Configuration Reference - nnstd/seaweedfs-operator GitHub Wiki

This page provides a comprehensive reference for all configuration options available in the SeaweedFS operator.

Seaweed Resource Configuration

Top-Level Configuration

Field Type Default Description
image string - Container image for SeaweedFS
version string - SeaweedFS version
metrics *MetricsSpec - Metrics configuration
master *MasterSpec - Master server configuration
volume *VolumeSpec - Volume server configuration
filer *FilerSpec - Filer server configuration
filerBackup *FilerBackupSpec - Filer backup configuration
admin *AdminSpec - Admin UI configuration
schedulerName string - Kubernetes scheduler name
pvReclaimPolicy *corev1.PersistentVolumeReclaimPolicy - PV reclaim policy
imagePullPolicy corev1.PullPolicy - Image pull policy
imagePullSecrets []corev1.LocalObjectReference - Image pull secrets
enablePVReclaim *bool - Enable PVC reclaim
hostNetwork *bool - Enable host networking
affinity *corev1.Affinity - Pod affinity rules
nodeSelector map[string]string - Node selector
annotations map[string]string - Pod annotations
tolerations []corev1.Toleration - Pod tolerations
statefulSetUpdateStrategy appsv1.StatefulSetUpdateStrategyType - StatefulSet update strategy
storage *StorageSpec - Storage configuration
hostSuffix *string - Host suffix for ingresses

MetricsSpec

Field Type Default Description
enabled bool false Enable metrics collection
metricsPort *int32 5555 Port for Prometheus metrics

StorageSpec

Field Type Default Description
storageClassName *string - Storage class for all components
volumeServerDiskCount int32 - Number of disks per volume server

MasterSpec

Field Type Default Description
replicas int32 1 Number of master replicas (minimum 1)
service *ServiceSpec - Service configuration
config *string - Raw TOML configuration
metricsPort *int32 - Prometheus metrics port
volumePreallocate *bool - Enable volume preallocation
volumeSizeLimitMB *int32 - Volume size limit in MB
garbageThreshold *string - Garbage collection threshold
pulseSeconds *int32 - Pulse interval in seconds
defaultReplication *string - Default replication setting
concurrentStart *bool - Enable concurrent startup

VolumeSpec

Field Type Default Description
replicas int32 1 Number of volume replicas (minimum 1)
service *ServiceSpec - Service configuration
storageClassName *string - Storage class for volumes
metricsPort *int32 - Prometheus metrics port
compactionMBps *int32 - Compaction speed in MB/s
fileSizeLimitMB *int32 - File size limit in MB
fixJpgOrientation *bool - Fix JPEG orientation
idleTimeout *int32 - Idle timeout in seconds
maxVolumeCounts *int32 - Maximum volume counts
minFreeSpacePercent *int32 - Minimum free space percentage

FilerSpec

Field Type Default Description
replicas int32 1 Number of filer replicas (minimum 1)
service *ServiceSpec - Service configuration
config *string - Raw TOML configuration
metricsPort *int32 - Prometheus metrics port
persistence *PersistenceSpec - Persistence configuration
maxMB *int32 - Maximum file size in MB
s3 bool true Enable S3 compatibility

AdminSpec

Field Type Default Description
replicas int32 1 Number of admin replicas (minimum 1)
service *ServiceSpec - Service configuration
port *int32 23646 Admin server port
masters string - Comma-separated master servers
dataDir string - Data directory
adminUser string "admin" Admin username
adminPassword string - Admin password
adminPasswordSecretRef *AdminPasswordSecretRef - Admin password secret reference
persistence *PersistenceSpec - Persistence configuration
tls *AdminTLSSpec - TLS configuration

ServiceSpec

Field Type Default Description
type corev1.ServiceType - Service type
annotations map[string]string - Service annotations
loadBalancerIP *string - Load balancer IP
clusterIP *string - Cluster IP

PersistenceSpec

Field Type Default Description
enabled bool false Enable persistence
existingClaim *string - Use existing PVC
mountPath *string "/data" Volume mount path
subPath *string "" Subdirectory to mount
accessModes []corev1.PersistentVolumeAccessMode ["ReadWriteOnce"] Volume access modes
selector *metav1.LabelSelector - Volume selector
resources corev1.VolumeResourceRequirements {requests:{storage:"4Gi"}} Volume resource requirements
volumeName string - Volume name
storageClassName *string - Storage class
volumeMode *corev1.PersistentVolumeMode - Volume mode
dataSource *corev1.TypedLocalObjectReference - Data source

