Diagnosing unreliable join-accept reception

Okay great. What more can I do to diagnose why the Conduit running MP Packet Forwarder is often not transmitting the join-accept over the air?

I note in the logs above that 2019-03-31 10:35:33 10:35:32 WARNING: a downlink was already scheduled, overwriting it... often appears. Could the downlink be being scheduled rather than transmitted?

I tried adding "logger": true to the .conf file but it didn’t seem to make any difference.

If we can crack this it seems we’d be on the verge of cracking many more mysteries, like:

https://www.thethingsnetwork.org/forum/t/most-otaa-join-request-fail-to-downlink-confirmation-to-node/6183