values.yaml

# The name of the service and pods/deployment/etc being deployed
name: volume-autoscaler

# The typical config variables for the volume-autoscaler service, and their respective defaults

# This is how we we can deploy this chart to a different namespace (or cluster) than Prometheus, by telling it where to access Prometheus
#   Eg: http://prometheus-server.namespace.svc.cluster.local.  This can be auto-detected if run in the same namespace as prometheus
# This is also how you can run this script remotely (eg: on your local machine to develop this tool against an remote Prometheus server)
prometheus_url: ""
# This is how we confiure the Slack integration, set the two values below
slack_webhook_url: "REPLACEME"
slack_channel: "REPLACEME"
# Our slack message prefix and suffix.  Optional, can use to help distinguish between different clusters or to alert someone
slack_message_prefix: ""
slack_message_suffix: ""
# Our scan interval
interval_time: "60"
# How many scan intervals in alert before scaling
scale_after_intervals: "5"
# How much full the disk must be before considering scaling
scale_above_percent: "80"
# How much to scale disks up by, as a percentage of their current size
scale_up_percent: "20"
# An minimum amount of bytes to scale up by (typically because providers like AWS only support 1GB increments in disk size)
scale_up_min_increment: "1000000000"
# An maximum amount of bytes to scale up by (helps prevent large disks from growing too fast/exponentially).  Set to 16TB by default, which basically means this is disabled
scale_up_max_increment: "16000000000000"
# The maximum size of disks to scale up to.  By default on AWS using EBS volumes this is set to 16TB as that is the EBS Max disk size.
scale_up_max_size: "16000000000000"
# How long to wait inbetween scaling activities.  AWS restricts you to one resize per-disk per 6 hour period, so this is 6 hours plus 10 minutes (as a buffer)
scale_cooldown_time: "22200"
# Set this to "true" to only print what it would have done, it will not send Slack messages either if you enable this
dry_run: ""
# What label selector to query on prometheus to limit the volumes you wish to use.  For example set this to "namespace=\"production\"" to limit to PVCs in production namespace
prometheus_label_match: ""
http_timeout: "15"
# For debugging
verbose: "false"
# Victoriametrics or prometheus
victoriametrics_mode: "false"
# Auth header for mimir or cortex
scope_orgid_auth_header: ""


# Pretty much ignore anything below here I'd say, unless you really know what you're doing.  :)

# Number of pods in deployment, we only support 1 running
replicaCount: 1

# Can only have 1 running at a time, this service is not highly-available
deploymentStrategy:
  type: Recreate

# Our image is on DockerHub, replace tag during builds automatically
image:
  repository: devopsnirvana/kubernetes-volume-autoscaler
  tag: latest  # This should be replaced with an semver tag release automatically during CI for releases

globalEnvs:
  ####################################
  # !!! Note: You don't need to do anything in here, please set the values at the top of this file instead
  ####################################

  # This is how we we can deploy this chart to a different namespace (or cluster) than Prometheus, by telling it where to access Prometheus
  - name: PROMETHEUS_URL
    value: "{{ .Values.prometheus_url }}"

  # This is how we confiure the Slack integration, set the values below
  - name: SLACK_WEBHOOK_URL
    value: "{{ .Values.slack_webhook_url }}"
  - name: SLACK_CHANNEL
    value: "{{ .Values.slack_channel }}"
  - name: SLACK_MESSAGE_PREFIX
    value: "{{ .Values.slack_message_prefix }}"
  - name: SLACK_MESSAGE_SUFFIX
    value: "{{ .Values.slack_message_suffix }}"

  # Our scan interval
  - name: INTERVAL_TIME
    value: "{{ .Values.interval_time }}"
  # How many scan intervals in alert before scaling
  - name: SCALE_AFTER_INTERVALS
    value: "{{  .Values.scale_after_intervals }}"

  # How much full the disk must be before considering scaling
  - name: SCALE_ABOVE_PERCENT
    value: "{{ .Values.scale_above_percent }}"

  # How much to scale disks up by, as a percentage of their current size
  - name: SCALE_UP_PERCENT
    value: "{{ .Values.scale_up_percent }}"

  # An minimum amount of bytes to scale up by (typically because providers like AWS only support 1GB increments in disk size)
  - name: SCALE_UP_MIN_INCREMENT
    value: "{{ .Values.scale_up_min_increment }}"
  # An maximum amount of bytes to scale up by (helps prevent large disks from growing too fast/exponentially).  Set to 16TB by default, which basically means this is disabled
  - name: SCALE_UP_MAX_INCREMENT
    value: "{{ .Values.scale_up_max_increment }}"
  # The maximum size of disks to scale up to.  By default on AWS using EBS volumes this is set to 16TB as that is the EBS Max disk size.
  - name: SCALE_UP_MAX_SIZE
    value: "{{ .Values.scale_up_max_size }}"
  # How long to wait inbetween scaling activities.  AWS restricts you to one resize per-disk per 6 hour period, so this is 6 hours plus 10 minutes (as a buffer)
  - name: SCALE_COOLDOWN_TIME
    value: "{{ .Values.scale_cooldown_time }}"
  # Set this to "true" to only print what it would have done, it will not send Slack messages either if you enable this
  - name: DRY_RUN
    value: "{{ .Values.dry_run }}"
  # What label selector to query on prometheus to limit the volumes you wish to use.  For example set this to "namespace=\"production\"" to limit to PVCs in production namespace
  - name: PROMETHEUS_LABEL_MATCH
    value: "{{ .Values.prometheus_label_match }}"
  # How long to wait for Prometheus and Kubernetes to get back to us for all API calls
  - name: HTTP_TIMEOUT
    value: "{{ .Values.http_timeout }}"

