Elephants on Automatic: HA Clustered PostgreSQL with Helm [I]
summary
Demonstrates automated HA PostgreSQL on Kubernetes using Patroni + Helm chart and a CitusDB sharded cluster (1 query + 6 shards), deploying statefulsets with persistent volumes and service endpoints. Shows a 3‑replica Patroni cluster where killing the master triggers immediate failover and client transactions continue, and a seven‑node Citus cluster used to run 1.4M Wikipedia edits tests. Helm parameterizes names, storage and credentials so clusters deploy in minutes.
AI-written summary · about
chapters (8)
- 0:00Start
- 1:301.5: StatefulSet 1. storage 2. Identity 3. discovery
- 2:47Custom Schedulers? • not packaged with app • kube expertise needed
- 5:16parameterization • deploy multiple instances • app registry • support for sidecars
- 5:57Patroni • HA PostgreSQL controller • Etod, Zookeeper or Consul
- 15:59More than autofailover • synchronous mode • Cascading replication • dynamic configuration
- 17:47Sharded Parallel Postgres • Sharded Postgres • partition data over nodes • shards duplicate for redundancy · query node distributos queries • mainly for data mining big data
- 18:40StatefulSet Setup • one query node (node) • 2 to N shards • service for query node only
related talks
54mContainerized Clustered PostgreSQL
33mCitus & Patroni: The Key to Scalable and Fault-Tolerant PostgreSQL
50mPatroni Deployment Patterns
30mUSING KUBERNETES/DOCKER/HELM TO DEPLOY ON-DEMAND POSTGRESQL STREAMING REPLICAS
50mThe Immortal Postgres Web
41mEnsuring High Availability in Self-Managed PostgreSQL with Patroni
44mPostgreSQL on PetSet
32mContaining an Elephant