRules crd (prometheus operator)
          namespace: ""
          # -- Additional labels to add for KEDA admission webhooks using prometheusRules crd (prometheus operator)
          additionalLabels: {}
          # -- Additional alerts to add for KEDA admission webhooks using prometheusRules crd (prometheus operator)
          alerts: []

    opentelemetry:
      collector:
        # -- Uri of OpenTelemetry Collector to push telemetry to
        uri: ""
      operator:
          # -- Enable pushing metrics to an OpenTelemetry Collector for operator
          enabled: false

    certificates:
      # -- Enables the self generation for KEDA TLS certificates inside KEDA operator
      autoGenerated: true
      # -- Secret name to be mounted with KEDA TLS certificates
      secretName: kedaorg-certs
      # -- Path where KEDA TLS certificates are mounted
      mountPath: /certs
      certManager:
        # -- Enables Cert-manager for certificate management
        enabled: false
        # -- Certificate duration
        duration: 8760h0m0s # 1 year
        # -- Certificate renewal time before expiration
        renewBefore: 5840h0m0s # 8 months
        # -- Generates a self-signed CA with Cert-manager.
        # If generateCA is false, the secret with the CA
        # has to be annotated with `cert-manager.io/allow-direct-injection: "true"`
        generateCA: true
        # -- Secret name where the CA is stored (generatedby cert-manager or user given)
        caSecretName: "kedaorg-ca"
        # -- Add labels/annotations to secrets created by Certificate resources
        # [docs](https://cert-manager.io/docs/usage/certificate/#creating-certificate-resources)
        secretTemplate: {}
          # annotations:
          #   my-secret-annotation-1: "foo"
          #   my-secret-annotation-2: "bar"
          # labels:
          #   my-secret-label: foo
        # -- Reference to custom Issuer. If issuer.generate is false, then issuer.group, issuer.kind and issuer.name are required
        issuer:
          # -- Generates an Issuer resource with Cert-manager
          generate: true
          # -- Custom Issuer name. Required when generate: false
          name: foo-org-ca
          # -- Custom Issuer kind. Required when generate: false
          kind: ClusterIssuer
          # -- Custom Issuer group. Required when generate: false
          group: cert-manager.io
      operator:
        # -- Location(s) of CA files for authentication of external TLS connections such as TLS-enabled metrics sources
        # caDirs:
        # - /custom/ca

    permissions:
      metricServer:
        restrict:
          # -- Restrict Secret Access for Metrics Server
          secret: false
      operator:
        restrict:
          # -- Restrict Secret Access for KEDA operator
          # if true, KEDA operator will be able to read only secrets in {{ .Release.Namespace }} namespace
          secret: false
          # -- Array of strings denoting what secrets the KEDA operator will be able to read, this takes into account
          # also the configured `watchNamespace`.
          # the default is an empty array -> no restriction on the secret name
          namesAllowList: []

    # -- Array of extra K8s manifests to deploy
    extraObjects: []
      # - apiVersion: keda.sh/v1alpha1
      #   kind: ClusterTriggerAuthentication
      #   metadata:
      #     name: aws-credentials
      #     namespace: keda
      #   spec:
      #     podIdentity:
      #       provider: aws-eks

    # -- Capability to turn on/off ASCII art in Helm installation notes
    asciiArt: true

    # -- When specified, each rendered resource will have `app.kubernetes.io/managed-by: ${this}` label on it. Useful, when using only helm template with some other solution.
    customManagedBy: ""

Подключение модуля Keda Config

Заметка

Модуль Keda Config содержит дополнительные настройки и требуется только для управляющего кластера.

Для подчиненного кластера модуль недоступен

Описание Yaml

Внимание!

Values без описания (1) являются продвинутыми настройками и редактировать их не рекомендовано

1. Описание

apiVersion: addon.bootsman.tech/v1alpha1
kind: Config
metadata:
  name: CLUSTER_NAME-keda-config
  namespace: CLUSTER_NAMESPACE
spec:
  enabled: true (1)

1. True - включено.

   False - выключено

Настройка в UI

Примеры использования

Документация

Более полная документация по модулю:
Keda Docs
Keda ScaledObject Spec

Основные тезисы о работе с Keda:

1. Объекты и правила скалирования определяются объектом ScaledObject.
2. Для каждого объекта кластера может быть только один ScaledObject

Автоматическое масштабирование экземпляров приложения (pods)

