Externalized Connection Stores
Decouple broker configurations from application containers with enterprise storage backends: PostgreSQL JDBC, Kubernetes Custom Resource Definitions, and encrypted file stores.
High Availability & Multi-Replica (REQ-101): By externalizing connection definitions, multiple EventOre pods run completely stateless behind an ingress or load balancer without configuration drift or manual file synchronization.
Supported Storage Backends
EventOre provides pluggable implementations of ConnectionStore:
- PostgreSQL JDBC Store: Enterprise-grade database storage using Flyway migrations, connection pooling (HikariCP), and optimistic concurrency control (
versioncolumn) to prevent concurrent write collisions across instances. - Kubernetes CRD Store: GitOps-native configuration where brokers are declared as
EventoreConnectioncustom resources managed by ArgoCD, Flux, or Helm. - File Store (Encrypted): Default local or persistent volume JSON/YAML store with AES-256 GCM credential encryption at rest.
Configuring PostgreSQL JDBC Store
In your application.yml or Helm values, select the jdbc store type:
eventore:
connection-store:
type: jdbc
jdbc:
url: jdbc:postgresql://postgres.eventore.svc.cluster.local:5432/eventoredb
username: ${DB_USER}
password: ${DB_PASSWORD}
pool-size: 10
auto-migrate: true
Kubernetes Custom Resource (CRD) Pattern
Deploy broker connections declaratively alongside your microservices:
apiVersion: eventore.io/v1alpha1
kind: EventoreConnection
metadata:
name: prod-kafka-cluster
namespace: eventore
spec:
protocol: KAFKA
brokerUrl: "b-1.prod.kafka.us-east-1.amazonaws.com:9096"
security:
protocol: SASL_SSL
saslMechanism: AWS_MSK_IAM
tags:
environment: "production"
team: "data-platform"
Live Store Migration
Upgrading from a single-node file store to clustered PostgreSQL JDBC can be performed without downtime using EventOre's live migration endpoint:
POST /api/v1/admin/connection-store/migrate
Content-Type: application/json
{
"sourceType": "file",
"targetType": "jdbc",
"dryRun": false
}