Skip to content

DatabaseServerPreset

DatabaseServerPreset is a cluster-scoped baseline configuration for DatabaseServer resources. You create it, or another tool creates it for you.

A preset holds one PostgreSQL shape: instance count, resources, storage, scheduling, monitoring, and the archive bucket. A platform team can publish a set of presets, for example small, standard, and large, and each team picks one. The PostgreSQL major version is not part of the shape. It comes from a CamundaRelease, so a version change never edits a preset.

A preset creates nothing and reports no status. A DatabaseServer uses it through spec.presetRef.

The smallest preset sets an instance count and a volume size:

apiVersion: core.camunda.io/v1
kind: DatabaseServerPreset
metadata:
  name: standard
spec:
  server:
    instances: 2
    storageSize: "64Gi"

graph LR
    DBS[DatabaseServer] -.->|presetRef| DBSP[DatabaseServerPreset]
    DBS -.->|releaseRef| CR[CamundaRelease]
Hold "Alt" / "Option" to enable pan & zoom

Merge rules

spec.server of the preset is the baseline. The CamundaRelease of releaseRef merges over it, and the server spec merges over both. A field set on the DatabaseServer replaces the value of the layer below for that field. A field left unset on the server comes from the layer below.

Every object and every map is replaced as a whole, never merged field by field or key by key. This includes resources, serviceAccount, scheduling, monitoring, archive, podLabels, and podAnnotations. A server that sets its own archive block drops the bucket and the retention of the preset with it. A server that sets one pod label drops every pod label of the preset.

An empty map (podLabels: {}, podAnnotations: {}) counts as unset, so it does not remove the map of the preset. To remove it, set the map that you want on the server. Or reference a preset without that map.

Fleet settings

A preset can set archive and platformConfigRef. One bucket then serves every server that references the preset, because each server writes its archive under a prefix of its own. One image repository serves them all as well.

Changes

An edit of a preset reaches every DatabaseServer that references it, with these exceptions:

  • A lower storageSize or walStorageSize does not shrink a running server. The server keeps its size and records a Warning event with reason StorageShrinkIgnored.
  • A cleared walStorageSize does not remove the write-ahead log volume of a running server. The server keeps the volume and records a Warning event with reason WALStorageKept.
  • An edit of archive waits while a server that uses the preset runs a rollback. That server reports Ready False with reason InvalidReference until the rollback is answered.

A new server uses the new baseline.

Deletion

Deleting a preset removes no server. Each DatabaseServer that references it reports Ready False with reason InvalidReference.

Status

A preset has no status. It reports no conditions and no status.observedGeneration. A problem with a preset shows on the Ready condition of the DatabaseServer that references it. Examples are a presetRef that names no preset, or a merge that lacks version or storageSize.

Spec reference

spec.server has the same type as the spec of DatabaseServer. The fields presetRef, releaseRef, databaseServerConfig, and suspend belong to one server and must stay unset in a preset. Validation rules has the rule for version. Every other field is inheritable.

Every field, with its type, whether it is required, and its default:

apiVersion: core.camunda.io/v1
kind: DatabaseServerPreset
metadata:
  name: standard
spec:
  # object (DatabaseServer spec type). Required. The baseline that servers inherit.
  server:
    # string. Optional. Name of a cluster-scoped CamundaPlatformConfig. Only its image settings are read.
    platformConfigRef: "my-platform-config"
    # integer. Optional, default: 1. Number of PostgreSQL instances, at least 1.
    instances: 2
    # object (corev1.ResourceRequirements). Optional. CPU and memory of each instance.
    resources:
      requests: { cpu: "1", memory: "2Gi" }
      limits: { memory: "2Gi" }
    # string (resource quantity). Optional. Size of the data volume of each instance.
    storageSize: "64Gi"
    # string. Optional, default: the default StorageClass of the Kubernetes cluster. StorageClass of the volumes.
    storageClassName: "ssd"
    # string (resource quantity). Optional. Size of a separate volume for the write-ahead log.
    walStorageSize: "8Gi"
    # object. Optional. The ServiceAccount <name>-postgres of the instance pods, which the operator creates. See DatabaseServer.
    serviceAccount:
      # map[string]string. Optional. Annotations for workload identity.
      annotations: {}
    # object. Optional. Scheduling constraints. A server that sets its own scheduling block replaces this one as a whole.
    scheduling:
      # object (corev1.NodeAffinity). Optional. Node affinity rules.
      nodeAffinity: {}
      # object (corev1.PodAffinity). Optional. Pod affinity rules.
      podAffinity: {}
      # list (corev1.Toleration). Optional. Tolerations of the pods.
      tolerations: []
    # map[string]string. Optional. Extra labels on the instance pods.
    podLabels: {}
    # map[string]string. Optional. Extra annotations on the instance pods.
    podAnnotations: {}
    # object. Optional. Prometheus scraping. A server that sets its own monitoring block replaces this one as a whole.
    monitoring:
      podMonitor:
        # boolean. Optional, default: false. Creates a PodMonitor over the instance pods.
        enabled: false
        # map[string]string. Optional. Extra labels on the PodMonitor.
        labels: {}
        # map[string]string. Optional. Extra annotations on the PodMonitor.
        annotations: {}
        # string. Optional, default: the Prometheus setting. Scrape interval, as a Prometheus duration.
        interval: "30s"
    # object. Optional. The continuous archive. See DatabaseServer.
    archive:
      # string. Required in this block. Name of an ObjectStorageConfig in the namespace of each server.
      objectStorageRef: my-backup-bucket
      # integer. Required in this block. How many days into the past a restore can reach, from 1 to 36500.
      retentionPeriodDays: 30
      # string. Optional, default: "0 0 2 * * *". Six-field cron in UTC, seconds first, or a descriptor such as "@daily". A five-field cron is rejected.
      baseBackupSchedule: "0 0 2 * * *"

Validation rules

  • spec.server must not set presetRef, releaseRef, databaseServerConfig, or suspend. An empty presetRef, an empty releaseRef, and suspend: false count as unset, so templated YAML that renders zero values still applies. An empty databaseServerConfig is rejected by the name pattern. Omit the field instead.
  • version is rejected in spec.server with version belongs to a CamundaRelease and must not be set in a preset. Move the version to a CamundaRelease, and point every server at it with releaseRef. An empty version is rejected by the bare-major pattern. Omit the field instead.
  • The no-shrink rule of DatabaseServer for storageSize and walStorageSize does not bind a preset. You can lower or clear them at any time. See Changes for the effect on a running server.
  • The DatabaseServer checks that the merged configuration is complete, not the preset.
  • Every other rule of the DatabaseServer schema applies to spec.server. instances must be at least 1, and archive.retentionPeriodDays must be from 1 to 36500. archive.baseBackupSchedule must be a six-field cron or a descriptor, as on the DatabaseServer.

A production-shaped example

apiVersion: core.camunda.io/v1
kind: DatabaseServerPreset
metadata:
  name: large
spec:
  server:
    platformConfigRef: my-platform-config
    instances: 3
    resources:
      requests: { cpu: "4", memory: "8Gi" }
      limits: { memory: "8Gi" }
    storageSize: "512Gi"
    storageClassName: "ssd"
    walStorageSize: "64Gi"
    scheduling:
      tolerations:
        - key: dedicated
          operator: Equal
          value: postgres
          effect: NoSchedule
    monitoring:
      podMonitor:
        enabled: true
        labels:
          release: prometheus
    archive:
      objectStorageRef: my-backup-bucket
      retentionPeriodDays: 30