Support often needs to analyse dumps of the config database. This usually involves running sh.status()/printShardingStatus() and passing in a reference to the mongorestored configDB, e.g. printShardingStatus(db.getSiblingDB("the_config_dump"))
However, the "Balancer" section (added in
SERVER-14882 to show balancer status, window, and recent activity) does not use the given configDB. This is because it has hard-coded references to the database named "config".
This means that if a database with a name other than "config" is passed in, then most of the output will relate to the correct db, but the balancer section will relate to the db named "config", which could have completely different details. Furthermore, if there is no db named "config" then printShardingStatus() will print mysterious warnings about config.settings and/or config.locks being missing, even though they are present in the given configDB. This is all very confusing and misleading when trying to analyse a config dump.