A
Okay seems like i fixed it or rather mitigated it.
After a lot of testing out and recording my results i noticed that the EAGAIN Error did happen only about 30% of the time and that i just gut unlucky in assuming that it happened particularly more often using a different power source.
Secondly, the EAGAIN Error only happened when the joining process took a long time. I assume that this does something that causes a faulty state in the Lopy4.
This 30% Error chance could be reduced to about 10% by setting the Datarate to 5 during sending.
s.setsockopt(socket.SOL_LORA, socket.SO_DR, 5)
After this i also increased the Datarate during joining by using
lora.join(activation=LoRa.OTAA, auth=(app_eui, app_key), timeout=0, dr=5)
I havent done extensive testing after this, but the error has not occured ever since.
Thank you again @jcaron!