War Stories: Fighting the PostreSQL Support Trenches
summary
Twenty years of PostreSQL support yields concrete failures and fixes: flying to Kiribati for a live system, getting held two and a half hours at United Nations Vienna security, a replica that fills the disk because someone ignored size checks, an initdb followed by create database that destroys data, a sequence gap that nearly ruins a company over invoices, rescuing 1.4 million life insurance records by fixing corrupted data files, and refusing impossible requests like similarity search in Chinese or customers who expect Oracle-priced hardware for six rows.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 1:06PostgreSQL support company
- 2:31Traveling to remote clients
- 5:42PostgreSQL training needs
- 6:49Security at UN Vienna
- 7:30Database migration struggles
- 8:54Replica setup disk space
- 10:38Fresh init file missing
- 12:47Chinese similarity search
- 14:13Character set mismatch errors
- 15:37High availability misunderstandings
- 17:23Fax support request
- 18:49Performance with minimal data
- 22:00Data models and WHERE clauses
- 23:45Index phobia in banking
- 24:27Semantic query errors
- 25:51Tax office invoicing gaps
- 29:04Life insurance database disaster
- 31:39Closing remarks
related talks
42mPostgres: stories that will scare you
36mTales from the Postgres Front, and What We Can Learn
44mConfessions of a DBA: worst and best things I've done in production
19mPostgreSQL Challenges
39mData Corruption: Stop the Evil Tribbles
54mLessons from two decades of hacking the proprietary value into open so
21mSQL Horror Stories- 21mSQL Horror Stories