Описание полей

ScaledObject.yaml

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: SO_NAME (1)
  namespace: SO_NAMESPACE (2)
spec:
  maxReplicaCount: MAX_REPLICA_COUNT (3)
  minReplicaCount: MIN_REPLICA_COUNT (4)
  scaleTargetRef: 
    apiVersion: API_VERSION_TARGET_KIND (5)
    kind: TARGET_KIND (6)
    name: TARGET_NAME (7)
  triggers: (8)
    - type: prometheus (9) # Тип триггера
      metadata: (10)
        (10)serverAddress: http://rancher-monitoring-prometheus.cattle-monitoring-system.svc.cluster.local:9090 # Адрес источника
        (11)threshold: '100'
        (12)query: sum(node_namespace_pod_container:container_cpu_usage_seconds_total:sum_irate{namespace="default"} * on(namespace,pod) group_left(workload, workload_type) namespace_workload_pod:kube_pod_owner:relabel{namespace="default",workload_type="deployment",workload="my-deployment"}) by (workload, workload_type)

1. Имя объекта ScaledObject
2. Имя Namespace, ScaledObject должен быть размещен в том же namespace, что и объект масштабирования
3. Опционально.

   Максимальное число реплик целевого ресурса.

   По умолчанию: 100

4. Опционально.

   Минимальное количество реплик целевого ресурса.

   По умолчанию: 0

5. Опционально.

   Версия API kind целевого ресурса.

   По умолчанию: apps/v1

6. Опционально.

   Тип (kind) объекта целевого ресурса.

   По умолчанию: Deployment

7. Имя целевого ресурса.
8. Список правил активации для масштабирования целевых ресурсов
9. Тип триггера.

   Со всеми поддерживаемыми типами можно ознакомится в документации

10. Данные для работы выбранного типа триггера
11. Актуально только для type: prometheus.

    Адрес источника, который содержит метрики prometheus

12. Актуально только для type: prometheus.

    Верхняя граница для запуска процесса масштабирования

13. Актуально только для type: prometheus.

    Prometheus запрос, на основе которого будет принято решение о масштабирования

Создание автоматически масштабируемого Deployment

Примените в кластере ScaledObject:

Пример заполненного ScaledObject

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: my-deployment-scaledobject
  namespace: default
  labels:
    deploymentName: dummy
spec:
  maxReplicaCount: 5
  minReplicaCount: 1
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: load
  triggers:
    - type: prometheus
      metadata:
        serverAddress: http://rancher-monitoring-prometheus.cattle-monitoring-system.svc.cluster.local:9090
        threshold: '1'
        query: sum(node_namespace_pod_container:container_cpu_usage_seconds_total:sum_irate{namespace="default"} * on(namespace,pod) group_left(workload, workload_type) namespace_workload_pod:kube_pod_owner:relabel{namespace="default",workload_type="deployment",workload="load"}) by (workload, workload_type)

Запустите Deployment, который следует масштабировать:

Пример приложения

kubectl create deployment load --image=vish/stress --replicas 1 -- /stress --cpus 2 --mem-total 512M --logtostderr

Проверить корректность настроек:

kubectl get hpa

Пример вывода

NAME                                  REFERENCE         TARGETS         MINPODS   MAXPODS   REPLICAS   AGE
keda-hpa-my-deployment-scaledobject   Deployment/load   1903m/1 (avg)   1         5         5          3m38s

Проверить текущие значения, на основе которых происходит масштабирование:

kubectl get hpa keda-hpa-my-deployment-scaledobject -o jsonpath="{.status.currentMetrics[*]}"

Пример вывода

{"external":{"current":{"averageValue":"1909m"},"metric":{"name":"s0-prometheus","selector":{"matchLabels":{"scaledobject.keda.sh/name":"my-deployment-scaledobject"}}}},"type":"External"}

Проверить текущий статус:

kubectl get scaledobject my-deployment-scaledobject -n default -o jsonpath="{.status.health}"

Пример вывода

{"s0-prometheus":{"numberOfFailures":0,"status":"Happy"}}

Автоматическое масштабирование узлов WorkerPool подчиненного кластера

Keda позволяет автоматически масштабировать WorkerPool подчиненного кластера

Для этого сначала нужно получить ID подчиненного кластера:

kubectl get clusters.provisioning.bootsman.tech SUBCLUSTER_NAME (1) -o=jsonpath='{.status.rancherClusterRef.name}'

