TTN v3 risk of abandoned devices?

Hi folks,
Im playing around with the new public v3 cluster as a part of my Bachelor thesis. The documentation says I have to create the device in 4 services via REST or gRPC.

What happens if something went wrong in the deletion process of the device (because of a bad client implementation)? Lets say my application failed to delete the device only on the application or join server? Is there a periodic clean up job which deletes all devices which are not present on the identity server or something like this?

Jolly good question although I think it’s only three - join, network & application servers. Oh, no wait, it includes the identity server. Depends on which bit of the docs you read!

I think by implication it’s current expected that we do the clean up - as there is a suggested creation and then deletion order.

As the SDK is barely started, it may be a while before we can script it all.