At a $job, I took down prod once and restored it twice after the database got emptied. Well, we didn't quite empty it - I imported an old dump into prod instead of a test database while trying to replicate an issue. My coworker, on the other hand, was running some script he had written to truncate a few tables... and ended up running TRUNCATE individually on every single table in the database.
They've been good tests of our backup systems, actually. In fact, one of the incidents revealed that our on-site backups had been broken for ~a couple of weeks. (Our off-sites use a different backup system and were fine, but we restored that gap from binlog instead, as they were still present locally and it was faster than the 100Mbps upload from our off-site...)
They've been good tests of our backup systems, actually. In fact, one of the incidents revealed that our on-site backups had been broken for ~a couple of weeks. (Our off-sites use a different backup system and were fine, but we restored that gap from binlog instead, as they were still present locally and it was faster than the 100Mbps upload from our off-site...)
Each one was 15-20 minutes of downtime.