1. Имя подчиненного кластера

Пример вывода

c-m-wb84fv65

Описание полей

ScaledObject.yaml

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: SO_NAME (1)
  namespace: SO_NAMESPACE (2)
  annotations:
    scaledobject.keda.sh/transfer-hpa-ownership: TRANSFER_HPA_OWNERSHIP (3)
    validations.keda.sh/hpa-ownership: HPA_OWNERSHIP (4)
    autoscaling.keda.sh/paused: PAUSED (5)
spec:
  scaleTargetRef:
    apiVersion: provisioning.bootsman.tech/v1alpha1
    kind: Workerpool
    name: WORKERPOOL_NAME (6)
  pollingInterval: POLLING_INTERVAL (7)
  cooldownPeriod: COOLDOWN_PERIOD (8)
  minReplicaCount: MIN_REPLICA_COUNT (9)
  maxReplicaCount: MAX_REPLICA_COUNT (10)
  fallback: (11)
    failureThreshold: FALLBACK_FAILURE_THRESHOLD (12)
    replicas: FALLBACK_REPLICAS (13)
  triggers: (14)
    - type: prometheus (15)
      metadata: (16)
        authModes: "basic"
        (17)serverAddress: https://rancher.cattle-system/k8s/clusters/SUBCLUSTER_ID/api/v1/namespaces/cattle-monitoring-system/services/http:rancher-monitoring-prometheus:9090/proxy
        unsafeSsl: "true"
        (18)query: sum(100 * avg(1 - rate(node_cpu_seconds_total{mode="idle",nodename=~".*-md-.*"}[1m])) by (nodename))
        (19)threshold: "50"
      authenticationRef:
        kind: ClusterTriggerAuthentication 
        (20)name: keda-trigger-auth

1. Имя объекта ScaledObject
2. Имя Namespace, ScaledObject должен быть размещен в том же namespace, что и объект масштабирования.

   В данном случае в том же namespace, что и подчиненный кластер

3. Опционально.

   Передача управления существующего HPA во владение объекту ScaledObject.

   По умолчанию: true

4. Опционально.

   Проверка владения HPA объектом ScaledObject.

   По умолчанию: true

5. Опционально.

   Приостановка работы автоматического масштабирования.

   По умолчанию: false

6. Имя Workerpool, который будет масштабироваться
7. Опционально.

   Интервал проверки в секундах.

   По умолчанию: 30

8. Опционально.

   Период ожидания после того, как последний триггер сообщил об активности, прежде чем масштабировать ресурс обратно до 0, в секундах

   По умолчанию: 300

9. Опционально.

   Минимальное количество узлов Workerpool.

   По умолчанию: 0

10. Опционально.

    Максимальное количество узлов Workerpool.

    По умолчанию: 100

11. Опционально.

    Поведение в случае наличия ошибок в масштабировании.

12. Число неудачных попыток получения данных, при которых значение узлов вернется к fallback.replacas
13. Число реплик, при ошибках при получения метрик
14. Список правил активации для масштабирования целевых ресурсов
15. Тип триггера.

    Со всеми поддерживаемыми типами можно ознакомится в документации

16. Данные для работы выбранного типа триггера
17. Путь для получения метрик подчиненного кластера.

    Нужно использовать SUBCLUSTER_ID

18. Запрос, основе которого применяется решение о масштабировании
19. Пороговое значение, превышение которого будет запускать процесс масштабирования
20. Имя Secret для авторизации. Не менять.

Создание автоматически масштабируемого Workerpool

Данные подчиненного кластера

SUBCLUSTER_ID=c-m-w4lfkvns
SUBCLUSTER_NAME=sub-example-worker
SUBCLUSTER_NAMESPACE=default

Средняя загрузка CPU на узлах более 80%

ScaledObject.yaml

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: cluster-a
  namespace: default
  annotations:
    scaledobject.keda.sh/transfer-hpa-ownership: "true"
    validations.keda.sh/hpa-ownership: "true"
    autoscaling.keda.sh/paused: "false"
spec:
  scaleTargetRef:
    apiVersion: provisioning.bootsman.tech/v1alpha1
    kind: WorkerPool
    name: sub-example-worker
  pollingInterval: 10
  cooldownPeriod: 150
  minReplicaCount: 3
  maxReplicaCount: 5
  fallback:
    failureThreshold: 3
    replicas: 5
  triggers:
    - type: prometheus
      metadata:
        authModes: "basic"
        serverAddress: https://rancher.cattle-system/k8s/clusters/c-m-w4lfkvns/api/v1/namespaces/cattle-monitoring-system/services/http:rancher-monitoring-prometheus:9090/proxy
        unsafeSsl: "true"
        query: sum(100 * avg(1 - rate(node_cpu_seconds_total{mode="idle",nodename=~".*-md-.*"}[1m])) by (nodename))
        threshold: "50"
      authenticationRef:
        kind: ClusterTriggerAuthentication
        name: keda-trigger-auth

