DC
That’s exactly the failure the archive section is meant to catch: everything reports green while nothing is actually landing. Finding it in a restore test instead of an incident is the best possible version of that story. It’s also why the post doesn’t treat a zero exit code as proof. It alerts when pg_stat_archiver shows last_failed_time newer than last_archived_time, and the scheduled restore test checks real data, not just that the server starts. And yes, the restore you practice is the restore you get. Thanks for sharing it.