BucketClaim Resource Configuration

BucketClaimSpec

Field Type Required Default Description
bucketName string Yes - Name of the bucket to be created
clusterRef ClusterReference Yes - Reference to the Seaweed cluster
replication string No "000" Replication setting for the bucket
collection string No - Collection setting for the bucket
ttlSec *int32 No - TTL in seconds for objects in the bucket
readOnly bool No false Whether the bucket should be read-only

ClusterReference

Field Type Required Description
name string Yes Name of the Seaweed cluster
namespace string No Namespace of the Seaweed cluster (defaults to BucketClaim namespace)

Component Configuration Inheritance

The SeaweedFS operator uses a hierarchical configuration system where component-level settings can override cluster-level settings:

ComponentSpec (Base for all components)

Field Type Description
version *string Override cluster-level version
imagePullPolicy *corev1.PullPolicy Override cluster-level imagePullPolicy
imagePullSecrets []corev1.LocalObjectReference Override cluster-level imagePullSecrets
hostNetwork *bool Override cluster-level hostNetwork
affinity *corev1.Affinity Override cluster-level affinity
priorityClassName *string Component-specific priority class
schedulerName *string Override cluster-level schedulerName
nodeSelector map[string]string Merge with cluster-level nodeSelector
annotations map[string]string Merge with cluster-level annotations
tolerations []corev1.Toleration Override cluster-level tolerations
env []corev1.EnvVar Additional environment variables
terminationGracePeriodSeconds *int64 Pod termination grace period
statefulSetUpdateStrategy appsv1.StatefulSetUpdateStrategyType Override cluster-level strategy

Resource Requirements

All components support Kubernetes resource requirements:

spec:
  master:
    resources:
      requests:
        memory: "256Mi"
        cpu: "250m"
      limits:
        memory: "512Mi"
        cpu: "500m"

Environment Variables

You can add custom environment variables to any component:

spec:
  filer:
    env:
    - name: CUSTOM_VAR
      value: "custom-value"
    - name: SECRET_VAR
      valueFrom:
        secretKeyRef:
          name: my-secret
          key: secret-key

Node Selectors

Place pods on specific nodes:

spec:
  nodeSelector:
    storage: ssd
    zone: us-west-1a

Tolerations

Allow pods to be scheduled on tainted nodes:

spec:
  tolerations:
  - key: "dedicated"
    operator: "Equal"
    value: "seaweedfs"
    effect: "NoSchedule"

Storage Classes

Configure storage for different components:

spec:
  storage:
    storageClassName: "fast-ssd"
  volume:
    storageClassName: "high-iops"  # Override for volume servers

TLS Configuration

Configure TLS for the admin UI:

spec:
  admin:
    tls:
      enabled: true
      certificateSecretRef:
        name: admin-tls-secret
        mapping:
          cert: tls.crt
          key: tls.key
          ca: ca.crt

Backup Configuration

Configure backup destinations:

spec:
  filerBackup:
    sink:
      s3:
        enabled: true
        region: "us-west-2"
        bucket: "seaweedfs-backup"
        directory: "backups"
        awsCredentialsSecretRef:
          name: aws-credentials
          mapping:
            awsAccessKeyID: access-key
            awsSecretAccessKey: secret-key

Validation Rules

Required Fields

  • BucketClaimSpec.bucketName: Must be a valid bucket name
  • BucketClaimSpec.clusterRef.name: Must reference an existing Seaweed cluster
  • SeaweedSpec.master.replicas: Must be at least 1
  • SeaweedSpec.volume.replicas: Must be at least 1
  • SeaweedSpec.filer.replicas: Must be at least 1

Default Values

  • BucketClaimSpec.replication: "000" (no replication)
  • BucketClaimSpec.readOnly: false
  • AdminSpec.adminUser: "admin"
  • AdminSpec.port: 23646
  • MetricsSpec.metricsPort: 5555
  • PersistenceSpec.mountPath: "/data"
  • PersistenceSpec.accessModes: ["ReadWriteOnce"]
  • PersistenceSpec.resources.requests.storage: "4Gi"

Best Practices

Resource Sizing

  • Start with small resource requests and adjust based on usage
  • Monitor resource usage and scale accordingly
  • Use resource limits to prevent resource exhaustion

Storage Configuration

  • Use appropriate storage classes for your workload
  • Consider using fast storage for volume servers
  • Use persistent storage for filer and admin components

Security

  • Use secrets for sensitive configuration
  • Enable TLS for admin UI in production
  • Configure appropriate RBAC permissions

Monitoring

  • Enable metrics for all components
  • Set up alerts for resource usage
  • Monitor cluster health and performance

Related Resources