Synchronous Replication

A walkthrough of running OpenBao on Kubernetes with CloudNativePG as its PostgreSQL storage backend. Every layer of this stack is open source, with no vendor lock-in: Kubernetes and CloudNativePG are both CNCF projects, authenticated entirely over TLS client certificates via the 1.30 DatabaseRole CRD, with no passwords anywhere in the stack. Co-authored with Rob Kenefeck from ControlPlane.

CloudNativePG 1.24 introduces a highly customisable approach to managing PostgreSQL synchronous replication through the new .spec.postgresql.synchronous stanza. In this article, I’ll guide you through configuring synchronous replication within a single Kubernetes cluster and across multiple clusters. I’ll explore quorum-based and priority-based replication methods, highlighting their benefits and trade-offs. Additionally, I’ll explain how to adjust the synchronous_commit setting to strike the right balance between data durability and performance.