Join-request to cluster-local Join Server failed DevNonce has already been used in registered node

Good morning, I have installed and correctly registered some water meters for several months (some in a building and others in a distant one, with a different gateway and connection), well, last October 10 I stopped receiving their readings, the gateways were working (some sensors are installed and their data was received that day without problem). In the following days, without doing anything, the readings have been received again but the emission counter has been reset, as if they had been self-registered again, except for two, which appear as registered (it seems like a bad or not active session), but when they emit it appears in TTN console "Join-request to cluster-local Join Server failed
DevNonce has already been used ". Any idea what might have happened? Any solution?

Maybe your water meters have a firmware that performs a join on every data upload (that is bad behaviour). On every join a “nonce” is used, basically a unique number. Your device cannot use the same nonce twice, and at some point you run out of unique numbers (it’s a 16-bit counter I think).

With newer versions of the LoRaWAN specification, the nonce should be incrementing (not just some random unique number). See also: