Kube-native Postgres [I]
summary
Patroni moves PostgreSQL to be Kubernetes-native by dropping etcd and using the Kubernetes API for leader election, storing leadership in ConfigMap+Endpoint, and running a 3-pod StatefulSet where one pod becomes master and others replicate; failover rebinds the master endpoint and replicas remaster automatically. There’s an alpha Spilo operator (manifest-driven) and a Crunchy alternative (CLI-driven). Production deployments exist, PGRewind support is present, and default is async replication unless you enable synchronous/quorum.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:00Goal
- 3:33Overview
- 5:33Goals
- 12:25Failover
- 19:56Environment variables
- 20:32Operators
- 20:52Why use an operator
- 21:59Stateful deployment
- 22:13Operator
- 22:41Config Map
- 28:52Comparison
- 29:56Future plans
- 30:49Prometheus integration
- 31:44Limitations
- 33:07Questions
related talks
34mRun PostgreSQL:The Kubernetes way
26mKubernetes from the Database Out
48mPostgres and Kubernetes: Past, Present and Future
37mData On Kubernetes, Deploying And Running PostgreSQL And Patterns For Databases In a Kubernetes Cluster.
50mDemystifying Kubernetes for Postgres DBAs; A Guide to Operators
48mPostgreSQL in Kubernetes: Operation - Day 2
28mLightning Talk: Cloud Native PostgreSQL - Running PostgreSQL on Kubernetes
54mContainerized Clustered PostgreSQL