Hey guys, we just updated our cluster to HA and one of our collections is not joining correctly (it is joining documents that it shouldn't) Is there a way to fix that with zero downtime?