The Elephantine Upgrade
summary
OVHcloud migrates 230 databases across 50+ clusters from PostgreSQL 9.6 to 14 using Ansible + AWX to orchestrate automated playbooks and a one-click promotion that moves a VIP and swaps load balancers for minimal downtime. Migration uses pg_logical logical replication for most clusters, with pg_dump/pg_restore or pg_upgrade for corner cases. Teams solve enum/oid and charset issues, reduce replication load with preloads, and work around missing Debian packages. Result is clustered automation, upgraded OS/hardware, and plans to adopt built-in logical replication.
AI-written summary · about
chapters (20)
- 0:00Introduction and context
- 6:12Motivation for the upgrade
- 9:48Upgrade methods compared
- 15:49Logical replication setup
- 18:02Automation with Ansible and AWX
- 21:14Dynamic inventory and Bastion
- 22:21Day-of migration steps
- 27:22Optimizing the process
- 29:06Alternative: PG upgrade method
- 30:33Unexpected type changes
- 32:20Compatibility challenges
- 34:30Internal software adaptations
- 36:03Handling schema constraints
- 37:52Identifier limitations
- 38:57Load impact of logical replication
- 40:28Dynamic table pitfalls
- 41:57Legacy charset conversion
- 43:04Future plans and mindset shift
- 45:18Team and hiring acknowledgment
- 46:24Q&A session
related talks
32mPostgreSQL Major Upgrades: Swapping the Elephant Without Breaking the Room
26mMastering Large-Scale PostgreSQL Migration
46mBig databases, many servers, data center or cloud - it's all here!
49mPractical Guide to Migrating +100 Large PostgreSQL to the Cloud
01h 01mAlways Online or Bust
29mTales from the Field - Oracle to PostgreSQL migrations
59mUsing Postgres XL
22mImproving the efficiency of more than 800 databases with o11y