Patroni: what the blog posts don't tell you...
summary
Shows concrete hardening and operations details: how to enable TLS and RBAC on etcd and protect Patroni’s REST API, why etcd needs odd quorum sizes (3 or 5) and must be on separate hosts, and why to use per-cluster etcd users not a global root. Covers backups (etcd snapshots, Barman/pgbackrest), controlled minor and major upgrade procedures (one node at a time, remove systemid in DCS for major upgrades), failover testing, proxy tradeoffs, and turning DCS failsafe mode on for survivability — all proven in long‑running production.
AI-written summary · about
chapters (14)
- 0:00Introduction and setup
- 3:56What is Patroni and why use it
- 9:11Setting up a basic Patroni cluster
- 11:02Securing ETCD with TLS and auth
- 15:57Securing Patroni REST API
- 18:04DCS failsafe mode and config placement
- 21:42Running ETCD and Patroni separately
- 22:43Multiple Patroni clusters with one ETCD
- 24:28Proxies and client connections
- 24:51Backups and disaster recovery
- 27:16Timelines and backup testing
- 28:03Upgrading ETCD and PostgreSQL
- 33:40Major PostgreSQL upgrades
- 34:10Q&A session
related talks
50mPatroni Deployment Patterns
22mWhat is Patroni, really?
41mProtecting your data with Patroni and pgBackRest
52mPatroni 3.0: What's New and Future Plans
19mDr etcd; or; How I Learned to Stop Worrying and Love the Datastore - Nick Young, VMware
20mLessons Learned from Operating ETCD
54mContainerized Clustered PostgreSQL
41mEnsuring High Availability in Self-Managed PostgreSQL with Patroni