Ecto with the Postgrex adapter allow us to run a callback right after connect, with which we can call a function where the first argument will be:

%DBConnection{
   conn_mod: Postgrex.Protocol,
   conn_ref: #Reference<0.384196318.1471152129.192608>,
   pool_mod: DBConnection.Task,
   pool_ref: {#PID<0.734.0>, #Reference<0.384196318.1471152129.192606>}
}

The PID in pool_ref refers to the checked out connection in the pool. This is only called after a successful connection which is fine. Now when the back-end postgrex adapter runs into a problem (say I just plug off the database), I keep getting flooded with [error] Postgrex.Protocol (#PID<0.737.0>) disconnected: ** (Postgrex.Error) FATAL 57P01 (admin_shutdown): terminating connection due to administrator command, in which the PID refers to the checked out connections in the pool.

But because both the Repo.Pool and Repo.Supervisor trap exits I can’t seem to find a way of being notified of these failures, since the pool connections never actually exit. The recovery on the side of Ecto/PGex works just fine, but I would like to be able to do some stuff if all of a sudden all the pool connections are error’ing out. Is there any way of getting those errors to bubble up somehow or any other way you handle this?

Thanks