How to investigate DB errors like "client timed out because it queued and checked out the connection for longer than..."

Questions / Help
,
14

Thanks to the discussion in this thread, I ended up doing something like this:

  def handle_event(
        [:my_app, :repo :query],
        _measurements,
        %{result: {:error, error}, query: query},
        _config
      ) do
    # Log the query
  end

That will capture more than just timeouts. If that’s a problem then you can go through the additional work to match on error or add some other filtering so that you only get timeouts. In my experience testing locally I saw a few different errors associated with timeouts but in production so far we’ve only seen this one:

Postgrex.Protocol (#PID<0.18044.0>) disconnected: ** (DBConnection.ConnectionError) client #PID<0.18926.0> timed out because it queued and checked out the connection for longer than 5ms