  # How verbose to be during runtime.  Off by default, when enabled it prints every volume and their specifications on every iteration.  Recommended for testing/development and/or dry-runs
  - name: VERBOSE
    value: "{{ .Values.verbose }}"

  # Which monitoring stack you're running
  - name: VICTORIAMETRICS_MODE
    value: "{{ .Values.victoriametrics_mode }}"

  # If you are using Cortex or Mimir you'll probably need to set an scope OrgID auth header for it
  - name: SCOPE_ORGID_AUTH_HEADER
    value: "{{ .Values.scope_orgid_auth_header }}"


# Additional pod annotations
podAnnotations: {}
  # tick: "1528451892"

# Additional labels put onto anything that can be labelled (pods, services)
labels: {}

# This adds lifecycle events for the deployment
lifecycle: {}

# Lower the ndots value, to reduce the search path expansion of DNS queries
# https://kubernetes.io/docs/concepts/services-networking/dns-pod-service/#pod-s-dns-config
dnsConfig:
  enabled: false
  ndots: 2

# container resource requests/limits
# The suggested values below have been tested against 100+ pvcs.  Your usage may vary, so
#    adjust this upwards if you need.
# Note: Limits are HARD Limits
#       Requests are "soft" limits and are what affects HPA (autoscaling) aggressiveness and are what resources are is guaranteed
# To use below, remove the {} and uncomment the following 6 lines and customize them as desired
resources: {}
  # limits:
  #   cpu: 20m
  #   memory: 75Mi
  # requests:
  #   cpu: 20m
  #   memory: 75Mi
  # NOTE: Above resources have been used successfully on numerous clients and customers, but, just incase and what the default for most
  #       helm charts is to keep this disabled, and let the user specify it.  So, feel free if you'd like

# Assign pods to nodes based on nodeSelector labels, define a default here if desired
# To use below, remove the {} and uncomment the following 1 lines and/or add your own lines
nodeSelector:
  # This runs on Linux only
  kubernetes.io/os: linux
# purpose: node-feature-name

# Assign pods to nodes based on tolerations and taints
tolerations: []

# Init container(s)
initContainers: []
  # - name: volume-mount-hack
  #   image: busybox
  #   command: ["sh", "-c", "chown -R 1000:1000 /var/data"]
  #   volumeMounts:
  #    - name: data
  #      mountPath: /var/data

# Additional containers to be added to the pod (eg: add the cloudsql proxy)
extraContainers: []

# Volumes added to the pod if needed
volumes: []
volumeMounts: []

# We use a service so we can have prometheus metrics
service:
  enabled: true
  type: ClusterIP
  port: 80
  targetPort: 8000
  name: default-service
  additionalPorts: []
  annotations:
    prometheus.io/scrape: 'true'

# Probes to detect failure
livenessProbe:
  enabled: true
  initialDelaySeconds: 120
  periodSeconds: 10
  timeoutSeconds: 9
  successThreshold: 1
  failureThreshold: 3
  probeType: httpGet
  scheme: HTTP
  path: /alive
  port: default-service
readinessProbe:
  enabled: true
  initialDelaySeconds: 5
  periodSeconds: 5
  timeoutSeconds: 4
  successThreshold: 2
  failureThreshold: 2
  probeType: httpGet
  scheme: HTTP
  path: /ready
  port: default-service

# For AWS Security Groups, if you wish to use this specify the sg-123123123 in an list here
securityGroupIDs: []

serviceAccount:
  enabled: true
  # Add annotations if desited to your service account
  annotations: {}
    # Here's an example of assigning an IAM role to a pod, remove the {} above and uncomment this if needed
    # eks.amazonaws.com/role-arn: arn:aws:iam::123123123:role/role-name-goes-here

# This is absolutely necessary, this is how this service reads and updates PVCs and talks to Kubernetes
# Do not change any of this, or the volume autoscaler will not function any more
rbac:
  create: true
  clusterWideAccess: true
  rules:
    # This is how we listen and update PVCs
    - apiGroups: ['*']
      resources: ['persistentvolumeclaims']
      verbs:
        - list
        - patch
    # This is so we can send events into Kubernetes viewable in the event viewer
    - apiGroups: [""]
      resources:
        - events
      verbs:
        - create
        - patch
    # So we can to check StorageClasses for if they have AllowVolumeExpansion set to true
    - apiGroups: ["storage.k8s.io"]
      resources:
        - storageclasses
      verbs:
        - list

# This is to use new kubernetes standard labeling, should be true on both always
usingNewRecommendedLabels: true
labelsEnableDefault: true
usingMemoryKibiBytesEnvs: false

# We don't use a lot of features from the upstream chart, so disable them
podDistuptionBudget:
  enabled: false
ingress:
  enabled: false
ingress_secondary:
  enabled: false
autoscaling:
  enabled: false
security:
  runAsNonRoot: false
sidecar:
  enabled: false