DBConnection.Holder.checkout deadlock?

tim2CF · June 21, 2019, 10:18am

Hello! Need help from Ecto experts!
I have a problem. I’m trying to test some cluster features of my app with ex_unit_clustered_case package. Local cluster starts successfully, but when I’m trying to call some Ecto Repo functions in slave node, Repo always returns error

%DBConnection.ConnectionError{
    message: "connection not available and request was dropped from queue after 1333ms. You can configure how long requests wait in the queue using :queue_target and :queue_interval. See DBConnection.start_link/2 for more information"
  }

I started to investigate what is happening on cluster nodes at this moment using observer tool and found interesting processes on top:

Can it be some deadlock in this receive expression?

github.com

elixir-ecto/db_connection/blob/9d2416a4b0e9f9f465e639e090027e9acc7d9d77/lib/db_connection/holder.ex#L237


      
          
              nil ->
                {:exit, :noproc}
            end
          end
          
          defp checkout_call(pid, callers, queue?, start, timeout) do
            lock = Process.monitor(pid)
            send(pid, {:db_connection, {self(), lock}, {:checkout, callers, start, queue?}})
          
            receive do
              {:"ETS-TRANSFER", holder, pool, {^lock, ref}} ->
                Process.demonitor(lock, [:flush])
                {deadline, ops} = start_deadline(timeout, pool, ref, holder, start)
                :ets.update_element(holder, :conn, [{conn(:lock) + 1, lock} | ops])
          
                pool_ref =
                  pool_ref(pool: pool, reference: ref, deadline: deadline, holder: holder, lock: lock)
          
                checkout_result(holder, pool_ref)

Maybe you have an idea how to fix it?