Backups, disaster recovery, and going live
A failover group protects you from losing a site. It does not protect you from losing the cluster, the bucket, or last Tuesday. This unit adds backups, PITR, verification, restore and encryption at rest, then the alerting and runbooks that make playground operable at 3am, and finishes with a go-live gate you would actually sign. By the end you can hand playground to an on-call rotation and state precisely what it will and will not do for them.
Topics
Unit test
Quick check: can you say which site a backup ran from and why, what a Succeeded verification did not prove, which site in playground cannot have its keyring rotated right now, and why an alert on bloodraven_replication_lag_seconds might page you for a reader doing exactly what it was designed to do?
Project
The go-live pack for playground
Assemble the artefacts you would actually hand an on-call rotation: a Prometheus rules file built only from metric names the operator really exports, a one-page alert-to-runbook-to-first-command map, and a DR drill record showing you restored playground and measured how far back you could reach. Then let a checker prove the thing that matters — that your rules stay silent while a reader soaks past three times maxLagSeconds.