New to KubeDB? Please start here.
Neo4jAutoscaler
What is Neo4jAutoscaler?
Neo4jAutoscaler is a Kubernetes custom resource that declares how KubeDB should automatically scale the compute resources and persistent storage of a Neo4j cluster. The Autoscaler operator translates its recommendations into Neo4jOpsRequest resources, which are executed by Ops Manager.
The following example enables both compute and storage autoscaling:
apiVersion: autoscaling.kubedb.com/v1alpha1
kind: Neo4jAutoscaler
metadata:
name: neo4j-autoscaler
namespace: demo
spec:
databaseRef:
name: neo4j-autoscale
opsRequestOptions:
apply: IfReady
timeout: 10m
maxRetries: 3
compute:
neo4j:
trigger: "On"
podLifeTimeThreshold: 5m
resourceDiffPercentage: 20
minAllowed:
cpu: 600m
memory: 1200Mi
maxAllowed:
cpu: "2"
memory: 2Gi
controlledResources:
- cpu
- memory
containerControlledValues: RequestsAndLimits
storage:
neo4j:
trigger: "On"
usageThreshold: 80
scalingThreshold: 50
expansionMode: Online
Specification
Like other Kubernetes resources, Neo4jAutoscaler contains apiVersion, kind, metadata, spec, and status. Users declare the policy in spec; the operator reports observed state in status.
spec.databaseRef
spec.databaseRef is required and identifies the Neo4j resource in the same namespace.
| Field | Description |
|---|---|
name | Name of the target KubeDB Neo4j resource. |
spec.opsRequestOptions
These options are copied to every Neo4jOpsRequest created by the Autoscaler.
| Field | Description |
|---|---|
apply | IfReady creates operations only while the database is ready; Always permits creation regardless of readiness. Defaults to IfReady. |
timeout | Maximum duration allowed for each operation step. |
maxRetries | Maximum retries for a failed operation. Defaults to 1. |
spec.compute
spec.compute.neo4j controls CPU and memory recommendations for the neo4j container.
| Field | Description |
|---|---|
trigger | Enables autoscaling when set to On; use Off to disable it without deleting the resource. |
minAllowed | Lower CPU and memory bounds for recommendations. |
maxAllowed | Upper CPU and memory bounds for recommendations. |
controlledResources | Resources controlled by the Autoscaler, normally cpu and memory. |
containerControlledValues | RequestsAndLimits updates both values; RequestsOnly updates only requests. |
resourceDiffPercentage | Minimum percentage difference between the current allocation and a recommendation before an update is applied. Defaults to 50. |
podLifeTimeThreshold | Minimum pod lifetime considered when deciding whether to apply a recommendation. Defaults to 15m. |
spec.compute.nodeTopology is optional. When set, the Autoscaler selects resources from the named NodeTopology instead of applying an arbitrary recommendation. scaleUpDiffPercentage and scaleDownDiffPercentage control when it moves between topology entries; their defaults are 15 and 25, respectively.
spec.storage
spec.storage.neo4j controls expansion of the Neo4j data volumes.
| Field | Description |
|---|---|
trigger | Enables storage autoscaling when set to On. |
usageThreshold | Used-capacity percentage at which expansion is triggered. Defaults to 80. |
scalingThreshold | Percentage by which the current volume is increased. Defaults to 50. |
scalingRules | Optional size-dependent rules. Each rule has an appliesUpto capacity and a threshold percentage or absolute quantity. |
upperBound | Optional maximum volume size. |
expansionMode | Required expansion strategy: Online or Offline. |
For example, the following rules grow smaller volumes proportionally and larger volumes by a fixed amount:
storage:
neo4j:
trigger: "On"
usageThreshold: 80
expansionMode: Online
upperBound: 2Ti
scalingRules:
- appliesUpto: 500Gi
threshold: 30pc
- appliesUpto: 1Ti
threshold: 20pc
- appliesUpto: ""
threshold: 100Gi
status
The status is managed by KubeDB and should not be edited. Important fields include:
| Field | Description |
|---|---|
phase | Current Autoscaler phase, such as InProgress, Current, or Failed. |
observedGeneration | Most recent resource generation processed by the operator. |
conditions | Events and outcomes reported by the Autoscaler controller. |
vpas | Current compute recommendations and their conditions. |
checkpoints | Historical CPU and memory samples used by the recommender. |