How to Avoid Pitfalls in Schema Upgrade with Percona XtraDB Cluster
summary
Percona XtraDB Cluster enforces 100% synchronized for DML but DDL is non‑transactional so schema changes must replicate fully before next access and can take hours. The safest method TOI blocks writes while RSU (rolling) is advertised non‑blocking but can fail due to concurrent updates even on unrelated tables, triggers, or replication state. Expect concrete examples, a short checklist of checks (active writers, triggers, replica bootstrap) and failure modes to avoid.
AI-written summary · about
chapters (6)
- 0:00Percona XtraDB Cluster upgrades
- 5:07Schema modification pitfalls
- 10:58DDL in a cluster
- 17:29Failed upgrade recovery
- 25:03Safe schema changes
- 29:41Q&A
related talks
37mDatabase Schema Migrations with Zero Downtime
33mDatabase Schema Migrations with Zero Downtime
38mIs Percona XtraDB Cluster Right for Any Application
24mDatabases beyond the tutorials
59mNavigating the Challenges of CI /CD for Relational Database Schema Changes
47mMeet Percona XtraDB Cluster 5.7
27mStrategies for Reliable Database Schema Migrations
40mrails db:migrate:safely