https://coreos.slack.com/archives/C0246P60U8H/p1667339454972789

I checked on stage and there are ~88 source_rhc_connections that don't have an associated source..so it seems like theres a bug where the source gets deleted, but not the join record.

EDIT:
turns out the issue was just related to how we send update messages during availability status checks - since we were loading the rhc record directly rather than through the DAO. By setting the field correctly everything seems to be good.