I created a test standby that was recovering from our prod primary via restored WAL files. When I broke recovery and opened up that test standby, I saw that records in pg_stat_all_tables for tables that contain many millions of rows (and have for years) were saying a few hundred rows, with last_analyze and last_autoanalyze being null. I know this isn't the case for pg_stat_all_tables on prod.
Do stats not automatically get preserved when a standby is promoted? Did something go wrong here or should I know/expect to have to manually gather DB stats after a failover (like we do with major upgrades)?
This is PG 9.2.22 on CentOS.
