In Staging, it happened that executing too many background jobs caused auto-scaled Zync pods to create too many DB connections.

We need to figure out why, because in theory they should be limited by a connection pool