Средняя загрузка RAM на узлах более 80%

ScaledObject.yaml

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: cluster-a
  namespace: default
  annotations:
    scaledobject.keda.sh/transfer-hpa-ownership: "true"
    validations.keda.sh/hpa-ownership: "true"
    autoscaling.keda.sh/paused: "false"
spec:
  scaleTargetRef:
    apiVersion: provisioning.bootsman.tech/v1alpha1
    kind: WorkerPool
    name: sub-example-worker
  pollingInterval: 10
  cooldownPeriod: 150
  minReplicaCount: 3
  maxReplicaCount: 5
  fallback:
    failureThreshold: 3
    replicas: 5
  triggers:
    - type: prometheus
      metadata:
        authModes: "basic"
        serverAddress: https://rancher.cattle-system/k8s/clusters/c-m-w4lfkvns/api/v1/namespaces/cattle-monitoring-system/services/http:rancher-monitoring-prometheus:9090/proxy
        unsafeSsl: "true"
        query: sum((1 - (node_memory_MemAvailable_bytes{nodename=~".*-md-.*"} / node_memory_MemTotal_bytes{nodename=~".*-md-.*"})) * 100)
        threshold: "40"
      authenticationRef:
        kind: ClusterTriggerAuthentication
        name: keda-trigger-auth

Среднее число подов на каждом узле более 200

ScaledObject.yaml

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: cluster-a
  namespace: default
  annotations:
    scaledobject.keda.sh/transfer-hpa-ownership: "true"
    validations.keda.sh/hpa-ownership: "true"
    autoscaling.keda.sh/paused: "false"
spec:
  scaleTargetRef:
    apiVersion: provisioning.bootsman.tech/v1alpha1
    kind: WorkerPool
    name: sub-example-worker
  pollingInterval: 10
  cooldownPeriod: 150
  minReplicaCount: 3
  maxReplicaCount: 5
  fallback:
    failureThreshold: 3
    replicas: 5
  triggers:
    - type: prometheus
      metadata:
        authModes: "basic"
        serverAddress: https://rancher.cattle-system/k8s/clusters/c-m-w4lfkvns/api/v1/namespaces/cattle-monitoring-system/services/http:rancher-monitoring-prometheus:9090/proxy
        unsafeSsl: "true"
        query: sum(kubelet_running_pods{node=~".*-md-.*"}) OR sum(kubelet_running_pod_count{node=~".*-md-.*"})
        threshold: "200"
      authenticationRef:
        kind: ClusterTriggerAuthentication
        name: keda-trigger-auth

Пример нескольких правил масштабирования в одном ScaledObject:

ScaledObject.yaml

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: cluster-a
  namespace: default
  annotations:
    scaledobject.keda.sh/transfer-hpa-ownership: "true"
    validations.keda.sh/hpa-ownership: "true"
    autoscaling.keda.sh/paused: "false"
spec:
  scaleTargetRef:
    apiVersion: provisioning.bootsman.tech/v1alpha1
    kind: WorkerPool
    name: sub-example-worker
  pollingInterval: 10
  cooldownPeriod: 150
  minReplicaCount: 3
  maxReplicaCount: 5
  fallback:
    failureThreshold: 3
    replicas: 5
  triggers:
    - type: prometheus
      metadata:
        authModes: "basic"
        serverAddress: https://rancher.cattle-system/k8s/clusters/c-m-w4lfkvns/api/v1/namespaces/cattle-monitoring-system/services/http:rancher-monitoring-prometheus:9090/proxy
        unsafeSsl: "true"
        query: sum(100 * avg(1 - rate(node_cpu_seconds_total{mode="idle",nodename=~".*-md-.*"}[1m])) by (nodename))
        threshold: "50"
      authenticationRef:
        kind: ClusterTriggerAuthentication
        name: keda-trigger-auth
    - type: prometheus
      metadata:
        authModes: "basic"
        serverAddress: https://rancher.cattle-system/k8s/clusters/c-m-w4lfkvns/api/v1/namespaces/cattle-monitoring-system/services/http:rancher-monitoring-prometheus:9090/proxy
        unsafeSsl: "true"
        query: sum((1 - (node_memory_MemAvailable_bytes{nodename=~".*-md-.*"} / node_memory_MemTotal_bytes{nodename=~".*-md-.*"})) * 100)
        threshold: "40"
      authenticationRef:
        kind: ClusterTriggerAuthentication
        name: keda-trigger-auth