DoltDB Operator

DoltDB brings Git-style versioning to your database — branches, diffs, merges, and a complete history of every change. Running it on Kubernetes, however, means managing StatefulSets, replication topology, failover, backups, and configuration by hand.

This operator takes care of all of that. Define your DoltDB cluster as a Kubernetes Custom Resource, and the operator handles provisioning, replication, automatic failover, rolling upgrades, and backups — so you can focus on your data instead of your infrastructure.

Features

• Provision DoltDB clusters with a single Custom Resource
• Automated primary failover with replication-aware candidate selection
• Single-instance and multi-replica modes with seamless scaling transitions
• Scheduled backups via VolumeSnapshots
• Zero-downtime rolling updates for data-plane upgrades
• Declarative SQL resource management: users, grants, and logical databases
• Install via Helm charts

Quick Start

Prerequisites

• Kubernetes 1.28+
• Helm 3.x
• A StorageClass that supports dynamic provisioning
• (Recommended) A service mesh with mTLS (e.g., Istio) for network encryption

Install with Helm

# Add the CRDs chart (install separately for lifecycle management)
helm install doltdb-operator-crds oci://ghcr.io/electronicarts/doltdb-operator-crds

# Install the operator
helm install doltdb-operator oci://ghcr.io/electronicarts/doltdb-operator

Deploy a DoltDB Cluster

apiVersion: k8s.dolthub.com/v1alpha
kind: DoltDB
metadata:
  name: my-dolt
spec:
  engineVersion: "1.57.2"
  image: dolthub/dolt
  replicas: 2
  storage:
    size: 10Gi
  server:
    listener:
      host: 0.0.0.0
      port: 3306
      maxConnections: 2048
    cluster:
      remotesAPI:
        port: 50051
  globalConfig:
    commitAuthor:
      name: "dolt operator"
      email: "dolt@operator.local"
  replication:
    enabled: true
    primary:
      podIndex: 0
      automaticFailover: true

kubectl apply -f my-dolt.yaml

Connect to DoltDB

# Port-forward the primary service
kubectl port-forward svc/my-dolt-primary 3306:3306

# Connect with any MySQL client
mysql -h 127.0.0.1 -P 3306 -u root

Custom Resources

Roadmap

• Backup-aware switchover — prevent primary switchover while a backup is running to avoid unnecessary retries
• Validating and mutating webhooks — webhook certificate management for CRD admission control
• Remote replication — support for cross-cluster replication via Dolt remotes
• Backup restore — declarative restore from S3/DoltHub backups via a BackupRestore CRD

Architecture

The operator follows the standard Kubernetes controller pattern built with kubebuilder. The main DoltDBReconciler delegates to specialized sub-reconcilers:

DoltDBReconciler
  |-- RBACReconciler        (ServiceAccount, Role, RoleBinding)
  |-- ConfigMapReconciler   (DoltDB server configuration)
  |-- ServiceReconciler     (primary, reader, internal headless services)
  |-- StatefulSetReconciler (pods, rolling updates)
  |-- StorageReconciler     (PVC management)
  |-- StatusReconciler      (condition updates)
  +-- ReplicationReconciler (primary election, failover, switchover)

SQL resources (Database, User, Grant) connect directly to DoltDB instances to execute DDL statements, with reconciliation tied to the parent DoltDB cluster health.

Helm Chart Configuration

See charts/doltdb-operator/values.yaml for all available values.

Key options:

CRDs Chart

CRDs are distributed as a separate Helm chart (doltdb-operator-crds) for independent lifecycle management. This allows upgrading CRDs separately from the operator.

Network Encryption

The operator does not implement application-level TLS for DoltDB connections. It is designed to run in environments with network-level encryption:

• Istio with strict mTLS (recommended)
• Linkerd, Cilium, or other service mesh / CNI with encryption

Health probes use --no-tls because they connect via loopback (127.0.0.1) within the same pod.

See SECURITY.md for more details.

Development

See CONTRIBUTING.md for development setup, testing, and contribution guidelines.

Quick commands:

make build          # Build the operator binary
make test           # Run unit tests
make lint-fix       # Lint and auto-fix
make cluster        # Create a kind cluster
make tiltci         # Run integration tests

License

BSD 3-Clause License - Copyright (c) 2025 Electronic Arts Inc.

Contributing

Before you can contribute, EA must have a Contributor License Agreement (CLA) on file that has been signed by each contributor. You can sign here: Go to CLA

See CONTRIBUTING.md for detailed guidelines.